Re: Eagle Microwave Antenna Arrays -- mechanical concepts
Tom,
Have we given up on the idea of using DSP techniques to do the phasing? If each antenna/amplifier had its own RF generator controlled by separate (I & Q) DACs, then it would be easy to control the phase of each element precisely with "infinite" interpolation between steps.
The same goes for the amplitude. So you could, for example, taper the amplitudes of the elements near the edges to reduce sidelobes. It should be possible to get a fully-symmetrical beam pattern to eliminate spin modulation.
This technique would allow full flexibility in antenna placement. The optimum phasings and amplitudes could be calculated before launch for all beam angles (every few degrees) and stored in a table. Software on the satellite would then interpolate between the table values.
Alan Bloom
On Fri, 2007-03-23 at 07:52, Tom Clark, K3IO wrote:
Grant Hodgson wrote:
Tom
Don't forget to claim back the expenses that you've incurred for these models...
More seriously - is the intention to have a separate phase shifter for each element?
Grant -- there are several basic ideas for doing the phasing: 1. A scheme which has been used in the past on electrically despun arrays is to have a discrete beam former with N beams and then discretely switch to the best of the beams as the s/c rotates. IMHO, this is a REALLY BAD :-P idea because there will be abrupt phase and amplitude discontinuities when switching from one beam to the next as the s/c spins. 2. A neat "zero click" adaptation of #1 can be done with a linear or square array. For this geometry, the "optimum" combiner is the Butler matrix which is the electrical realization of the Cooley-Tukey FFT. Assume that tap X is the beam now, and that Y is best for the next rotation step. If we use an in-phase variable power splitter that can linearly interpolate between the X & Y position, we can smoothly move the beam with no discontinuities. The interpolation is done in POWER with fractions [P] and [1-P] split between the X & Y taps. To build this for an NxN array, we build the combiner that makes NxN beams (of which [N-1]x[N-1] will be used -- we have no need to make use of the beam on the array's "horizon"). I've tested (in MATLAB) this idea for an 8x1 and 8x8 array. I haven't had a magic idea on a Butler-like matrix for hexagonal geometry. 3. We could devise some continuous phase shifter to be applied to each element. The required phase shift for any given pointing direction is a linear phase gradient across the aperture -- i.e. when viewed from the target (earth), we need to compensate for the geometrical phase offsets due to the plane of the array. [ Note: If we can generate the phase gradient easily, then we can free ourselves from any geometric constraints -- the elements an be located anywhere on the spacecraft.] Ideas are solicited!
73, Tom
Via the Eagle mailing list courtesy of AMSAT-NA Eagle@amsat.org http://amsat.org/mailman/listinfo/eagle
At 12:42 PM 3/23/2007, Alan Bloom wrote:
The same goes for the amplitude. So you could, for example, taper the amplitudes of the elements near the edges to reduce sidelobes. It
Side lobes of the spacecraft antenna do no harm, so there is no value in adding complexity to reduce them.
Good point. However I and Q signals may not be needed for each antenna. The transmitter is BPSK so there can be one signal source at the final RF output frequency. Analog phase shifter ICs (see attachments) can be used to generate signals for individual power amplifiers at each antenna. The transmitter power amplifiers should be operated at saturation for best efficiency so the small amplitude variations in the phase shifters don't matter.
For the receiver, an LNA, mixer and LO phase shifter could be placed at each antenna. Amplitude variations in the LO have little effect as the LO input to the mixer is limiter. The mixers could be quadrature mixer ICs (such as an HMC597) if a low-IF receiver is used or single Gilbert-cell mixers (such as an LT5560) if the a high IF is to be used. A high IF probably uses less power and the conversion to I and Q baseband signals could occur after the outputs of all the mixers are summed. It might even be possible to sum the RF directly and use only an LNA, phase shifter and variable attenuator for each antenna element. The attenuators would compensate for phase shifter gain variation.
73,
John KD6OZH
----- Original Message ----- From: "Alan Bloom" n1al@cds1.net To: K3IO@verizon.net Cc: "AMSAT Eagle" Eagle@amsat.org Sent: Friday, March 23, 2007 19:42 UTC Subject: [eagle] Re: Eagle Microwave Antenna Arrays -- mechanical concepts
Tom,
Have we given up on the idea of using DSP techniques to do the phasing? If each antenna/amplifier had its own RF generator controlled by separate (I & Q) DACs, then it would be easy to control the phase of each element precisely with "infinite" interpolation between steps.
The same goes for the amplitude. So you could, for example, taper the amplitudes of the elements near the edges to reduce sidelobes. It should be possible to get a fully-symmetrical beam pattern to eliminate spin modulation.
This technique would allow full flexibility in antenna placement. The optimum phasings and amplitudes could be calculated before launch for all beam angles (every few degrees) and stored in a table. Software on the satellite would then interpolate between the table values.
Alan Bloom
On Fri, 2007-03-23 at 07:52, Tom Clark, K3IO wrote:
Grant Hodgson wrote:
Tom
Don't forget to claim back the expenses that you've incurred for these models...
More seriously - is the intention to have a separate phase shifter for each element?
Grant -- there are several basic ideas for doing the phasing: 1. A scheme which has been used in the past on electrically despun arrays is to have a discrete beam former with N beams and then discretely switch to the best of the beams as the s/c rotates. IMHO, this is a REALLY BAD :-P idea because there will be abrupt phase and amplitude discontinuities when switching from one beam to the next as the s/c spins. 2. A neat "zero click" adaptation of #1 can be done with a linear or square array. For this geometry, the "optimum" combiner is the Butler matrix which is the electrical realization of the Cooley-Tukey FFT. Assume that tap X is the beam now, and that Y is best for the next rotation step. If we use an in-phase variable power splitter that can linearly interpolate between the X & Y position, we can smoothly move the beam with no discontinuities. The interpolation is done in POWER with fractions [P] and [1-P] split between the X & Y taps. To build this for an NxN array, we build the combiner that makes NxN beams (of which [N-1]x[N-1] will be used -- we have no need to make use of the beam on the array's "horizon"). I've tested (in MATLAB) this idea for an 8x1 and 8x8 array. I haven't had a magic idea on a Butler-like matrix for hexagonal geometry. 3. We could devise some continuous phase shifter to be applied to each element. The required phase shift for any given pointing direction is a linear phase gradient across the aperture -- i.e. when viewed from the target (earth), we need to compensate for the geometrical phase offsets due to the plane of the array. [ Note: If we can generate the phase gradient easily, then we can free ourselves from any geometric constraints -- the elements an be located anywhere on the spacecraft.] Ideas are solicited!
73, Tom
Via the Eagle mailing list courtesy of AMSAT-NA Eagle@amsat.org http://amsat.org/mailman/listinfo/eagle
Via the Eagle mailing list courtesy of AMSAT-NA Eagle@amsat.org http://amsat.org/mailman/listinfo/eagle
At 02:05 PM 3/23/2007, John B. Stephensen wrote:
It might even be possible to sum the RF directly and use only an LNA, phase shifter and variable attenuator for each antenna element. The attenuators would compensate for phase shifter gain variation.
I note that the spec for gain variation vs angle choice is 3 dB on these things! Yipes. Are there better ones available that don't have this flaw? We could certainly calibrate it out as you suggest, but maybe we could buy more expensive shifters that have it already calibrated out?
There are probably better phase shifters available. I just happened to see these while looking at an MA/COM catalog. LNA MMICs have a gain variation of about 3 dB and at least 1 dB of the variation is probably temperature related, so the phase shifter loss doesn't need to be controlled too precisely.
73,
John KD6OZH
----- Original Message ----- From: "Franklin Antonio" antonio@qualcomm.com To: "John B. Stephensen" kd6ozh@comcast.net Cc: n1al@cds1.net; K3IO@verizon.net; "AMSAT Eagle" Eagle@amsat.org Sent: Friday, March 23, 2007 21:37 UTC Subject: Re: [eagle] Re: Eagle Microwave Antenna Arrays -- mechanical concepts
At 02:05 PM 3/23/2007, John B. Stephensen wrote:
It might even be possible to sum the RF directly and use only an LNA, phase shifter and variable attenuator for each antenna element. The attenuators would compensate for phase shifter gain variation.
I note that the spec for gain variation vs angle choice is 3 dB on these things! Yipes. Are there better ones available that don't have this flaw? We could certainly calibrate it out as you suggest, but maybe we could buy more expensive shifters that have it already calibrated out?
At 03:24 PM 3/23/2007, John B. Stephensen wrote:
There are probably better phase shifters available. I just happened to see these while looking at an MA/COM catalog. LNA MMICs have a gain variation of about 3 dB and at least 1 dB of the variation is probably temperature related, so the phase shifter loss doesn't need to be controlled too precisely.
A related thought... My first thought was that the programmable attenuator was an absurd complexity. I was reacting to the thought that it was just there to fix this stupid feature of the phase shifter. However, there will be other things that make the levels slightly different at each antenna. Losses, splitter imbalances, god knows what. A programmable attenuator per element could shim all these things. Once per cycle, the software would just write a few bits to each antenna element, some of the bits going to the digital phaseshift chip, and some going to the digital attenuator chip. Maybe the attenuator is a good thing.
Still... 3dB !!! Such a big phase shifter loss VARIATION with phase is something we'd want to compensate for no matter what the LNA gain variation with temperature or whatever was.
At 12:42 PM 3/23/2007, Alan Bloom wrote:
If each antenna/amplifier had its own RF generator controlled by separate (I & Q) DACs, then it would be easy to control the phase of each element precisely with "infinite" interpolation between steps.
That's a lot of DACs. 2 per antenna element. High speed DACs consume a lot of power.
A much better approach would be to use a digitally controlled phase shifter at the input of each PA. I believe that 3-bit phase shifters will prove adequate, but hey, you could use phase shifters with more bits if you want. Remember, all of these phase shifters don't jump at once as the spacecraft rotates. Therefore the phase discontinuity in the resulting composite signal is much smaller than the phase discontinuity at one element.
I don't think its necessary to do I & Q. Why not just put the phase shifter in the RF path right before the PA? That avoids a zillion mixers.
There are many possible solutions. The trick is to choose one that has a low power consumption and weight and complexity (hence high reliability).
And the same scheme would work with the receiver with the addition of an attenuator and switch. The LNA feeds the phase shifter and attenuator. The attenuator compensates for LNA gain variation and phase shifter loss variation. The switch would be used to remove antenna elements from the receive path so each element could be calibrated individually. All outputs would feed a passive combiner.
73,
John KD6OZH
----- Original Message ----- From: "Franklin Antonio" antonio@qualcomm.com To: n1al@cds1.net Cc: "AMSAT Eagle" Eagle@amsat.org; K3IO@verizon.net Sent: Friday, March 23, 2007 21:32 UTC Subject: [eagle] Re: Eagle Microwave Antenna Arrays -- mechanical concepts
At 12:42 PM 3/23/2007, Alan Bloom wrote:
If each antenna/amplifier had its own RF generator controlled by separate (I & Q) DACs, then it would be easy to control the phase of each element precisely with "infinite" interpolation between steps.
That's a lot of DACs. 2 per antenna element. High speed DACs consume a lot of power.
A much better approach would be to use a digitally controlled phase shifter at the input of each PA. I believe that 3-bit phase shifters will prove adequate, but hey, you could use phase shifters with more bits if you want. Remember, all of these phase shifters don't jump at once as the spacecraft rotates. Therefore the phase discontinuity in the resulting composite signal is much smaller than the phase discontinuity at one element.
I don't think its necessary to do I & Q. Why not just put the phase shifter in the RF path right before the PA? That avoids a zillion mixers.
There are many possible solutions. The trick is to choose one that has a low power consumption and weight and complexity (hence high reliability).
Via the Eagle mailing list courtesy of AMSAT-NA Eagle@amsat.org http://amsat.org/mailman/listinfo/eagle
On Fri, 2007-03-23 at 13:05, John B. Stephensen wrote:
Good point. However I and Q signals may not be needed for each antenna. The transmitter is BPSK so there can be one signal source at the final RF output frequency.
...
On Fri, 2007-03-23 at 18:39, Tom Clark, K3IO wrote: ...
Franklin is right in noting that the element amplitude does not need to be tuned. We will always be in the mode of desiring maximum gain, and hence all individual PAs should be running at maximum efficiency.
...
I've been assuming that the BPSK modulation must be filtered to reduce the bandwidth. That means that the power amplifiers can't be run in class C since that would clip the rise/fall waveform of the modulation (unless we did some fancy pre-compensation).
But maybe unfiltered is OK. The satellite signal will be quite weak by the time it reaches earth, so maybe the unfiltered sin(x)/x sidebands would be acceptable.
If so, that really simplifies things. The modulator is simply a 180-degree phase shifter. Something like a double-balanced mixer fed with a bipolar digital signal (on the DC-coupled "IF" port). The PAs running in class C would have high efficiency and nice stable output power levels even if the input drive level is not very constant.
Then the only problem left is to be able to vary the RF phase of each element. Does anyone know a clever technique to control the phase of a microwave oscillator directly? It's easy to do over a 90-degree or so range using a PLL, but we need to do it over several cycles of phase.
Another method:
... It wouldn't even be necessary to do the phasing in DSP -- programmable NCOs could be used. ...
The NCOs would have to run at a lower RF frequency and then be heterodyned up to C-band. Still it wouldn't be too horrible a block diagram. Let's see if I can draw it in ASCII:
___________________ | C-band oscillator | |___________________| | _________V_________ | LO | | Balanced mixer IF|<--BPSK modulation signal |_________RF________| | | ________ | Phase 1-->| RF NCO | | |________| | | | ________V_________ |----->| Mixer upconverter |---> Element 1 amplifier | |___________________| | | ________ | Phase 2-->| RF NCO | | |________| | | | _________V_________ |----->| Mixer upconverter |---> Element 2 amplifier | |___________________| | | ... etc.
The phases would be generated in real time by the DSP and downloaded to the NCOs. There would have to be at least a little filtering at the upconverter outputs to reduce the image. Also, the 36-way splitter at the C-band oscillator output might get kind of interesting...
By the way, the same block diagram works even with filtered modulation as long as all stages are linear. Amplitude matching of the elements would have to be dealt with somehow.
Also, the same block diagram works in reverse for the S-band receiver if "BPSK modulation signal" is replaced with the proper LO to heterodyne down to the final IF.
Alan
======================================================================
On Fri, 2007-03-23 at 18:39, Tom Clark, K3IO wrote:
Let me try to answer a number of points. First, Alan wrote:
Tom,
Have we given up on the idea of using DSP techniques to do the phasing? If each antenna/amplifier had its own RF generator controlled by separate (I & Q) DACs, then it would be easy to control the phase of each element precisely with "infinite" interpolation between steps.
The same goes for the amplitude. So you could, for example, taper the amplitudes of the elements near the edges to reduce sidelobes. It should be possible to get a fully-symmetrical beam pattern to eliminate spin modulation.
The full DSP approach has not been discounted, but it comes with a lot of pain. First of all, it would make sense to do the phasing at IF and then up-convert to microwave with a single, coherent LO. The problem is that we would need 40-ish D/A converters, and this would cost a lot of power. It wouldn't even be necessary to do the phasing in DSP -- programmable NCOs could be used. If you think about it, the phase modulation pattern needed for all these LOs is a sinusoid, with the phase (with respect to the spin rate) and phase amplitude being a function of that particular patch's geometry. And yes, even if an array is off the spin-axis, the spin modulation can be removed too.
Franklin is right in noting that the element amplitude does not need to be tuned. We will always be in the mode of desiring maximum gain, and hence all individual PAs should be running at maximum efficiency. If we have too much TX power, we simply turn the unnecessary elements power off, saving valuable watts and making less heat. The presence of sidelobes is of no concern. We want sidelobe levels to be controlled, but that's just to prevent wasting power beaming it at the Venusians.
The problem with the phase shifters Jon mentioned is that their spec shows a nominal=4 dB, max=10 dB insertion loss. I can't find info in the spec sheet about power consumption. Again, if discrete phase shifters are to be used, doesn't it make sense to apply them at IF where power is of less concern?
Regardless (just speaking of the TX side), for N (a number like 40 ±4) patches in an array, we need N ~1 watt high efficiency amplifiers which will have at most 15 dB of gain. Somehow, we need to generate N drive signals thru some sort of power splitter. If we split at microwave, this will introduce at least 16 dB of loss, even if the splitter is perfect. Hence we will need a few watts of RF just to get enough power to the PAs at the array elements. And if we use John's lossy phase shifters, we need even more watts in the driver.
Back on Feb 13, I circulated an idea for doing the phasing in something like a Butler Matrix (if you don't have that message, I can resend it -- the subject was "Eagle Microwave Antenna Concepts"). It suggested a way to do the power splitter and array phasing using a Butler matrix (or something like that). That concept included the concept that, if the Butler array had ports for all possible beams, then beam steering could be done with something like a crossbar switch. I ended that message with a set of 7 questions I am still pondering:
So here are some challenges to make something like this work: 1. Is there some better way to do the phasing that using Butler matrices? 2. The antenna array phase needed to point at a given location is a simple linear phase gradient. If we don't use a Butler combiner, is there a simple way to invoke a linear phase "tilt" on a bunch of elements. 3. How complicated is it to fabricate the 2N Butler matrices that we will need in a microstrip structure? 4. The current baseline calls for 36 elements, arranged as either a 6x6 square or as a 37 element "bee hive" hexagon. What do Butler-like matrices look like for the cases where the array does not have 2^N elements? 5. Can we invent the microwave widget that allows us to linearly interpolate power between 2 (or preferably 4) of the (N-1)² beams? 6. Remember that we need a similar structure for the S2 receiver, with LNA's instead of PA's at each antenna. 7. Have I gone off the deep end with these ideas?
73, Tom
Via the Eagle mailing list courtesy of AMSAT-NA Eagle@amsat.org http://amsat.org/mailman/listinfo/eagle
The only disadvantage of the NCOs is the power required. Analog Devices makes a quad 500 Msps DDS but it consumes up to 80 mW per channel and the upconversion mixers will consume additional power. The MA/COM phase shifters consume less than 50 mW per channel.
However, NCOs aren't needed for the transmitter. The IF inputs could be fixed-frequency square waves with adjustable time delays. Harmonics could be cleaned up by low-pass filters preceeding the upconversion mixers. Spartan-3 FPGAs have up to 8 digital clock managers with 256-step time delays in 15-60 ps increments. A 15 ns maximum delay would allow a 360-degree phase shift at frequencies as low as 67 MHz and the FPGA can be clocked at 200 MHz. Virtex FPGAs run at 500 MHz and have more DCMs.
73,
John KD6OZH
----- Original Message ----- From: "Alan Bloom" n1al@cds1.net To: K3IO@verizon.net Cc: "AMSAT Eagle" Eagle@amsat.org Sent: Saturday, March 24, 2007 06:26 UTC Subject: [eagle] Re: Eagle Microwave Antenna Arrays -- mechanical concepts
On Fri, 2007-03-23 at 13:05, John B. Stephensen wrote:
Good point. However I and Q signals may not be needed for each antenna. The transmitter is BPSK so there can be one signal source at the final RF output frequency.
...
On Fri, 2007-03-23 at 18:39, Tom Clark, K3IO wrote: ...
Franklin is right in noting that the element amplitude does not need to be tuned. We will always be in the mode of desiring maximum gain, and hence all individual PAs should be running at maximum efficiency.
...
I've been assuming that the BPSK modulation must be filtered to reduce the bandwidth. That means that the power amplifiers can't be run in class C since that would clip the rise/fall waveform of the modulation (unless we did some fancy pre-compensation).
But maybe unfiltered is OK. The satellite signal will be quite weak by the time it reaches earth, so maybe the unfiltered sin(x)/x sidebands would be acceptable.
If so, that really simplifies things. The modulator is simply a 180-degree phase shifter. Something like a double-balanced mixer fed with a bipolar digital signal (on the DC-coupled "IF" port). The PAs running in class C would have high efficiency and nice stable output power levels even if the input drive level is not very constant.
Then the only problem left is to be able to vary the RF phase of each element. Does anyone know a clever technique to control the phase of a microwave oscillator directly? It's easy to do over a 90-degree or so range using a PLL, but we need to do it over several cycles of phase.
Another method:
... It wouldn't even be necessary to do the phasing in DSP -- programmable NCOs could be used. ...
The NCOs would have to run at a lower RF frequency and then be heterodyned up to C-band. Still it wouldn't be too horrible a block diagram. Let's see if I can draw it in ASCII:
| C-band oscillator | |___________________| | _________V_________ | LO | | Balanced mixer IF|<--BPSK modulation signal |_________RF________| | | ________ | Phase 1-->| RF NCO | | |________| | | | ________V_________ |----->| Mixer upconverter |---> Element 1 amplifier | |___________________| | | ________ | Phase 2-->| RF NCO | | |________| | | | _________V_________ |----->| Mixer upconverter |---> Element 2 amplifier | |___________________| | | ... etc.
The phases would be generated in real time by the DSP and downloaded to the NCOs. There would have to be at least a little filtering at the upconverter outputs to reduce the image. Also, the 36-way splitter at the C-band oscillator output might get kind of interesting...
By the way, the same block diagram works even with filtered modulation as long as all stages are linear. Amplitude matching of the elements would have to be dealt with somehow.
Also, the same block diagram works in reverse for the S-band receiver if "BPSK modulation signal" is replaced with the proper LO to heterodyne down to the final IF.
Alan
======================================================================
On Fri, 2007-03-23 at 18:39, Tom Clark, K3IO wrote:
Let me try to answer a number of points. First, Alan wrote:
Tom,
Have we given up on the idea of using DSP techniques to do the phasing? If each antenna/amplifier had its own RF generator controlled by separate (I & Q) DACs, then it would be easy to control the phase of each element precisely with "infinite" interpolation between steps.
The same goes for the amplitude. So you could, for example, taper the amplitudes of the elements near the edges to reduce sidelobes. It should be possible to get a fully-symmetrical beam pattern to eliminate spin modulation.
The full DSP approach has not been discounted, but it comes with a lot of pain. First of all, it would make sense to do the phasing at IF and then up-convert to microwave with a single, coherent LO. The problem is that we would need 40-ish D/A converters, and this would cost a lot of power. It wouldn't even be necessary to do the phasing in DSP -- programmable NCOs could be used. If you think about it, the phase modulation pattern needed for all these LOs is a sinusoid, with the phase (with respect to the spin rate) and phase amplitude being a function of that particular patch's geometry. And yes, even if an array is off the spin-axis, the spin modulation can be removed too.
Franklin is right in noting that the element amplitude does not need to be tuned. We will always be in the mode of desiring maximum gain, and hence all individual PAs should be running at maximum efficiency. If we have too much TX power, we simply turn the unnecessary elements power off, saving valuable watts and making less heat. The presence of sidelobes is of no concern. We want sidelobe levels to be controlled, but that's just to prevent wasting power beaming it at the Venusians.
The problem with the phase shifters Jon mentioned is that their spec shows a nominal=4 dB, max=10 dB insertion loss. I can't find info in the spec sheet about power consumption. Again, if discrete phase shifters are to be used, doesn't it make sense to apply them at IF where power is of less concern?
Regardless (just speaking of the TX side), for N (a number like 40 ±4) patches in an array, we need N ~1 watt high efficiency amplifiers which will have at most 15 dB of gain. Somehow, we need to generate N drive signals thru some sort of power splitter. If we split at microwave, this will introduce at least 16 dB of loss, even if the splitter is perfect. Hence we will need a few watts of RF just to get enough power to the PAs at the array elements. And if we use John's lossy phase shifters, we need even more watts in the driver.
Back on Feb 13, I circulated an idea for doing the phasing in something like a Butler Matrix (if you don't have that message, I can resend it -- the subject was "Eagle Microwave Antenna Concepts"). It suggested a way to do the power splitter and array phasing using a Butler matrix (or something like that). That concept included the concept that, if the Butler array had ports for all possible beams, then beam steering could be done with something like a crossbar switch. I ended that message with a set of 7 questions I am still pondering:
So here are some challenges to make something like this work: 1. Is there some better way to do the phasing that using Butler matrices? 2. The antenna array phase needed to point at a given location is a simple linear phase gradient. If we don't use a Butler combiner, is there a simple way to invoke a linear phase "tilt" on a bunch of elements. 3. How complicated is it to fabricate the 2N Butler matrices that we will need in a microstrip structure? 4. The current baseline calls for 36 elements, arranged as either a 6x6 square or as a 37 element "bee hive" hexagon. What do Butler-like matrices look like for the cases where the array does not have 2^N elements? 5. Can we invent the microwave widget that allows us to linearly interpolate power between 2 (or preferably 4) of the (N-1)² beams? 6. Remember that we need a similar structure for the S2 receiver, with LNA's instead of PA's at each antenna. 7. Have I gone off the deep end with these ideas?
73, Tom
Via the Eagle mailing list courtesy of AMSAT-NA Eagle@amsat.org http://amsat.org/mailman/listinfo/eagle
--------------------------------------------------------------------------------
Via the Eagle mailing list courtesy of AMSAT-NA Eagle@amsat.org http://amsat.org/mailman/listinfo/eagle
On Sat, 2007-03-24 at 00:28, John B. Stephensen wrote:
The only disadvantage of the NCOs is the power required. Analog Devices makes a quad 500 Msps DDS but it consumes up to 80 mW per channel and the upconversion mixers will consume additional power. The MA/COM phase shifters consume less than 50 mW per channel.
80 mW doesn't sound too bad for a signal chain with a 1W power amplifier.
However, NCOs aren't needed for the transmitter. The IF inputs could be fixed-frequency square waves with adjustable time delays. Harmonics could be cleaned up by low-pass filters preceeding the upconversion mixers. Spartan-3 FPGAs have up to 8 digital clock managers with 256-step time delays in 15-60 ps increments. A 15 ns maximum delay would allow a 360-degree phase shift at frequencies as low as 67 MHz and the FPGA can be clocked at 200 MHz. Virtex FPGAs run at 500 MHz and have more DCMs.
But how much power do the FPGAs require? Of course, it depends on what is programmed into them and the clock rate. But I know that in some applications they require a heat sink, so it can be quite a lot (several watts) of power.
Alan
The subject of the thermal design of the active antenna elements that are being planned will require some quite serious work to solve the issues being presented.
Dick Jansson, KD1K kd1k@amsat.org ---------------------------
-----Original Message----- From: eagle-bounces@amsat.org [ mailto:eagle-bounces@amsat.org mailto:eagle-bounces@amsat.org] On Behalf Of Alan Bloom Sent: Sunday, 25 March, 2007 0540 To: John B. Stephensen Cc: AMSAT Eagle; K3IO@verizon.net Subject: [eagle] Re: Eagle Microwave Antenna Arrays -- mechanical concepts
But how much power do the FPGAs require? Of course, it depends on what is programmed into them and the clock rate. But I know that in some applications they require a heat sink, so it can be quite a lot (several watts) of power.
Alan
The amount of power required by the FPGA depends on the number of logic elements used and the speed of operation. Anything that isn't clocked in the FPGA consumes almost no power so the thermal dissipation is controlled by the logic designer. DCMs require a lot less power than an external DDS and an FPGA would be a good place to put other logic. A $15 FPGA has over a million equivalent gates and the DCMs are 1 or 2 percent of that. Essentially, this eliminates the power consumed by the ROMs and DACs in the DDS chips as there is no need to synthesize sine waves or to vary the frequency. Discrete clock distribution chips also perform the same function but a lot of power and PCB area would be cnsumed to drive them.
73,
John KD6OZH
----- Original Message ----- From: "Alan Bloom" n1al@cds1.net To: "John B. Stephensen" kd6ozh@comcast.net Cc: K3IO@verizon.net; "AMSAT Eagle" Eagle@amsat.org Sent: Sunday, March 25, 2007 05:40 UTC Subject: Re: [eagle] Re: Eagle Microwave Antenna Arrays-- mechanical concepts
On Sat, 2007-03-24 at 00:28, John B. Stephensen wrote:
The only disadvantage of the NCOs is the power required. Analog Devices makes a quad 500 Msps DDS but it consumes up to 80 mW per channel and the upconversion mixers will consume additional power. The MA/COM phase shifters consume less than 50 mW per channel.
80 mW doesn't sound too bad for a signal chain with a 1W power amplifier.
However, NCOs aren't needed for the transmitter. The IF inputs could be fixed-frequency square waves with adjustable time delays. Harmonics could be cleaned up by low-pass filters preceeding the upconversion mixers. Spartan-3 FPGAs have up to 8 digital clock managers with 256-step time delays in 15-60 ps increments. A 15 ns maximum delay would allow a 360-degree phase shift at frequencies as low as 67 MHz and the FPGA can be clocked at 200 MHz. Virtex FPGAs run at 500 MHz and have more DCMs.
But how much power do the FPGAs require? Of course, it depends on what is programmed into them and the clock rate. But I know that in some applications they require a heat sink, so it can be quite a lot (several watts) of power.
Alan
John B. Stephensen wrote:
The amount of power required by the FPGA depends on the number of logic elements used and the speed of operation. Anything that isn't clocked in the FPGA consumes almost no power so the thermal dissipation is controlled by the logic designer.
This is not true anymore. Small geometry devices (< 90nm) can consume a lot of static power.
DCMs require a lot less power than an external DDS and an FPGA would be a good place to put other logic.
DCMs have horrible phase noise. You wouldn't want to use them for generating clocks for ADCs, DACs, LOs, or PLLs.
Matt
The big current spike is during start up and the static current afterwards is low enough so that you can manage power dissiation. I wouldn't use DCMs for the reciver as it must deal with multiple narrow-bandwidth signals. The downlink will be about 2 Mbaud BPSK so phase noise isn't a big issue in generating low-frequency signals to be upconverted.
73
John KD6OZH
----- Original Message ----- From: "Matt Ettus" matt@ettus.com To: "AMSAT Eagle" Eagle@amsat.org Sent: Sunday, March 25, 2007 21:40 UTC Subject: [eagle] Re: Eagle Microwave Antenna Arrays-- mechanical concepts
John B. Stephensen wrote:
The amount of power required by the FPGA depends on the number of logic elements used and the speed of operation. Anything that isn't clocked in the FPGA consumes almost no power so the thermal dissipation is controlled by the logic designer.
This is not true anymore. Small geometry devices (< 90nm) can consume a lot of static power.
DCMs require a lot less power than an external DDS and an FPGA would be a good place to put other logic.
DCMs have horrible phase noise. You wouldn't want to use them for generating clocks for ADCs, DACs, LOs, or PLLs.
Matt _______________________________________________ Via the Eagle mailing list courtesy of AMSAT-NA Eagle@amsat.org http://amsat.org/mailman/listinfo/eagle
This is an elegant and simple solution. However I'm worried that the DCM phase may not be accurate and stable enough. From the Xilinx "Virtex-4 data sheet: DC and Switching Characteristics", page 37, tables 45 and 46:
Output clock synthesis period jitter: +/-100 ps Output clock phase offset between any DCM outputs: +/-140 ps
With a 400 MHz clock, +/-140 ps is +/-20 degrees of phase. It doesn't seem like a 40 degree variation would be good enough for accurate array element phasing. Also the +/-100 ps jitter might adversely degrade the quality of the transmitted signal.
However those are worst-case specifications. Perhaps phase innacuracy can be compensated by measuring the indivicual FPGAs we plan to use and including calibration tables in software. We would have to measure the parts over all environmental parameter ranges to be sure the calibrated values are stable and repeatable.
Another solution might be to pick a lower clock frequency so that the innacuracy is a smaller percentage of a clock cycle. The tradeoff is that the IF image would be harder to filter out.
This is a good enough idea that it is probably worth doing some investigation to see if it can be made to work. (Sorry, I'm not volunteering. :=)
Alan
On Sun, 2007-03-25 at 16:06, John B. Stephensen wrote:
The big current spike is during start up and the static current afterwards is low enough so that you can manage power dissiation. I wouldn't use DCMs for the reciver as it must deal with multiple narrow-bandwidth signals. The downlink will be about 2 Mbaud BPSK so phase noise isn't a big issue in generating low-frequency signals to be upconverted.
73
John KD6OZH
----- Original Message ----- From: "Matt Ettus" matt@ettus.com To: "AMSAT Eagle" Eagle@amsat.org Sent: Sunday, March 25, 2007 21:40 UTC Subject: [eagle] Re: Eagle Microwave Antenna Arrays-- mechanical concepts
John B. Stephensen wrote:
The amount of power required by the FPGA depends on the number of logic elements used and the speed of operation. Anything that isn't clocked in the FPGA consumes almost no power so the thermal dissipation is controlled by the logic designer.
This is not true anymore. Small geometry devices (< 90nm) can consume a lot of static power.
DCMs require a lot less power than an external DDS and an FPGA would be a good place to put other logic.
DCMs have horrible phase noise. You wouldn't want to use them for generating clocks for ADCs, DACs, LOs, or PLLs.
Matt _______________________________________________ Via the Eagle mailing list courtesy of AMSAT-NA Eagle@amsat.org http://amsat.org/mailman/listinfo/eagle
Via the Eagle mailing list courtesy of AMSAT-NA Eagle@amsat.org http://amsat.org/mailman/listinfo/eagle
I was thinking that the DCM output could be around 100 MHz to minimize jitter. It consumes less power than a bank of DDSs. However, I still like the analog phase shifters as that alternative may consume even less power.
73,
John KD6OZH
----- Original Message ----- From: "Alan Bloom" n1al@cds1.net To: "John B. Stephensen" kd6ozh@comcast.net Cc: "Matt Ettus" matt@ettus.com; "AMSAT Eagle" Eagle@amsat.org Sent: Monday, March 26, 2007 04:26 UTC Subject: Re: [eagle] Re: Eagle Microwave AntennaArrays-- mechanical concepts
This is an elegant and simple solution. However I'm worried that the DCM phase may not be accurate and stable enough. From the Xilinx "Virtex-4 data sheet: DC and Switching Characteristics", page 37, tables 45 and 46:
Output clock synthesis period jitter: +/-100 ps Output clock phase offset between any DCM outputs: +/-140 ps
With a 400 MHz clock, +/-140 ps is +/-20 degrees of phase. It doesn't seem like a 40 degree variation would be good enough for accurate array element phasing. Also the +/-100 ps jitter might adversely degrade the quality of the transmitted signal.
However those are worst-case specifications. Perhaps phase innacuracy can be compensated by measuring the indivicual FPGAs we plan to use and including calibration tables in software. We would have to measure the parts over all environmental parameter ranges to be sure the calibrated values are stable and repeatable.
Another solution might be to pick a lower clock frequency so that the innacuracy is a smaller percentage of a clock cycle. The tradeoff is that the IF image would be harder to filter out.
This is a good enough idea that it is probably worth doing some investigation to see if it can be made to work. (Sorry, I'm not volunteering. :=)
Alan
On Sun, 2007-03-25 at 16:06, John B. Stephensen wrote:
The big current spike is during start up and the static current afterwards is low enough so that you can manage power dissiation. I wouldn't use DCMs for the reciver as it must deal with multiple narrow-bandwidth signals. The downlink will be about 2 Mbaud BPSK so phase noise isn't a big issue in generating low-frequency signals to be upconverted.
73
John KD6OZH
----- Original Message ----- From: "Matt Ettus" matt@ettus.com To: "AMSAT Eagle" Eagle@amsat.org Sent: Sunday, March 25, 2007 21:40 UTC Subject: [eagle] Re: Eagle Microwave Antenna Arrays-- mechanical concepts
John B. Stephensen wrote:
The amount of power required by the FPGA depends on the number of logic elements used and the speed of operation. Anything that isn't clocked in the FPGA consumes almost no power so the thermal dissipation is controlled by the logic designer.
This is not true anymore. Small geometry devices (< 90nm) can consume a lot of static power.
DCMs require a lot less power than an external DDS and an FPGA would be a good place to put other logic.
DCMs have horrible phase noise. You wouldn't want to use them for generating clocks for ADCs, DACs, LOs, or PLLs.
Matt _______________________________________________ Via the Eagle mailing list courtesy of AMSAT-NA Eagle@amsat.org http://amsat.org/mailman/listinfo/eagle
Via the Eagle mailing list courtesy of AMSAT-NA Eagle@amsat.org http://amsat.org/mailman/listinfo/eagle
John B. Stephensen wrote:
The big current spike is during start up and the static current afterwards is low enough so that you can manage power dissiation.
Simply not true. Large 65nm FPGAs dissipate several watts of standby power.
I wouldn't use DCMs for the reciver as it must deal with multiple narrow-bandwidth signals. The downlink will be about 2 Mbaud BPSK so phase noise isn't a big issue in generating low-frequency signals to be upconverted.
I can speak from experience here. Clocks coming out of the FPGA are simply not clean enough to use for this purpose.
Matt
Alan Bloom wrote:
On Fri, 2007-03-23 at 13:05, John B. Stephensen wrote:
Good point. However I and Q signals may not be needed for each antenna. The transmitter is BPSK so there can be one signal source at the final RF output frequency.
...
On Fri, 2007-03-23 at 18:39, Tom Clark, K3IO wrote: ...
Franklin is right in noting that the element amplitude does not need to be tuned. We will always be in the mode of desiring maximum gain, and hence all individual PAs should be running at maximum efficiency.
...
I've been assuming that the BPSK modulation must be filtered to reduce the bandwidth. That means that the power amplifiers can't be run in class C since that would clip the rise/fall waveform of the modulation (unless we did some fancy pre-compensation).
But maybe unfiltered is OK. The satellite signal will be quite weak by the time it reaches earth, so maybe the unfiltered sin(x)/x sidebands would be acceptable.
It is okay and we are planning on hard limited amplifiers and BPSK.
If so, that really simplifies things. The modulator is simply a 180-degree phase shifter. Something like a double-balanced mixer fed with a bipolar digital signal (on the DC-coupled "IF" port). The PAs running in class C would have high efficiency and nice stable output power levels even if the input drive level is not very constant.
Then the only problem left is to be able to vary the RF phase of each element. Does anyone know a clever technique to control the phase of a microwave oscillator directly? It's easy to do over a 90-degree or so range using a PLL, but we need to do it over several cycles of phase.
Another method:
... It wouldn't even be necessary to do the phasing in DSP -- programmable NCOs could be used. ...
The NCOs would have to run at a lower RF frequency and then be heterodyned up to C-band. Still it wouldn't be too horrible a block diagram. Let's see if I can draw it in ASCII:
This was all discussed in our San Diego meeting of last summer. We traded off then and we are still trading off analog shifters versus tuning the phase of each element in the NCO's.
My plan is to have us start building actual components THIS SUMMER in a hard push to present project management with some real engineering trade offs so we can make hard decisions and build stuff.
NOW.
Alan
Bob
One solution for loss in a passive splitter is to have 30 dB of gain at each antenna element by using 2 MMICs. Another would be to build an active splitter using 2 layers of passive splitters and intervening amplifiers. The MA/COM phase shifter data sheet says only that they consume less than 10 mA at -5 V
Another way of doing the phase shift is a lumped-constant delay line with varactors. They draw less current, but the amplitude variation is similar.
I did find 2 papers on implementing microstrip Butler matrices. However, since they work by cascading a series of 180, 90, 45... degree phase shifts, I don't see a way of doing anything other than a 4 or 8-way power split. Like FFTs they seem to only work in integer powers of 2. A 100-point FFT is done by adding 28 zero-value samples and doing a 128-pont FFT. You still end up with 128 output bins. If the Butler matix is operated at low signal levels, the power from 2 of the output ports can be dumped into termination resistors.
73,
John KD6OZH ----- Original Message ----- From: Tom Clark, K3IO To: AMSAT Eagle Sent: Saturday, March 24, 2007 02:39 UTC Subject: [eagle] Re: Eagle Microwave Antenna Arrays -- mechanical concepts
Let me try to answer a number of points. First, Alan wrote: Tom,
Have we given up on the idea of using DSP techniques to do the phasing? If each antenna/amplifier had its own RF generator controlled by separate (I & Q) DACs, then it would be easy to control the phase of each element precisely with "infinite" interpolation between steps.
The same goes for the amplitude. So you could, for example, taper the amplitudes of the elements near the edges to reduce sidelobes. It should be possible to get a fully-symmetrical beam pattern to eliminate spin modulation. The full DSP approach has not been discounted, but it comes with a lot of pain. First of all, it would make sense to do the phasing at IF and then up-convert to microwave with a single, coherent LO. The problem is that we would need 40-ish D/A converters, and this would cost a lot of power. It wouldn't even be necessary to do the phasing in DSP -- programmable NCOs could be used. If you think about it, the phase modulation pattern needed for all these LOs is a sinusoid, with the phase (with respect to the spin rate) and phase amplitude being a function of that particular patch's geometry. And yes, even if an array is off the spin-axis, the spin modulation can be removed too.
Franklin is right in noting that the element amplitude does not need to be tuned. We will always be in the mode of desiring maximum gain, and hence all individual PAs should be running at maximum efficiency. If we have too much TX power, we simply turn the unnecessary elements power off, saving valuable watts and making less heat. The presence of sidelobes is of no concern. We want sidelobe levels to be controlled, but that's just to prevent wasting power beaming it at the Venusians.
The problem with the phase shifters Jon mentioned is that their spec shows a nominal=4 dB, max=10 dB insertion loss. I can't find info in the spec sheet about power consumption. Again, if discrete phase shifters are to be used, doesn't it make sense to apply them at IF where power is of less concern?
Regardless (just speaking of the TX side), for N (a number like 40 ±4) patches in an array, we need N ~1 watt high efficiency amplifiers which will have at most 15 dB of gain. Somehow, we need to generate N drive signals thru some sort of power splitter. If we split at microwave, this will introduce at least 16 dB of loss, even if the splitter is perfect. Hence we will need a few watts of RF just to get enough power to the PAs at the array elements. And if we use John's lossy phase shifters, we need even more watts in the driver.
Back on Feb 13, I circulated an idea for doing the phasing in something like a Butler Matrix (if you don't have that message, I can resend it -- the subject was "Eagle Microwave Antenna Concepts"). It suggested a way to do the power splitter and array phasing using a Butler matrix (or something like that). That concept included the concept that, if the Butler array had ports for all possible beams, then beam steering could be done with something like a crossbar switch. I ended that message with a set of 7 questions I am still pondering:
So here are some challenges to make something like this work:
1.. Is there some better way to do the phasing that using Butler matrices? 2.. The antenna array phase needed to point at a given location is a simple linear phase gradient. If we don't use a Butler combiner, is there a simple way to invoke a linear phase "tilt" on a bunch of elements.
3.. How complicated is it to fabricate the 2N Butler matrices that we will need in a microstrip structure? 4.. The current baseline calls for 36 elements, arranged as either a 6x6 square or as a 37 element "bee hive" hexagon. What do Butler-like matrices look like for the cases where the array does not have 2^N elements?
5.. Can we invent the microwave widget that allows us to linearly interpolate power between 2 (or preferably 4) of the (N-1)² beams? 6.. Remember that we need a similar structure for the S2 receiver, with LNA's instead of PA's at each antenna.
7.. Have I gone off the deep end with these ideas? 73, Tom
------------------------------------------------------------------------------
_______________________________________________ Via the Eagle mailing list courtesy of AMSAT-NA Eagle@amsat.org http://amsat.org/mailman/listinfo/eagle
At 08:52 AM 3/23/2007, Tom Clark, K3IO wrote:
IMHO, this is a REALLY BAD :-P idea because there will be abrupt phase and amplitude discontinuities when switching from one beam to the next as the s/c spins.
You should not fear small phase discontinuities from steps in the beam forming. These phase jumps will be very small, and will have only a tiny effect on coded BPSK modulation.
We do not need to do anything fancy to eliminate small phase steps that will have no impact on performance.
I agree. If you had 5-bits of phase control (11.25 degree steps) the noise would be 24 dB down.
73,
John KD6OZH
----- Original Message ----- From: "Franklin Antonio" antonio@qualcomm.com To: K3IO@verizon.net Cc: "AMSAT Eagle" Eagle@amsat.org Sent: Friday, March 23, 2007 19:53 UTC Subject: [eagle] Re: Eagle Microwave Antenna Arrays -- mechanical concepts
At 08:52 AM 3/23/2007, Tom Clark, K3IO wrote:
IMHO, this is a REALLY BAD :-P idea because there will be abrupt phase and amplitude discontinuities when switching from one beam to the next as the s/c spins.
You should not fear small phase discontinuities from steps in the beam forming. These phase jumps will be very small, and will have only a tiny effect on coded BPSK modulation.
We do not need to do anything fancy to eliminate small phase steps that will have no impact on performance.
Via the Eagle mailing list courtesy of AMSAT-NA Eagle@amsat.org http://amsat.org/mailman/listinfo/eagle
I think I've figured out a good scheme, which may be the right way to build the C-band antenna. It's a bit different than you guys have been thinking.
You've been thinking about how to accomplish the phasing of the RF signal, but ignoring the physical distribution of RF. Physical distribution of C-band RF signal to 35 or so elements is nontrivial. It could be done with splitters and cables and connectors, or printed splitters and microstrip lines, but with any of these schemes there will be amplitude mismatches and various phase shifts associated with just the distribution. If we distribute RF, we'll have to compensate for those amplitude and phase variations. Can be done of course.
How about this. Instead of distributing RF, we distribute a much lower reference frequency which is multiplied up by a PLL at each antenna element. A PLL is one chip these days, so it certainly doesn't cost much in dollars weight or power. We've already got a PA and phase control stuff at each element after all. So maybe we distribute 100 MHz or something easy. It is only gonna be used as a reference, so we don't need to match amplitudes. We design the reference distribution network for high isolation between the outputs, so the individual element circuits have least chance to couple. We don't need to match phases of the distributed reference because we're gonna adjust the phase of each element under software control anyway. (We'll build a calibration table which the software will use as a term it adds into the total phase shift it specifies for each antenna element.)
At each element we have a PLL to generate the RF frequency, followed by a digitally controlled analog phase shifter chip, and a digitally controlled attenuator chip (to adjust out variations in the phase shifter chip vs control input), and a balanced mixer to generate the 180 degree phase shift of BPSK. (If the phase shift chip has a 180 degree input, we can leave out the balanced mixer.) Next, of course, the signal goes to the local PA.
This entire circuit fits on a small board smaller than the antenna element (ie patch) itself, and bolts to the back of the antenna element.
This design is highly modular. The only things distributed are power, reference, data-to-be-modulated, and data to control the individual elements. The controls are on/off, phase, and gain.
It is also possible to move the digital phase shifter to BEFORE the PLL, at which point it may be possible to remove the digital attenuator entirely. You would no longer need it to compensate for the changing attenuation of the phase shifter vs control input, as long as the PLL could handle the range of possible amplitudes. (On the other hand you might still want the attenuator as an easy way to balance the gain and/or power output level of the PAs. With the digitally controlled phase shifter operating at a lower frequency it would likely be more accurate and have less amplitude variation anyway.
This scheme works no matter what the geometry of the array. Software just generates the phases, and sends 'em down some bus.
The thing I have described easily makes unfiltered BPSK or even QPSK. Somebody was talking about possible filtering. That's harder. I hope we don't need filtering. Distributed filtering is not an easy thing.
What do you think?
This sounds like a scheme we could live with. It is very much like what I thought we would need. I especially like the part about "This scheme works no matter what the geometry of the array. Software just generates the phases, and sends 'em down some bus". Being modular adds greatly to the overall reliability.
I must remind everybody that power is very critical and will ultimately determine the size and shape of the spacecraft. For the 170w average power predicted at the San Diego meeting, it will require a 250w solar power capability. This could be reduced by reducing the power during eclipse or by relaxing the requirement for operating at all possible sun angles.
Lou McFadin W5DID w5did@mac.com
On Apr 2, 2007, at 1:56 AM, Franklin Antonio wrote:
I think I've figured out a good scheme, which may be the right way to build the C-band antenna. It's a bit different than you guys have been thinking.
You've been thinking about how to accomplish the phasing of the RF signal, but ignoring the physical distribution of RF. Physical distribution of C-band RF signal to 35 or so elements is nontrivial. It could be done with splitters and cables and connectors, or printed splitters and microstrip lines, but with any of these schemes there will be amplitude mismatches and various phase shifts associated with just the distribution. If we distribute RF, we'll have to compensate for those amplitude and phase variations. Can be done of course.
How about this. Instead of distributing RF, we distribute a much lower reference frequency which is multiplied up by a PLL at each antenna element. A PLL is one chip these days, so it certainly doesn't cost much in dollars weight or power. We've already got a PA and phase control stuff at each element after all. So maybe we distribute 100 MHz or something easy. It is only gonna be used as a reference, so we don't need to match amplitudes. We design the reference distribution network for high isolation between the outputs, so the individual element circuits have least chance to couple. We don't need to match phases of the distributed reference because we're gonna adjust the phase of each element under software control anyway. (We'll build a calibration table which the software will use as a term it adds into the total phase shift it specifies for each antenna element.)
At each element we have a PLL to generate the RF frequency, followed by a digitally controlled analog phase shifter chip, and a digitally controlled attenuator chip (to adjust out variations in the phase shifter chip vs control input), and a balanced mixer to generate the 180 degree phase shift of BPSK. (If the phase shift chip has a 180 degree input, we can leave out the balanced mixer.) Next, of course, the signal goes to the local PA.
This entire circuit fits on a small board smaller than the antenna element (ie patch) itself, and bolts to the back of the antenna element.
This design is highly modular. The only things distributed are power, reference, data-to-be-modulated, and data to control the individual elements. The controls are on/off, phase, and gain.
It is also possible to move the digital phase shifter to BEFORE the PLL, at which point it may be possible to remove the digital attenuator entirely. You would no longer need it to compensate for the changing attenuation of the phase shifter vs control input, as long as the PLL could handle the range of possible amplitudes. (On the other hand you might still want the attenuator as an easy way to balance the gain and/or power output level of the PAs. With the digitally controlled phase shifter operating at a lower frequency it would likely be more accurate and have less amplitude variation anyway.
This scheme works no matter what the geometry of the array. Software just generates the phases, and sends 'em down some bus.
The thing I have described easily makes unfiltered BPSK or even QPSK. Somebody was talking about possible filtering. That's harder. I hope we don't need filtering. Distributed filtering is not an easy thing.
What do you think?
Via the Eagle mailing list courtesy of AMSAT-NA Eagle@amsat.org http://amsat.org/mailman/listinfo/eagle
The PLL would have to be really good to have sufficiently low phase noise. This will become the complex issue to be dealt with in this scheme. This sounds really good but some analysis of the expected noise floor will have to be done.
Bob
Louis McFadin wrote:
This sounds like a scheme we could live with. It is very much like what I thought we would need. I especially like the part about "This scheme works no matter what the geometry of the array. Software just generates the phases, and sends 'em down some bus". Being modular adds greatly to the overall reliability.
I must remind everybody that power is very critical and will ultimately determine the size and shape of the spacecraft. For the 170w average power predicted at the San Diego meeting, it will require a 250w solar power capability. This could be reduced by reducing the power during eclipse or by relaxing the requirement for operating at all possible sun angles.
Lou McFadin W5DID w5did@mac.com mailto:w5did@mac.com
On Apr 2, 2007, at 1:56 AM, Franklin Antonio wrote:
I think I've figured out a good scheme, which may be the right way to build the C-band antenna. It's a bit different than you guys have been thinking.
You've been thinking about how to accomplish the phasing of the RF signal, but ignoring the physical distribution of RF. Physical distribution of C-band RF signal to 35 or so elements is nontrivial. It could be done with splitters and cables and connectors, or printed splitters and microstrip lines, but with any of these schemes there will be amplitude mismatches and various phase shifts associated with just the distribution. If we distribute RF, we'll have to compensate for those amplitude and phase variations. Can be done of course.
How about this. Instead of distributing RF, we distribute a much lower reference frequency which is multiplied up by a PLL at each antenna element. A PLL is one chip these days, so it certainly doesn't cost much in dollars weight or power. We've already got a PA and phase control stuff at each element after all. So maybe we distribute 100 MHz or something easy. It is only gonna be used as a reference, so we don't need to match amplitudes. We design the reference distribution network for high isolation between the outputs, so the individual element circuits have least chance to couple. We don't need to match phases of the distributed reference because we're gonna adjust the phase of each element under software control anyway. (We'll build a calibration table which the software will use as a term it adds into the total phase shift it specifies for each antenna element.)
At each element we have a PLL to generate the RF frequency, followed by a digitally controlled analog phase shifter chip, and a digitally controlled attenuator chip (to adjust out variations in the phase shifter chip vs control input), and a balanced mixer to generate the 180 degree phase shift of BPSK. (If the phase shift chip has a 180 degree input, we can leave out the balanced mixer.) Next, of course, the signal goes to the local PA.
This entire circuit fits on a small board smaller than the antenna element (ie patch) itself, and bolts to the back of the antenna element.
This design is highly modular. The only things distributed are power, reference, data-to-be-modulated, and data to control the individual elements. The controls are on/off, phase, and gain.
It is also possible to move the digital phase shifter to BEFORE the PLL, at which point it may be possible to remove the digital attenuator entirely. You would no longer need it to compensate for the changing attenuation of the phase shifter vs control input, as long as the PLL could handle the range of possible amplitudes. (On the other hand you might still want the attenuator as an easy way to balance the gain and/or power output level of the PAs. With the digitally controlled phase shifter operating at a lower frequency it would likely be more accurate and have less amplitude variation anyway.
This scheme works no matter what the geometry of the array. Software just generates the phases, and sends 'em down some bus.
The thing I have described easily makes unfiltered BPSK or even QPSK. Somebody was talking about possible filtering. That's harder. I hope we don't need filtering. Distributed filtering is not an easy thing.
What do you think?
Via the Eagle mailing list courtesy of AMSAT-NA Eagle@amsat.org mailto:Eagle@amsat.org http://amsat.org/mailman/listinfo/eagle
Via the Eagle mailing list courtesy of AMSAT-NA Eagle@amsat.org http://amsat.org/mailman/listinfo/eagle
Good phase noise shouldn't be too hard. With a high reference frequency you can use a wide loop bandwidth and basically get the same phase noise as the reference +20log(Frf/Fref).
The other issue is that the phase shifter needs a full 360-degree range. It would be easy to do at the reference frequency with an NCO.
Alan
On Mon, 2007-04-02 at 09:26, Robert McGwier wrote:
The PLL would have to be really good to have sufficiently low phase noise. This will become the complex issue to be dealt with in this scheme. This sounds really good but some analysis of the expected noise floor will have to be done.
Bob
Louis McFadin wrote:
This sounds like a scheme we could live with. It is very much like what I thought we would need. I especially like the part about "This scheme works no matter what the geometry of the array. Software just generates the phases, and sends 'em down some bus". Being modular adds greatly to the overall reliability.
I must remind everybody that power is very critical and will ultimately determine the size and shape of the spacecraft. For the 170w average power predicted at the San Diego meeting, it will require a 250w solar power capability. This could be reduced by reducing the power during eclipse or by relaxing the requirement for operating at all possible sun angles.
Lou McFadin W5DID w5did@mac.com mailto:w5did@mac.com
On Apr 2, 2007, at 1:56 AM, Franklin Antonio wrote:
I think I've figured out a good scheme, which may be the right way to build the C-band antenna. It's a bit different than you guys have been thinking.
You've been thinking about how to accomplish the phasing of the RF signal, but ignoring the physical distribution of RF. Physical distribution of C-band RF signal to 35 or so elements is nontrivial. It could be done with splitters and cables and connectors, or printed splitters and microstrip lines, but with any of these schemes there will be amplitude mismatches and various phase shifts associated with just the distribution. If we distribute RF, we'll have to compensate for those amplitude and phase variations. Can be done of course.
How about this. Instead of distributing RF, we distribute a much lower reference frequency which is multiplied up by a PLL at each antenna element. A PLL is one chip these days, so it certainly doesn't cost much in dollars weight or power. We've already got a PA and phase control stuff at each element after all. So maybe we distribute 100 MHz or something easy. It is only gonna be used as a reference, so we don't need to match amplitudes. We design the reference distribution network for high isolation between the outputs, so the individual element circuits have least chance to couple. We don't need to match phases of the distributed reference because we're gonna adjust the phase of each element under software control anyway. (We'll build a calibration table which the software will use as a term it adds into the total phase shift it specifies for each antenna element.)
At each element we have a PLL to generate the RF frequency, followed by a digitally controlled analog phase shifter chip, and a digitally controlled attenuator chip (to adjust out variations in the phase shifter chip vs control input), and a balanced mixer to generate the 180 degree phase shift of BPSK. (If the phase shift chip has a 180 degree input, we can leave out the balanced mixer.) Next, of course, the signal goes to the local PA.
This entire circuit fits on a small board smaller than the antenna element (ie patch) itself, and bolts to the back of the antenna element.
This design is highly modular. The only things distributed are power, reference, data-to-be-modulated, and data to control the individual elements. The controls are on/off, phase, and gain.
It is also possible to move the digital phase shifter to BEFORE the PLL, at which point it may be possible to remove the digital attenuator entirely. You would no longer need it to compensate for the changing attenuation of the phase shifter vs control input, as long as the PLL could handle the range of possible amplitudes. (On the other hand you might still want the attenuator as an easy way to balance the gain and/or power output level of the PAs. With the digitally controlled phase shifter operating at a lower frequency it would likely be more accurate and have less amplitude variation anyway.
This scheme works no matter what the geometry of the array. Software just generates the phases, and sends 'em down some bus.
The thing I have described easily makes unfiltered BPSK or even QPSK. Somebody was talking about possible filtering. That's harder. I hope we don't need filtering. Distributed filtering is not an easy thing.
What do you think?
Via the Eagle mailing list courtesy of AMSAT-NA Eagle@amsat.org mailto:Eagle@amsat.org http://amsat.org/mailman/listinfo/eagle
Via the Eagle mailing list courtesy of AMSAT-NA Eagle@amsat.org http://amsat.org/mailman/listinfo/eagle
It shouldn't be hard at all for the tranmitter as the downlink will be operating at about 2 Mbaud. VCO MMICs may work. Extendin this scheme to the receiver would be harder but 1/2-inch square VCOs may be adequate.
73,
John KD6OZH
----- Original Message ----- From: "Alan Bloom" n1al@cds1.net To: "Robert McGwier" rwmcgwier@gmail.com Cc: "AMSAT Eagle" eagle@amsat.org; "Louis McFadin" w5did@amsat.org Sent: Monday, April 02, 2007 18:30 UTC Subject: [eagle] Re: Eagle Microwave Antenna Arrays -- RF concepts
Good phase noise shouldn't be too hard. With a high reference frequency you can use a wide loop bandwidth and basically get the same phase noise as the reference +20log(Frf/Fref).
The other issue is that the phase shifter needs a full 360-degree range. It would be easy to do at the reference frequency with an NCO.
Alan
On Mon, 2007-04-02 at 09:26, Robert McGwier wrote:
The PLL would have to be really good to have sufficiently low phase noise. This will become the complex issue to be dealt with in this scheme. This sounds really good but some analysis of the expected noise floor will have to be done.
Bob
Louis McFadin wrote:
This sounds like a scheme we could live with. It is very much like what I thought we would need. I especially like the part about "This scheme works no matter what the geometry of the array. Software just generates the phases, and sends 'em down some bus". Being modular adds greatly to the overall reliability.
I must remind everybody that power is very critical and will ultimately determine the size and shape of the spacecraft. For the 170w average power predicted at the San Diego meeting, it will require a 250w solar power capability. This could be reduced by reducing the power during eclipse or by relaxing the requirement for operating at all possible sun angles.
Lou McFadin W5DID w5did@mac.com mailto:w5did@mac.com
On Apr 2, 2007, at 1:56 AM, Franklin Antonio wrote:
I think I've figured out a good scheme, which may be the right way to build the C-band antenna. It's a bit different than you guys have been thinking.
You've been thinking about how to accomplish the phasing of the RF signal, but ignoring the physical distribution of RF. Physical distribution of C-band RF signal to 35 or so elements is nontrivial. It could be done with splitters and cables and connectors, or printed splitters and microstrip lines, but with any of these schemes there will be amplitude mismatches and various phase shifts associated with just the distribution. If we distribute RF, we'll have to compensate for those amplitude and phase variations. Can be done of course.
How about this. Instead of distributing RF, we distribute a much lower reference frequency which is multiplied up by a PLL at each antenna element. A PLL is one chip these days, so it certainly doesn't cost much in dollars weight or power. We've already got a PA and phase control stuff at each element after all. So maybe we distribute 100 MHz or something easy. It is only gonna be used as a reference, so we don't need to match amplitudes. We design the reference distribution network for high isolation between the outputs, so the individual element circuits have least chance to couple. We don't need to match phases of the distributed reference because we're gonna adjust the phase of each element under software control anyway. (We'll build a calibration table which the software will use as a term it adds into the total phase shift it specifies for each antenna element.)
At each element we have a PLL to generate the RF frequency, followed by a digitally controlled analog phase shifter chip, and a digitally controlled attenuator chip (to adjust out variations in the phase shifter chip vs control input), and a balanced mixer to generate the 180 degree phase shift of BPSK. (If the phase shift chip has a 180 degree input, we can leave out the balanced mixer.) Next, of course, the signal goes to the local PA.
This entire circuit fits on a small board smaller than the antenna element (ie patch) itself, and bolts to the back of the antenna element.
This design is highly modular. The only things distributed are power, reference, data-to-be-modulated, and data to control the individual elements. The controls are on/off, phase, and gain.
It is also possible to move the digital phase shifter to BEFORE the PLL, at which point it may be possible to remove the digital attenuator entirely. You would no longer need it to compensate for the changing attenuation of the phase shifter vs control input, as long as the PLL could handle the range of possible amplitudes. (On the other hand you might still want the attenuator as an easy way to balance the gain and/or power output level of the PAs. With the digitally controlled phase shifter operating at a lower frequency it would likely be more accurate and have less amplitude variation anyway.
This scheme works no matter what the geometry of the array. Software just generates the phases, and sends 'em down some bus.
The thing I have described easily makes unfiltered BPSK or even QPSK. Somebody was talking about possible filtering. That's harder. I hope we don't need filtering. Distributed filtering is not an easy thing.
What do you think?
Via the Eagle mailing list courtesy of AMSAT-NA Eagle@amsat.org mailto:Eagle@amsat.org http://amsat.org/mailman/listinfo/eagle
Via the Eagle mailing list courtesy of AMSAT-NA Eagle@amsat.org http://amsat.org/mailman/listinfo/eagle
Via the Eagle mailing list courtesy of AMSAT-NA Eagle@amsat.org http://amsat.org/mailman/listinfo/eagle
At 11:30 AM 4/2/2007, Alan Bloom wrote:
The other issue is that the phase shifter needs a full 360-degree range. It would be easy to do at the reference frequency with an NCO.
It is easy to get full 360 degree phase control. You just buy this chip that does it.
It is also possible to put the phase shifter before the PLL, in which case you do not need 360 degree phase control, but 360/n where n is the PLL multiplication factor.
Good phase noise shouldn't be too hard. With a high reference frequency you can use a wide loop bandwidth and basically get the same phase noise as the reference +20log(Frf/Fref).
Yep. There are a couple of issues here. I would do the data modulation after the PLL, although it can be done either way. If the data modulation is done before the PLL, you would want the PLL bandwidth to be many times wider than the modulation, because you want the PLL output phase to change in a small fraction of a modulation symbol time. (The numbers might work out ok.) If you put the modulation after the PLL then the modulation puts no constraint on the PLL loop bandwidth. Why would you care? Well because if we have a 1 Msymbol/sec modulation, and you want the phase to change in less than 1/10th of a modulation symbol, then you might choose a PLL bandwidth 20x the modulation rate, which would get you to 20MHz, but if the reference frequency of the PLL is only 100 MHz, and you want the bandwidth of the PLL to be a small fraction of the reference frequency (sampling issue) then you're already stuck with limited choices. Might work ok, but its not like there's a lot of room, but you'd need more careful analysis. I prefer an approach with a lot of room, and easy analysis, so I was preferring the modulation after the PLL approach.
As for phase noise, as long as the PLL loop bandwidth is wider than the modulation, then the phase noise THAT MATTERS (ie the phase noise within the modulation bandwidth) will be controlled by the reference, not the VCO in the PLL.
I forget what our modulation symbol rate was gonna be. Suppose it is 1 Msymbol/second. Then the loop bandwidth needs to be wider than about 1 MHz to ensure that the phase noise is dominated by the reference, and that seems trivially easy to accomplish. Hey, you can throw a factor of 2 or 3 in there for good measure. A 3 MHz loop bandwidth works out fine with a 100 MHz reference frequency, as that's 33:1.
So phase noise would seem to be not a problem. You can use cheap VCOs. The loops are wide enough to make 'em track the reference.
And while Lou reminds us of power consumption, it should also be noted that getting rid of the dissipated power can also be a significant issue.
Dick Jansson, KD1K kd1k@amsat.org kd1k@arrl.net --------------------------- -----Original Message----- From: eagle-bounces@amsat.org [mailto:eagle-bounces@amsat.org] On Behalf Of Louis McFadin Sent: Monday, 02 April, 2007 1455 To: Franklin Antonio Cc: AMSAT Eagle Subject: [eagle] Re: Eagle Microwave Antenna Arrays -- RF concepts
I must remind everybody that power is very critical and will ultimately determine the size and shape of the spacecraft. For the 170w average power predicted at the San Diego meeting, it will require a 250w solar power capability. This could be reduced by reducing the power during eclipse or by relaxing the requirement for operating at all possible sun angles.
Lou McFadin W5DID w5did@mac.com
At 10:05 AM 4/2/2007, Dick Jansson-rr wrote:
And while Lou reminds us of power consumption, it should also be noted that getting rid of the dissipated power can also be a significant issue.
Absolutely correct. However, all of the C-band antenna architectures we have discussed have a PA at each antenna element, and these PAs dissipate most of the power, so the thermal engineering issues are the same in all the architectures.
Ater thinking about this a bit more it seems that the use of multiple PLLs will be a problem. It will be almost impossible to prevent coupling and the VCOs will tend to lock to each other. In order to measure the phase noise of a phase-locked VCXO, I had to operate two copies on battery power with common-base amplifiers (very low S12), attenuators and ferrite beads on the cables to isolate them from each other. Each was in a shielded box so coupling paths were through the power supply, the DBM and backwards through amplifiers.
It should be possible to distribute PSK-modulated RF efficiently if the design is modular. If the antenna array is rectangular, it can be constructed as 9 modules with each containing 4 elements. Electronics for each module is on one PCB containing a 4-way splitter, 4 driver amplifiers, 4 phase shifters and 4 power amplifiers. If the PCB can be made to fit on the back of the antenna module, RF could be distributed from the center and control could be distributed serially around the edges. There are then only 9 RF cables and 9 DC power and control cables back to a common transmitter module.
The receiver array will be larger so the mechanical design could be similar. There is more space so that an RF attenuator can be included or the RF mixed down to a lower frequency and then applied to 4 vaiable-gain IF amplifiers before summing on the module.
73,
John KD6OZH
----- Original Message ----- From: "Franklin Antonio" antonio@qualcomm.com To: K3IO@verizon.net Cc: "AMSAT Eagle" Eagle@amsat.org Sent: Monday, April 02, 2007 05:56 UTC Subject: [eagle] Eagle Microwave Antenna Arrays -- RF concepts
I think I've figured out a good scheme, which may be the right way to build the C-band antenna. It's a bit different than you guys have been thinking.
You've been thinking about how to accomplish the phasing of the RF signal, but ignoring the physical distribution of RF. Physical distribution of C-band RF signal to 35 or so elements is nontrivial. It could be done with splitters and cables and connectors, or printed splitters and microstrip lines, but with any of these schemes there will be amplitude mismatches and various phase shifts associated with just the distribution. If we distribute RF, we'll have to compensate for those amplitude and phase variations. Can be done of course.
How about this. Instead of distributing RF, we distribute a much lower reference frequency which is multiplied up by a PLL at each antenna element. A PLL is one chip these days, so it certainly doesn't cost much in dollars weight or power. We've already got a PA and phase control stuff at each element after all. So maybe we distribute 100 MHz or something easy. It is only gonna be used as a reference, so we don't need to match amplitudes. We design the reference distribution network for high isolation between the outputs, so the individual element circuits have least chance to couple. We don't need to match phases of the distributed reference because we're gonna adjust the phase of each element under software control anyway. (We'll build a calibration table which the software will use as a term it adds into the total phase shift it specifies for each antenna element.)
At each element we have a PLL to generate the RF frequency, followed by a digitally controlled analog phase shifter chip, and a digitally controlled attenuator chip (to adjust out variations in the phase shifter chip vs control input), and a balanced mixer to generate the 180 degree phase shift of BPSK. (If the phase shift chip has a 180 degree input, we can leave out the balanced mixer.) Next, of course, the signal goes to the local PA.
This entire circuit fits on a small board smaller than the antenna element (ie patch) itself, and bolts to the back of the antenna element.
This design is highly modular. The only things distributed are power, reference, data-to-be-modulated, and data to control the individual elements. The controls are on/off, phase, and gain.
It is also possible to move the digital phase shifter to BEFORE the PLL, at which point it may be possible to remove the digital attenuator entirely. You would no longer need it to compensate for the changing attenuation of the phase shifter vs control input, as long as the PLL could handle the range of possible amplitudes. (On the other hand you might still want the attenuator as an easy way to balance the gain and/or power output level of the PAs. With the digitally controlled phase shifter operating at a lower frequency it would likely be more accurate and have less amplitude variation anyway.
This scheme works no matter what the geometry of the array. Software just generates the phases, and sends 'em down some bus.
The thing I have described easily makes unfiltered BPSK or even QPSK. Somebody was talking about possible filtering. That's harder. I hope we don't need filtering. Distributed filtering is not an easy thing.
What do you think?
Via the Eagle mailing list courtesy of AMSAT-NA Eagle@amsat.org http://amsat.org/mailman/listinfo/eagle
Really glad to see discussion finally happening regarding the phased array concept.
I said earlier that this effort will be a "significant" challenge and we need to step out on feasibility "hardware" as soon as we can all settle on an approach. I predict this S2 and C effort will take the most money and time, the most power and the largest chunk of satellite real estate.
Now that we seem to narrowing the possibilities, I'd like to weight in on a few issues to be considered.
I agree with John that multiple LOs can be questionable for the reasons he stated. Assuming you eliminate the injection locking with much mechanical and electrical isolation, you throw in the differential change of phase between all of them due to differences in heat (temperature changes will translate to phase changes in the components of PLL) and you can be chasing phase adjustments all day.
An approach that uses one LO and does the phase changing after it will reduce the "degrees" of complexity we need to handle. It also won't hurt in the categories of cost, power and real estate.
Regards...Bill - N6GHz
John B. Stephensen wrote:
Ater thinking about this a bit more it seems that the use of multiple PLLs will be a problem. It will be almost impossible to prevent coupling and the VCOs will tend to lock to each other. In order to measure the phase noise of a phase-locked VCXO, I had to operate two copies on battery power with common-base amplifiers (very low S12), attenuators and ferrite beads on the cables to isolate them from each other. Each was in a shielded box so coupling paths were through the power supply, the DBM and backwards through amplifiers.
It should be possible to distribute PSK-modulated RF efficiently if the design is modular. If the antenna array is rectangular, it can be constructed as 9 modules with each containing 4 elements. Electronics for each module is on one PCB containing a 4-way splitter, 4 driver amplifiers, 4 phase shifters and 4 power amplifiers. If the PCB can be made to fit on the back of the antenna module, RF could be distributed from the center and control could be distributed serially around the edges. There are then only 9 RF cables and 9 DC power and control cables back to a common transmitter module.
The receiver array will be larger so the mechanical design could be similar. There is more space so that an RF attenuator can be included or the RF mixed down to a lower frequency and then applied to 4 vaiable-gain IF amplifiers before summing on the module.
73,
John KD6OZH
----- Original Message ----- From: "Franklin Antonio" antonio@qualcomm.com To: K3IO@verizon.net Cc: "AMSAT Eagle" Eagle@amsat.org Sent: Monday, April 02, 2007 05:56 UTC Subject: [eagle] Eagle Microwave Antenna Arrays -- RF concepts
I think I've figured out a good scheme, which may be the right way to build the C-band antenna. It's a bit different than you guys have been thinking.
You've been thinking about how to accomplish the phasing of the RF signal, but ignoring the physical distribution of RF. Physical distribution of C-band RF signal to 35 or so elements is nontrivial. It could be done with splitters and cables and connectors, or printed splitters and microstrip lines, but with any of these schemes there will be amplitude mismatches and various phase shifts associated with just the distribution. If we distribute RF, we'll have to compensate for those amplitude and phase variations. Can be done of course.
How about this. Instead of distributing RF, we distribute a much lower reference frequency which is multiplied up by a PLL at each antenna element. A PLL is one chip these days, so it certainly doesn't cost much in dollars weight or power. We've already got a PA and phase control stuff at each element after all. So maybe we distribute 100 MHz or something easy. It is only gonna be used as a reference, so we don't need to match amplitudes. We design the reference distribution network for high isolation between the outputs, so the individual element circuits have least chance to couple. We don't need to match phases of the distributed reference because we're gonna adjust the phase of each element under software control anyway. (We'll build a calibration table which the software will use as a term it adds into the total phase shift it specifies for each antenna element.)
At each element we have a PLL to generate the RF frequency, followed by a digitally controlled analog phase shifter chip, and a digitally controlled attenuator chip (to adjust out variations in the phase shifter chip vs control input), and a balanced mixer to generate the 180 degree phase shift of BPSK. (If the phase shift chip has a 180 degree input, we can leave out the balanced mixer.) Next, of course, the signal goes to the local PA.
This entire circuit fits on a small board smaller than the antenna element (ie patch) itself, and bolts to the back of the antenna element.
This design is highly modular. The only things distributed are power, reference, data-to-be-modulated, and data to control the individual elements. The controls are on/off, phase, and gain.
It is also possible to move the digital phase shifter to BEFORE the PLL, at which point it may be possible to remove the digital attenuator entirely. You would no longer need it to compensate for the changing attenuation of the phase shifter vs control input, as long as the PLL could handle the range of possible amplitudes. (On the other hand you might still want the attenuator as an easy way to balance the gain and/or power output level of the PAs. With the digitally controlled phase shifter operating at a lower frequency it would likely be more accurate and have less amplitude variation anyway.
This scheme works no matter what the geometry of the array. Software just generates the phases, and sends 'em down some bus.
The thing I have described easily makes unfiltered BPSK or even QPSK. Somebody was talking about possible filtering. That's harder. I hope we don't need filtering. Distributed filtering is not an easy thing.
What do you think?
Via the Eagle mailing list courtesy of AMSAT-NA Eagle@amsat.org http://amsat.org/mailman/listinfo/eagle
Via the Eagle mailing list courtesy of AMSAT-NA Eagle@amsat.org http://amsat.org/mailman/listinfo/eagle
At 01:35 PM 4/2/2007, Bill Ress wrote:
you throw in the differential change of phase between all of them due to differences in heat (temperature changes will translate to phase changes in the components of PLL) and you can be chasing phase adjustments all day.
Uh... That's a pretty nonquantitative statement. How much phase change do you think will occur due to temperature? Why do you narrow in on a PLL as something that might have phase change due to temperature? All components do! The PA for example does. Cables do. Any architecture for this antenna has phase shifts vs temperature. I believe that the temperature differential between elements will be small (after all, the PAs are all dissipating the same amount of power, and the satlelite is spinning), and also the phase change as a function of temperature will be very small. In the end it will be irrelevant.
Tom can do some simulation to demonstrate how much phase error we can tolerate. I was only gonna put phase shifters with something like 4 bits of control anyway. (Probably overkill.) Those are 22.5 degree steps. That means the phase error due to quantization is between 0 and 11.25 degrees, and will average 6.125 degrees. There is another error term caused by the phase shifter's internal error. Maybe that's a few degrees. If you want to throw in a few degrees for temperature variations, everything will still work fine.
Remember that the resulting signal is the sum of 35 or so elements, so a phase error in one element has a very small affect on the composite signal. This summing is a huge effect.
At 12:45 PM 4/2/2007, John B. Stephensen wrote:
Ater thinking about this a bit more it seems that the use of multiple PLLs will be a problem. It will be almost impossible to prevent coupling and the VCOs will tend to lock to each other.
Nah. They are "locked to each other" by design. All running at exactly the same frequency.
The frequencies should converge rapidly. I'm not sure if they would end up at a constant phase or wander around each other.
73,
John KD6OZH
----- Original Message ----- From: "Franklin Antonio" antonio@qualcomm.com To: "John B. Stephensen" kd6ozh@comcast.net Cc: K3IO@verizon.net; "AMSAT Eagle" Eagle@amsat.org Sent: Monday, April 02, 2007 21:11 UTC Subject: Re: [eagle] Eagle Microwave Antenna Arrays -- RF concepts -- quad element RF modules
At 12:45 PM 4/2/2007, John B. Stephensen wrote:
Ater thinking about this a bit more it seems that the use of multiple PLLs will be a problem. It will be almost impossible to prevent coupling and the VCOs will tend to lock to each other.
Nah. They are "locked to each other" by design. All running at exactly the same frequency.
participants (9)
-
Alan Bloom -
Bill Ress -
Dick Jansson-rr -
Franklin Antonio -
John B. Stephensen -
Louis McFadin -
Matt Ettus -
Robert McGwier -
Tom Clark, K3IO