IMD, Phase Noise, a New Spur, and EMI suggestions
All,
I've started a new log here http://www.juanr.com/pages/hobbies/ham_radio/Eagle/IMD+Phase_Noise.htm .
I've repeated some preliminary phase noise tests using a simulated PAVE PAWS signal, along with a simulated Ham uplink signal. In the process of looking at those signals with the SDR-IQ I discovered another spur directly in the middle of the IF. I don't know the exact source but it is generated inside the receiver, and definitely not from the CAN-Do module.
I've also come to one general conclusion that I would like to bring up for discussion... I believe all switching power supplies on Eagle should be running at frequencies as high as possible - around 1 MHz would be a good ball park. That would put most spurs outside the passband of the payloads, and make EMI filtering much easier. The 5 kHz CAN-Do switching power supply needs to be completely redesigned in my opinion. I believe that if it is not cleaned up the impact will be devastating, and the full scope will likely not be known until integration and test when it is too late to do anything about it.
Having said all that, I also believe that we need to come up with EMI standards for everything on Eagle. Specifically we need:
1) A definition of the noise amplitude and spectral content to be expected from the Eagle power distribution point
2) A repeatable way to reproduce that noisy DC source in our labs for testing
3) A simple and effective way to test to the spec that we create
I think the way to make this work is to create a noise source that consists of several representative switching power supplies that are intended to be flown on Eagle. The exact makeup of this test fixture would be specified in the EMI requirements. Then, all payloads would need to be designed to operate without impairment while powered by this noisy power source. The EMI requirement would also spell out exactly what constituted impairment. They also need to keep conducted noise leaving the payload down to some specified level. More on that in a minute...
Here's a possible layout for a noise source:
The one ohm resistors account for lead resistance and allow the three switching power supplies to all modulate the power bus, along with whatever conducted EMI comes back from the devise under test (your payload.) It would be easy to crank out three or four of these noise sources for distribution to groups that need them. That way everyone would be testing using the same conducted EMI.
The gold standards for EMI are MIL-STD-461 and -462. They break EMI into four categories:
* CS - Conducted Susceptibility (the 70 cm Receiver suffers from this)
* RS - Radiated Susceptibility (the 70 cm Receiver suffers from this)
* CE - Conducted Emissions (The CAN-Do Module is guilty of this)
* RE - Radiated Emissions (The CAN-Do Module is guilty of this)
This noise source would provide a way to test for compliance with our CS requirement. RS and RE can be dealt with by shielding. That leaves CE. We'd need a way to insure that noise coming out of a payload is within spec. We'd need another test fixture for this.
I hope this stimulated some discussion. I'm eager to hear what you all think...
73,
Juan - WA6HTP
Juan:
I believe you are taking a very professional and intelligent approach to the problem. Your description of the situation and what should be done, brings back memories from quite a few years ago. It all sounds familiar.
All of this needs to go into an interface document and into specifications for the various payloads and other devices.
73,
Bill, W3XO ----- Original Message ----- From: Juan Rivera To: [email protected] ; Bill Ress ; Dave Black (Home) ; Dave Black (Work) ; Dave hartzell ; David Smith ; Don Ferguson ; Juan. Rivera (Home) ; Juan.Rivera (Work) ; Samsonoff@Mac. Com Cc: Juan.Rivera (Work) Sent: Wednesday, June 27, 2007 11:36 PM Subject: [eagle] IMD, Phase Noise, a New Spur, and EMI suggestions
All,
I've started a new log here.
I've repeated some preliminary phase noise tests using a simulated PAVE PAWS signal, along with a simulated Ham uplink signal. In the process of looking at those signals with the SDR-IQ I discovered another spur directly in the middle of the IF. I don't know the exact source but it is generated inside the receiver, and definitely not from the CAN-Do module.
I've also come to one general conclusion that I would like to bring up for discussion... I believe all switching power supplies on Eagle should be running at frequencies as high as possible - around 1 MHz would be a good ball park. That would put most spurs outside the passband of the payloads, and make EMI filtering much easier. The 5 kHz CAN-Do switching power supply needs to be completely redesigned in my opinion. I believe that if it is not cleaned up the impact will be devastating, and the full scope will likely not be known until integration and test when it is too late to do anything about it.
Having said all that, I also believe that we need to come up with EMI standards for everything on Eagle. Specifically we need:
1) A definition of the noise amplitude and spectral content to be expected from the Eagle power distribution point
2) A repeatable way to reproduce that noisy DC source in our labs for testing
3) A simple and effective way to test to the spec that we create
I think the way to make this work is to create a noise source that consists of several representative switching power supplies that are intended to be flown on Eagle. The exact makeup of this test fixture would be specified in the EMI requirements. Then, all payloads would need to be designed to operate without impairment while powered by this noisy power source. The EMI requirement would also spell out exactly what constituted impairment. They also need to keep conducted noise leaving the payload down to some specified level. More on that in a minute...
Here's a possible layout for a noise source:
The one ohm resistors account for lead resistance and allow the three switching power supplies to all modulate the power bus, along with whatever conducted EMI comes back from the devise under test (your payload.) It would be easy to crank out three or four of these noise sources for distribution to groups that need them. That way everyone would be testing using the same conducted EMI.
The gold standards for EMI are MIL-STD-461 and -462. They break EMI into four categories:
· CS - Conducted Susceptibility (the 70 cm Receiver suffers from this)
· RS - Radiated Susceptibility (the 70 cm Receiver suffers from this)
· CE - Conducted Emissions (The CAN-Do Module is guilty of this)
· RE - Radiated Emissions (The CAN-Do Module is guilty of this)
This noise source would provide a way to test for compliance with our CS requirement. RS and RE can be dealt with by shielding. That leaves CE. We'd need a way to insure that noise coming out of a payload is within spec. We'd need another test fixture for this.
I hope this stimulated some discussion. I'm eager to hear what you all think...
73,
Juan - WA6HTP
------------------------------------------------------------------------------
_______________________________________________ Via the Eagle mailing list courtesy of AMSAT-NA [email protected] http://amsat.org/mailman/listinfo/eagle
Juan:
If we have had this discussion before, forgive me. I have misplaced both the interaction and my memory.
What is the generator for the "Pave Paws" source? I ask this because the phase noise for the "Pave Paws" source does not seem to be included in your consideration.
Pave Paws is a pulsed instrument. It will not have its energy concentrated at a single frequency. Its energy will be spread considerably and much of that energy will fall outside of the front end filter of the 70 cm RX. A more realistic test would compute that a -70.2 dBm Pave Paw signal will put (say) -90 dBm inside our passband and further, it will be on/off. Since it is on/off, there will be gaps where the amateur signal is not drastically impacted. A receive system on the ground, necessarily narrow band, will provide processing gain. If we do a good job of writing code, and we do not run out of processing cycles, it would be nice to do pave paws pulse detection and subtraction.
My point here is, the pave paws situation is a great deal more complex than your test is capable of revealing.
Bob
Bob,
I'm only simulating a clean CW signal of the amplitude predicted by John as worst case. I've been thinking of a field trip up to Marysville to try to capture actual PAVE PAWS using the Matt Ettus USRP board. That might give us some real world signals to work with. But until then I have no idea what PAVE PAWS phase noise would look like, or if the data in the ATP adendum is correct or not. It's derived from open source information released during environmental impact proceedings for a proposed PAVE PAWS upgrade.
Have you looked at the PAVE PAWS addendum in the ATP? Do you think is is accurate? I want to make sure I'm on the same page with you. Pages 50 through 53 spell out the bandplan, pulse duration for both modes, repetition rates, etc. The tracking mode pulses are 16 milliseconds in duraction and chirp 1 MHz. Yes, that's a pulse, but it's extrememly long. There are four channels that look like they will have the most impact - 434.4, 435.6, and 436.8 and 438.0 MHz.
Assuming that this information is accurate, and assuming that the uplinked data decoder dies when the receiver phase noise I'm seeing buries the uplink signal, then my thinking goes like this:
Based on the test results I posted the other night, a clean signal at -43 dBm (PAVE PAWS) will bury a clean adjacent signal at -112 dBm (ham uplink) when it gets within about 7 kHz. So if we do the math - 16 mSec to move 1 MHz = 0.24 milliseconds to move 15 kHz (the distance required to move from one side of the uplink to the other) The rep rate for tracking mode is 54 milliseconds so I would expect to see a .24 mSec drop out every 54 mSec when illuminated with PAVE PAWS tracking mode EMI, under worst case conditions. I would think that if you bulit in enough forward error correction to deal with that then you could uplink right through PAVE PAWS, assuming that my simple test produced accurate results.
Something else to consider is the recovery time when the receiver is hit by a strong pulse. I haven't looked into that yet but it could add some time to the .24 mSec drop out.
Bottom line - Yes, I know my test is simpistic but it's the best I can do with what I have available now. If we can capture actual PAVE PAWS using the USRP board then we can conduct much more sophisticated tests using that digitized PAVE PAWS signal as the EMI source.
73,
Juan
-----Original Message----- From: Robert McGwier [mailto:[email protected]] Sent: Thursday, June 28, 2007 5:01 AM To: [email protected] Cc: [email protected]; Bill Ress; Dave Black (Home); Dave Black (Work); Dave hartzell; David Smith; Don Ferguson; Rivera, Juan A.; Samsonoff@Mac. Com Subject: Re: [eagle] IMD, Phase Noise, a New Spur, and EMI suggestions
Juan:
If we have had this discussion before, forgive me. I have misplaced both the interaction and my memory.
What is the generator for the "Pave Paws" source? I ask this because the phase noise for the "Pave Paws" source does not seem to be included in your consideration.
Pave Paws is a pulsed instrument. It will not have its energy concentrated at a single frequency. Its energy will be spread considerably and much of that energy will fall outside of the front end filter of the 70 cm RX. A more realistic test would compute that a -70.2 dBm Pave Paw signal will put (say) -90 dBm inside our passband and further, it will be on/off. Since it is on/off, there will be gaps where the amateur signal is not drastically impacted. A receive system on the ground, necessarily narrow band, will provide processing gain. If we do a good job of writing code, and we do not run out of processing cycles, it would be nice to do pave paws pulse detection and subtraction.
My point here is, the pave paws situation is a great deal more complex than your test is capable of revealing.
Bob
-- AMSAT Director and VP Engineering. Member: ARRL, AMSAT-DL, TAPR, Packrats, NJQRP, QRP ARCI, QCWA, FRC. ARRL SDR WG Chair "If you're going to be crazy, you have to get paid for it or else you're going to be locked up." Hunter S. Thompson
It will be useful to see an actual PAVE PAWS signal in order to determine the level of spurious signals in the transmitter's output. If we are unlucky, we will see spurs in the PAVE PAWS signal that are higher than those generated in the receiver and it will impede our effort to mitigate its effects. If the signal is very clean, we may not be able to just play it back. The DACs on the USRP will have spurs that are 80 to 90 dB down.
73,
John KD6OZH
----- Original Message ----- From: "Rivera, Juan A." [email protected] To: "Robert McGwier" [email protected]; [email protected] Cc: "David Smith" [email protected]; "Dave Black (Work)" [email protected]; "Dave Black (Home)" [email protected]; [email protected]; "Samsonoff@Mac. Com" [email protected] Sent: Thursday, June 28, 2007 15:03 UTC Subject: [eagle] Re: IMD, Phase Noise, a New Spur, and EMI suggestions
Bob,
I'm only simulating a clean CW signal of the amplitude predicted by John as worst case. I've been thinking of a field trip up to Marysville to try to capture actual PAVE PAWS using the Matt Ettus USRP board. That might give us some real world signals to work with. But until then I have no idea what PAVE PAWS phase noise would look like, or if the data in the ATP adendum is correct or not. It's derived from open source information released during environmental impact proceedings for a proposed PAVE PAWS upgrade.
Have you looked at the PAVE PAWS addendum in the ATP? Do you think is is accurate? I want to make sure I'm on the same page with you. Pages 50 through 53 spell out the bandplan, pulse duration for both modes, repetition rates, etc. The tracking mode pulses are 16 milliseconds in duraction and chirp 1 MHz. Yes, that's a pulse, but it's extrememly long. There are four channels that look like they will have the most impact - 434.4, 435.6, and 436.8 and 438.0 MHz.
Assuming that this information is accurate, and assuming that the uplinked data decoder dies when the receiver phase noise I'm seeing buries the uplink signal, then my thinking goes like this:
Based on the test results I posted the other night, a clean signal at -43 dBm (PAVE PAWS) will bury a clean adjacent signal at -112 dBm (ham uplink) when it gets within about 7 kHz. So if we do the math - 16 mSec to move 1 MHz = 0.24 milliseconds to move 15 kHz (the distance required to move from one side of the uplink to the other) The rep rate for tracking mode is 54 milliseconds so I would expect to see a .24 mSec drop out every 54 mSec when illuminated with PAVE PAWS tracking mode EMI, under worst case conditions. I would think that if you bulit in enough forward error correction to deal with that then you could uplink right through PAVE PAWS, assuming that my simple test produced accurate results.
Something else to consider is the recovery time when the receiver is hit by a strong pulse. I haven't looked into that yet but it could add some time to the .24 mSec drop out.
Bottom line - Yes, I know my test is simpistic but it's the best I can do with what I have available now. If we can capture actual PAVE PAWS using the USRP board then we can conduct much more sophisticated tests using that digitized PAVE PAWS signal as the EMI source.
73,
Juan
-----Original Message----- From: Robert McGwier [mailto:[email protected]] Sent: Thursday, June 28, 2007 5:01 AM To: [email protected] Cc: [email protected]; Bill Ress; Dave Black (Home); Dave Black (Work); Dave hartzell; David Smith; Don Ferguson; Rivera, Juan A.; Samsonoff@Mac. Com Subject: Re: [eagle] IMD, Phase Noise, a New Spur, and EMI suggestions
Juan:
If we have had this discussion before, forgive me. I have misplaced both the interaction and my memory.
What is the generator for the "Pave Paws" source? I ask this because the phase noise for the "Pave Paws" source does not seem to be included in your consideration.
Pave Paws is a pulsed instrument. It will not have its energy concentrated at a single frequency. Its energy will be spread considerably and much of that energy will fall outside of the front end filter of the 70 cm RX. A more realistic test would compute that a -70.2 dBm Pave Paw signal will put (say) -90 dBm inside our passband and further, it will be on/off. Since it is on/off, there will be gaps where the amateur signal is not drastically impacted. A receive system on the ground, necessarily narrow band, will provide processing gain. If we do a good job of writing code, and we do not run out of processing cycles, it would be nice to do pave paws pulse detection and subtraction.
My point here is, the pave paws situation is a great deal more complex than your test is capable of revealing.
Bob
-- AMSAT Director and VP Engineering. Member: ARRL, AMSAT-DL, TAPR, Packrats, NJQRP, QRP ARCI, QCWA, FRC. ARRL SDR WG Chair "If you're going to be crazy, you have to get paid for it or else you're going to be locked up." Hunter S. Thompson
Via the Eagle mailing list courtesy of AMSAT-NA [email protected] http://amsat.org/mailman/listinfo/eagle
The receiver wasn't designed to be linear with a full-strength PAVE PAWS signal in the passband. The effect of strong signals needs to be measured so that the in-band dynamic range (blocking and two-tone) can be determined, but the receiver will only operate linearly with radar pulses that are not aimed directly at Eagle.
PAVE PAWS signals offset by 750 kHz or more from the center frequency should have little effect (less than 1 dB increase in noise). This does require that the signal generator have low phase noise at this offset. The receiver first LO phase noise at a 750 kHz offset must be less than -132.4 dBc/Hz to acheive this so the phase noise of the signal source must be less than -138.4 dBc/Hz to acheive a 1 dB measurment error. It should be much lower so that we can see the true characteristics of the receiver.
The spur at -134 dBm could be a problem for text-mode uplinks so its source needs to be determined. Is the internal switching regulator on?
Table 4 in the new requirements document provides a good summary of expected signal levels at the input and output of the receiver. Spurs may need to be lower than originally assumed as the text-mode uplinks are very low power.
73,
John KD6OZH ----- Original Message ----- From: Juan Rivera To: [email protected] ; Bill Ress ; Dave Black (Home) ; Dave Black (Work) ; Dave hartzell ; David Smith ; Don Ferguson ; Juan. Rivera (Home) ; Juan.Rivera (Work) ; Samsonoff@Mac. Com Cc: Juan.Rivera (Work) Sent: Thursday, June 28, 2007 04:36 UTC Subject: [eagle] IMD, Phase Noise, a New Spur, and EMI suggestions
All,
I've started a new log here.
I've repeated some preliminary phase noise tests using a simulated PAVE PAWS signal, along with a simulated Ham uplink signal. In the process of looking at those signals with the SDR-IQ I discovered another spur directly in the middle of the IF. I don't know the exact source but it is generated inside the receiver, and definitely not from the CAN-Do module.
I've also come to one general conclusion that I would like to bring up for discussion... I believe all switching power supplies on Eagle should be running at frequencies as high as possible - around 1 MHz would be a good ball park. That would put most spurs outside the passband of the payloads, and make EMI filtering much easier. The 5 kHz CAN-Do switching power supply needs to be completely redesigned in my opinion. I believe that if it is not cleaned up the impact will be devastating, and the full scope will likely not be known until integration and test when it is too late to do anything about it.
Having said all that, I also believe that we need to come up with EMI standards for everything on Eagle. Specifically we need:
1) A definition of the noise amplitude and spectral content to be expected from the Eagle power distribution point
2) A repeatable way to reproduce that noisy DC source in our labs for testing
3) A simple and effective way to test to the spec that we create
I think the way to make this work is to create a noise source that consists of several representative switching power supplies that are intended to be flown on Eagle. The exact makeup of this test fixture would be specified in the EMI requirements. Then, all payloads would need to be designed to operate without impairment while powered by this noisy power source. The EMI requirement would also spell out exactly what constituted impairment. They also need to keep conducted noise leaving the payload down to some specified level. More on that in a minute...
Here's a possible layout for a noise source:
The one ohm resistors account for lead resistance and allow the three switching power supplies to all modulate the power bus, along with whatever conducted EMI comes back from the devise under test (your payload.) It would be easy to crank out three or four of these noise sources for distribution to groups that need them. That way everyone would be testing using the same conducted EMI.
The gold standards for EMI are MIL-STD-461 and -462. They break EMI into four categories:
· CS - Conducted Susceptibility (the 70 cm Receiver suffers from this)
· RS - Radiated Susceptibility (the 70 cm Receiver suffers from this)
· CE - Conducted Emissions (The CAN-Do Module is guilty of this)
· RE - Radiated Emissions (The CAN-Do Module is guilty of this)
This noise source would provide a way to test for compliance with our CS requirement. RS and RE can be dealt with by shielding. That leaves CE. We'd need a way to insure that noise coming out of a payload is within spec. We'd need another test fixture for this.
I hope this stimulated some discussion. I'm eager to hear what you all think...
73,
Juan - WA6HTP
------------------------------------------------------------------------------
_______________________________________________ Via the Eagle mailing list courtesy of AMSAT-NA [email protected] http://amsat.org/mailman/listinfo/eagle
John,
See results HERE http://www.juanr.com/pages/hobbies/ham_radio/Eagle/IMD+Phase_Noise.htm .
Juan
_____
From: John B. Stephensen [mailto:[email protected]] Sent: Thursday, June 28, 2007 12:26 PM To: [email protected]; [email protected]; Bill Ress; Dave Black (Home); Dave Black (Work); Dave hartzell; David Smith; Don Ferguson; Juan.Rivera (Work); Samsonoff@Mac. Com Cc: Juan.Rivera (Work) Subject: Re: [eagle] IMD, Phase Noise, a New Spur, and EMI suggestions
The receiver wasn't designed to be linear with a full-strength PAVE PAWS signal in the passband. The effect of strong signals needs to be measured so that the in-band dynamic range (blocking and two-tone) can be determined, but the receiver will only operate linearly with radar pulses that are not aimed directly at Eagle.
PAVE PAWS signals offset by 750 kHz or more from the center frequency should have little effect (less than 1 dB increase in noise). This does require that the signal generator have low phase noise at this offset. The receiver first LO phase noise at a 750 kHz offset must be less than -132.4 dBc/Hz to acheive this so the phase noise of the signal source must be less than -138.4 dBc/Hz to acheive a 1 dB measurment error. It should be much lower so that we can see the true characteristics of the receiver.
The spur at -134 dBm could be a problem for text-mode uplinks so its source needs to be determined. Is the internal switching regulator on?
Table 4 in the new requirements document provides a good summary of expected signal levels at the input and output of the receiver. Spurs may need to be lower than originally assumed as the text-mode uplinks are very low power.
73,
John
KD6OZH
----- Original Message -----
From: Juan mailto:[email protected] Rivera
To: [email protected] ; Bill Ress mailto:[email protected] ; Dave Black mailto:[email protected] (Home) ; Dave Black (Work) mailto:[email protected] ; Dave hartzell mailto:[email protected] ; David Smith mailto:[email protected] ; Don Ferguson mailto:[email protected] ; Juan. mailto:[email protected] Rivera (Home) ; Juan.Rivera (Work) mailto:[email protected] ; Samsonoff@Mac. Com mailto:Samsonoff@Mac.%20Com
Cc: Juan.Rivera mailto:[email protected] (Work)
Sent: Thursday, June 28, 2007 04:36 UTC
Subject: [eagle] IMD, Phase Noise, a New Spur, and EMI suggestions
All,
I've started a new log here http://www.juanr.com/pages/hobbies/ham_radio/Eagle/IMD+Phase_Noise.htm .
I've repeated some preliminary phase noise tests using a simulated PAVE PAWS signal, along with a simulated Ham uplink signal. In the process of looking at those signals with the SDR-IQ I discovered another spur directly in the middle of the IF. I don't know the exact source but it is generated inside the receiver, and definitely not from the CAN-Do module.
I've also come to one general conclusion that I would like to bring up for discussion... I believe all switching power supplies on Eagle should be running at frequencies as high as possible - around 1 MHz would be a good ball park. That would put most spurs outside the passband of the payloads, and make EMI filtering much easier. The 5 kHz CAN-Do switching power supply needs to be completely redesigned in my opinion. I believe that if it is not cleaned up the impact will be devastating, and the full scope will likely not be known until integration and test when it is too late to do anything about it.
Having said all that, I also believe that we need to come up with EMI standards for everything on Eagle. Specifically we need:
1) A definition of the noise amplitude and spectral content to be expected from the Eagle power distribution point
2) A repeatable way to reproduce that noisy DC source in our labs for testing
3) A simple and effective way to test to the spec that we create
I think the way to make this work is to create a noise source that consists of several representative switching power supplies that are intended to be flown on Eagle. The exact makeup of this test fixture would be specified in the EMI requirements. Then, all payloads would need to be designed to operate without impairment while powered by this noisy power source. The EMI requirement would also spell out exactly what constituted impairment. They also need to keep conducted noise leaving the payload down to some specified level. More on that in a minute...
Here's a possible layout for a noise source:
The one ohm resistors account for lead resistance and allow the three switching power supplies to all modulate the power bus, along with whatever conducted EMI comes back from the devise under test (your payload.) It would be easy to crank out three or four of these noise sources for distribution to groups that need them. That way everyone would be testing using the same conducted EMI.
The gold standards for EMI are MIL-STD-461 and -462. They break EMI into four categories:
* CS - Conducted Susceptibility (the 70 cm Receiver suffers from this)
* RS - Radiated Susceptibility (the 70 cm Receiver suffers from this)
* CE - Conducted Emissions (The CAN-Do Module is guilty of this)
* RE - Radiated Emissions (The CAN-Do Module is guilty of this)
This noise source would provide a way to test for compliance with our CS requirement. RS and RE can be dealt with by shielding. That leaves CE. We'd need a way to insure that noise coming out of a payload is within spec. We'd need another test fixture for this.
I hope this stimulated some discussion. I'm eager to hear what you all think...
73,
Juan - WA6HTP
_____
_______________________________________________ Via the Eagle mailing list courtesy of AMSAT-NA [email protected] http://amsat.org/mailman/listinfo/eagle
participants (5)
-
Bill Tynan -
John B. Stephensen -
Juan Rivera -
Rivera, Juan A. -
Robert McGwier