Re: inquiry about homebrew az-el systems
Unless we want lightning fast antenna rotation, we're going to need some sort of gearbox with virtually any motor we're likely to find. We could gear the drive to a simpler pot as well.
Gus, I used a pulse width modulator to slow my DC motor down with out having to loose a lot of power. This eliminated the resistor and the motor ran cooler. Mine cost $80.00 usd out of Canada but I understand that they are nothing more then modified fluorescent light dimmers. My thought would be to count the pulses instead of a pot. I think we will be stuck with gear motors if the goal is to go small and have lots of torque. Even with the PWM mine was a gear motor but it beats a big restive load hands down.
I would like to move away from stops and find another way to establish 0~360az and 0~90/180ex indication so as to enable continuous rotation to do continuous sky scan or search and rescue with out having to wind up and damage my coax. I am not sure what battle ship radar uses for constant rotation but that would be the ticket.
Bob Campbell KB3PMR
On 03/02/2013 06:55 AM, Robert C. Campbell wrote:
My thought would be to count the pulses instead of a pot. I think we will be stuck with gear motors if the goal is to go small and have lots of torque.
Counting pulses will only be practical if there is a direct correlation between pulse count and angular displacement. That means you are pulsing a stepper and counting those pulses, or you are activating an ordinary motor and the pulse train is coming back from some sort of shaft encoder. But I prefer this approach (digital) to reading the voltage on a pot (linear/analog). I am mentally working on a straightforward, easy-to-build shaft-encoder. But how accurate do we need to be, if we're building a system for small antennas with reasonable wide beam-widths?
I am coming to the conclusion that for a globally ubiquitous supply of second hand motors, we will have to turn to the automobile junk yard. Windshield wiper motors and window wider motors spring instantly to mind. Newer cars have motors everywhere. Repositioning seats, adjusting rear view mirrors, opening and closing tailgates. Perhaps central locking mechanisms can be adopted to lock the rotors in position between moves to prevent weather-cocking. I'm fairly sure that virtually anywhere in the world you could get your hands on two wiper motors from a junk yard without breaking the bank.
I would like to move away from stops and find another way to establish 0~360az and 0~90/180ex indication so as to enable continuous rotation to do continuous sky scan or search and rescue with out having to wind up and damage my coax.
Continuous rotation probably requires some sort of coaxial slip-ring system. Google "coax rotary joint." They look expensive, and I can't help worrying about insertion loss. But perhaps limit switch is the wrong word. But with a cam on your shaft and a micro-switch, you have hard position-detection of two places on the circle: when the switch goes ON and when the switch goes OFF. (In theory, enough switches and cams and you could detect any number of places on the circle.) I think it's useful, e3specially with a pulse counting system, to be able to confirm your position at least a couple points around your circle. This would allow your controller to calibrate itself when ever it needed to, and possibly to double-check itself during normal operation. Whether or not you choose to wire a hard-shutoff based on the output of these switches depends on whether you have a continuous-rotation coax joint or not.
I am not sure what battle ship radar uses for constant rotation but that would be the ticket.
Bob Campbell KB3PMR
MPJA has a large selection of stepper motors available:
http://www.mpja.com/Stepper-Motors/products/101/
Bob Martinson, K1REM
Just noticed this thread and caught up.
While rotor controllers are indeed a dime a dozen, I think we could do a lot better than any of them.
Your typical Yaesu/Kenpro rotor uses a 24V AC 2-phase induction motor. The control box applies 50/60 Hz AC directly to one winding and to the other through a capacitor. The capacitor creates a phase shift in the current through the second winding, creating a rotating magnetic field within the motor that drags the rotor in one direction or the other. You reverse the motor by applying AC directly to one winding or the other.
Although this design is extremely common, it has several highly non-ideal features. First, the current through the second winding isn't actually in phase quadrature (90 degrees) with the first. It's somewhat less due to the series resistances of the winding and capacitor.
Second, the current amplitudes in the two windings are not the same, and for the same reason -- series resistances. This means less torque and more motor heating than could otherwise be produced for the same input voltage.
Third, the motor has only one synchronous speed: 50 or 60 Hz. Stalled rotor torque is rather low, especially for a non-ideal supply.
What you *really* want is a variable frequency, variable voltage (VFVV) inverter producing two phases in exact quadrature (same amplitude, 90 degrees with respect to each other). You can smoothly vary the speed from a dead stop to faster than 60 Hz and with more torque at every speed, making it easy to track a continuously moving satellite with a narrow antenna. And you don't wear out the brakes and constantly flex the masts and booms until the clamps all work loose.
You can even use the motors as brakes by sending a small amount of DC current through them. It doesn't take much, as this essentially creates a DC generator with a shorted output, and that torque is amplified by the gear train.
The necessary waveforms could be generated with the PWM channels in an Arduino or similar microcontroller and amplified with the power MOSFET H-bridges common in robotics.
I do see several rotors using DC motors, plus several people suggesting them here. While they're somewhat easier to vary in speed (you just vary the average DC voltage with a PWM drive) you have to remember these motors contain brushes rubbing on commutators, and that makes them far less reliable than AC induction motors, which are famously simple, rugged and reliable. There's a reason AC motors are universal in the modern generation of hybrid and battery electric vehicles even though most hobby conversions still use DC motors.
As for position feedback, what about one of the cheap, modern IMU devices, like the Pololu MinIMU-9. I've been playing with this particular board, which contains a 3-axis accelerometer, magnetometer and rotational gyro. Just mount one on the antenna boom and directly measure the antenna position. The accelerometer will give elevation without any calibration at all. The magnetometer can read azimuth with a lookup table for your local magnetic declination, and any local magnetic distortions could be removed with a one-time calibration. And the gyro will quickly tell you if the antenna is out of balance or has stalled.
--Phil
Phil,
I'm glad you find the idea of interest, because I'm sure you could greatly contribute towards the idea.
Please note that we haven't simply been thinking of designing a better 5400/5500. We've also been thinking about a design that could be used in the field and after a disaster/in an emergency. And a design that could be replicated in countries around the globe. Hence 12v automotive motors and bicycle sprockets were all part of the brainstormed recipe!
Your 9dof IMU idea is sexy! Just think -- with TWO of them, you could compensate for the motion of the station, when operating from, say, a boat or a vehicle under way. (Nobody say "GPS" please!) One RTC chip, a USB interface to the laptop or bluetooth interface to the Android tablet... But it sounds less and less like you will be able to source much of it in the "third world." Which is where I happen to live.
Still, it sure sounds interesting! What do you think it would cost to put one together?
On 03/08/2013 02:35 AM, Phil Karn wrote:
Just noticed this thread and caught up.
While rotor controllers are indeed a dime a dozen, I think we could do a lot better than any of them.
Your typical Yaesu/Kenpro rotor uses a 24V AC 2-phase induction motor. The control box applies 50/60 Hz AC directly to one winding and to the other through a capacitor. The capacitor creates a phase shift in the current through the second winding, creating a rotating magnetic field within the motor that drags the rotor in one direction or the other. You reverse the motor by applying AC directly to one winding or the other.
Although this design is extremely common, it has several highly non-ideal features. First, the current through the second winding isn't actually in phase quadrature (90 degrees) with the first. It's somewhat less due to the series resistances of the winding and capacitor.
Second, the current amplitudes in the two windings are not the same, and for the same reason -- series resistances. This means less torque and more motor heating than could otherwise be produced for the same input voltage.
Third, the motor has only one synchronous speed: 50 or 60 Hz. Stalled rotor torque is rather low, especially for a non-ideal supply.
What you *really* want is a variable frequency, variable voltage (VFVV) inverter producing two phases in exact quadrature (same amplitude, 90 degrees with respect to each other). You can smoothly vary the speed from a dead stop to faster than 60 Hz and with more torque at every speed, making it easy to track a continuously moving satellite with a narrow antenna. And you don't wear out the brakes and constantly flex the masts and booms until the clamps all work loose.
You can even use the motors as brakes by sending a small amount of DC current through them. It doesn't take much, as this essentially creates a DC generator with a shorted output, and that torque is amplified by the gear train.
The necessary waveforms could be generated with the PWM channels in an Arduino or similar microcontroller and amplified with the power MOSFET H-bridges common in robotics.
I do see several rotors using DC motors, plus several people suggesting them here. While they're somewhat easier to vary in speed (you just vary the average DC voltage with a PWM drive) you have to remember these motors contain brushes rubbing on commutators, and that makes them far less reliable than AC induction motors, which are famously simple, rugged and reliable. There's a reason AC motors are universal in the modern generation of hybrid and battery electric vehicles even though most hobby conversions still use DC motors.
As for position feedback, what about one of the cheap, modern IMU devices, like the Pololu MinIMU-9. I've been playing with this particular board, which contains a 3-axis accelerometer, magnetometer and rotational gyro. Just mount one on the antenna boom and directly measure the antenna position. The accelerometer will give elevation without any calibration at all. The magnetometer can read azimuth with a lookup table for your local magnetic declination, and any local magnetic distortions could be removed with a one-time calibration. And the gyro will quickly tell you if the antenna is out of balance or has stalled.
--Phil
Sent via [email protected]. Opinions expressed are those of the author. Not an AMSAT-NA member? Join now to support the amateur satellite program! Subscription settings: http://amsat.org/mailman/listinfo/amsat-bb
On 03/08/2013 12:08 AM, Gus wrote:
Still, it sure sounds interesting! What do you think it would cost to put one together?
Dunno. I'd have to build one. It would run from a DC supply, because the inverter would convert that to AC at the necessary voltage and frequency. Because the rotor motors take a nominal 24V AC the DC supply would have to be higher. A 24V RMS sine wave has a peak-to-peak amplitude of about 68V, so that's the DC rail voltage needed if the rotors have to be driven in a single-ended fashion. This is the case for the rotors I've seen as they usually have three wires, one being a common to both windings. If each winding had separate wires, you could drive each one with an H-bridge and use a 34V DC supply. Either way, a DC-DC converter could still be used to power the system from 12V or whatever.
Its major advantage is in continuously tracking at a low and variable speed without constant starting and stopping. But this design could easily go faster than the nominal 50 or 60 Hz speed if the DC rail voltage is increased proportionately. (The voltage and frequency in a VVVF AC motor drive are varied together to push constant current through the inductive reactance of the motor windings and provide uniform torque at all speeds.) But I wouldn't say *how* fast until I built and tested it.
Also dunno why you would need two IMUs. A single one on the antenna itself would tell you which way "down" is and which way "north" was. Platform acceleration (not mere motion) might be a problem but I'd have to think about how to compensate for it. Other than that, the only thing I'm concerned about is RFI from the transmitter getting into the sensor. You could simply not read it when transmitting.
I've also been thinking of using one of these IMU sensors for automating the setup of my Meade LX-200 telescope -- which I also want to use for satellite tracking. Not having to center that damn bubble level and find north would be nice. The IMU would probably be good enough to bring a pair of bright alignment stars into view so I can fine-tune the orientation. (I haven't checked but I wouldn't be surprised if scopes are already available that do all this. But I want to see if I can use my existing scope.)
A GPS will still be almost mandatory for both satellite antennas and telescopes for accurate time and location. This is needed not only for the pointing calculations but also to look up magnetic declination and inclination to interpret the magnetometer data. Then the magnetometer and accelerometer together give you a 3-axis orientation in space without calibration, assuming you don't have anything nearby to distort the earth's magnetic field.
--Phil, KA9Q
On 03/08/2013 01:46 PM, Phil Karn wrote:
On 03/08/2013 12:08 AM, Gus wrote:
Still, it sure sounds interesting! What do you think it would cost to put one together?
Dunno. I'd have to build one.
Willing to have a go at it? I'd contribute towards parts for a prototype...
Also dunno why you would need two IMUs.
Because I foolishly thought to compare data from the antenna and the base, to get pointing angles. Only after posting did I realize that one IMU would give antenna position data in the earth frame of reference (not the vehicle frame of reference).
Platform acceleration (not mere motion) might be a problem but I'd have to think about how to compensate for it.
Are you familiar with the UAV Dev Board? They do all manner of clever tricks and don't even have a magnetometer!
Other than that, the only thing I'm concerned about is RFI from the transmitter getting into the sensor. You could simply not read it when transmitting.
Won't the IMU work in a Faraday cage? Yes, but power has to get in and sensor data has to get out, so RF will still be a problem. What about auto-sensing the RF and delaying the output from the IMU or telling the CPU not to read them? Could be a problem for big-mouthed rag-chewers like myself. Also, in a Field Day type environment with several nearby transmitters operating, your tracker could be offline for an entire pass. Of course! Fibre optic control cable! Obvious, isn't it? :-)
A GPS will still be almost mandatory for both satellite antennas and telescopes for accurate time and location. This is needed not only for the pointing calculations but also to look up magnetic declination and inclination to interpret the magnetometer data. Then the magnetometer and accelerometer together give you a 3-axis orientation in space without calibration, assuming you don't have anything nearby to distort the earth's magnetic field.
I've got a couple uBlox 5's around here somewhere...
participants (4)
-
Bob Martinson -
Gus -
Phil Karn -
Robert C. Campbell