B-52 Bomber Astro Tracker - Part 2: Power up and gyro-stabilization
Hello and welcome back. If you follow the channel, you know that we opened up a Kollsman Astrotracker, with the welcome help from our friend Michel, from the “Le Labo de Michel” YouTube channel. This astrotracker is but only one part of the complex Kollsman MD-1 astrocompass, that was flown on B-52 bombers. It was a Cold War era celestial navigation system that used analog mechanical computers. In Part 1, I explained how celestial navigation works, by sighting the altitude and azimuth of the sun, moon or stars, using a sextant, then doing some complicated calculations to correct an approximate assumed position into a precise true position.
The star tracker we have is an automated sextant, which is able to sight, center, and maintain lock on a star while the bomber is flying, providing an automated and continuous true heading to the navigation system, and also a position error estimate. When we cracked it open in the last episode, we were greeted by a complex arrangement of analog, electro-mechanical components. The regular viewers of the channel will have noted the similarity to the Bendix air data computer that we previously restored. The many cylindrical thingies that you see all over the place, are synchros, motors, generators, and control transformers. They are the key electrical components of these analog computers. We explained them in our previous episodes on the Air Data Computer,
link in the doodley-doo, and we’ll come back to them in due time. But this astro-tracker goes one step further: it adds optics to the mix. And sure enough, the centerpiece of our freshly opened astro-tracker is the optical telescope. [Marc] Ah, oui, c''est un prisme. It's a prism. So, looks from the side here. So the prism is probably all the way down. It should be able to move up. Oh, okay. It does spin freely.
[Michel] Normally it's possible to have a complete turn. [Marc] It's moving when I am turning it. Do you see that? It's going up, and then it's going down. Oh, that's interesting. Our astrotracker telescope does not look like a telescope at all. Fortunately, Kollsman patent 2,998,529 gives us some insight into how it might work. The telescope optics are actually pointed straight up, mostly hidden from view in the middle of a cylinder. What we just saw was the movable prism on top of the telescope optics,
item 71 in the patent drawing. This prism functions as a mirror, forming a periscope which allows the telescope to point to any part of the sky, looking through the small curved transparent dome, which is the only part of the system that peeks through the wing. The prism rotates around for azimuth sighting, and tilts up and down for altitude sighting. There are two motor-generators, called MG in the drawing, for controlling the prism position. Item 77 is the altitude motor, and item 81 is the azimuth motor.
In addition, each motor is associated with two Control Transformers, marked CT in the drawing. These are the position sensors. There is a x1 CT, the coarse position sensor, and a x25 CT, the fine position sensor with 25 times finer resolution. Via internal reflection, the prism directs the light of the small patch of sky it’s pointing at into the telescope optics. If our pointing computer did its job correctly - remember, that’s the one that Dave teared down on the EEVblog - this small patch of sky should contain the star that we want to track. The star will not be perfectly centered, but it should be close enough to be in the narrow field of view.
The light from that patch of sky is then focused onto detector 110, which is a Photomultiplier tube, or PMT. PMTs are ultra-sensitive light detectors that we also had previously encountered, in our episode about the Eskalab spectrophotometer. You might remember, from that episode, that the PMT is not an image sensor at all. It is just a light level detector, a very sensitive one at that, but it provides no image information. By itself, it cannot tell where the star is in the image. It can only measure the total amount of light in the patch of sky, just like it measured the total amount of light going through our spectrometer sample.
Obviously, so more trickery is needed to derive the star position. That’s when the rotating disks 106 and 107 come into play. We could spy both of them after we removed the PMT. One disk has radial slots, and is called the raster. The other one has a round aperture and is called the shutter. They rotate continuously at precise speeds. It’s going to take a whole video to explain how this devilish analog star position encoding scheme works, but for now, just know it’s there, and that it needs a motor M to rotate the disks, and an associated position sensor P.
So this is what our star tracker really is, if we ignore the many motors and sensors that automate it. A fixed telescope with a movable prism on top, a rotating shutter and raster disks to encode the position of the star in the field of view, a few filters for viewing the sun or masking out auroras, and a photomultiplier. And lots of motors and control transformers. The various motors and control transformers are the cylindrical thingies that you see at the base of the telescope, surrounding the PMT. So far, we have identified 2 Motor Generators, 4 Control
Transformers, one Motor and one Position sensor. That should be 8 cylindrical thingies. But you can count 10 of them here, so there are two more - a sure hint that the actual implementation is more complicated than what’s disclosed in the patent. We’ll get to discover what they are in due time. But we are not quite done. There is additional stuff at the bottom. We can see two more Motor
Generators, and their associated Control Transformers. It turns out that the entire telescope platform is mounted on a two axis gimbal mount. This is used to keep the telescope platform level while the plane is flying, using a gyroscope. This system will keep the telescope perfectly vertical through +/- 15 degrees of movement in pitch and roll. We already demonstrated the movement of the gimbal platform in the previous episode. For that, we first had to release the gimbal’s electro-magnetic brake, and then move the platform manually so we could access a hidden screw. You can spy the gimbal motors and control transformers.
So we’re now talking about a grand total of 4 motor generators, 8 control transformers, 1 motor, 1 position sensor and 2 mysterious extras. That’s 16 cylindrical thingies, total. Each of them has many wires: a control transformer has 4 coils and 5 wires. As we’ll see later, our motors are more than simple AC servo motors, they are motor-generator combos, for a total of 4 coils and 8 wires. And there is a brake, a few solenoids, the PMT control, and extra electrical stuff not talked about in the patent. That’s a lot of wires, none of which we know how it is hooked up. Undaunted, Michel got right to work on reverse engineering the wiring. I soon learned that Michel is a reverse engineering beast. He must be Master Ken’s
forgotten French cousin. For days on end, he would be in the lab from 7 am to midnight, beeping connections tirelessly, which, by the way, you can only do by piercing the wires, as there is no other place to probe them. He then carefully wrote his findings in his notebook. After a few days of this, he had reverse engineered the entire wiring. Thanks to Michel’s drawings, we can start to make sense of what we see at the back of the telescope platform.
There is the photomultiplier tube of course. The first cylinder is the shutter position sensor, which is implemented with a permanent magnet generator. Then we have the shutter synchronous motor, which should be driven at 250 Hz to rotate the shutter disks at a precise speed.
Then we have the prism altitude angle coarse position sensor, implemented by a control transformer. Here is our first mystery item. It happens to be a potentiometer which serves as another altitude angle sensor for the star position detection electronics. Then comes the prism altitude motor generator.
Then the fine control transformer for the prism altitude angle. It is followed by our second mystery item, which this time is a resolver, an animal that we have already encountered. This time, it is coupled to the azimuth angle. We’ll see later that it is used as a coordinate rotator for the star position detection electronics. Then we have the coarse control transformer for the prism azimuth angle, the corresponding motor generator, and the fine control transformer for that axis. There you go. This complicated contraption is starting to make
sense, assuming you are still with me. I think we are ready to start experimenting with the star tracker. But before we do that, we need to build a test stand. [Marc] All right, so let's try to put in a mount, see if we can take it from there. So, we've made a couple of rails here.
I think that will... Oops, you are already hooked up. That will work. [Michel] Perfect! [Marc] There we go, we have it. So I can either clamp it, or put some screws. I think clamp it. Then you can put your gyro at the bottom, power supply, stuff like that. We can adjust the height. There you go. Test stand!
The first step is to get the gyro stabilization working, so we can erect the telescope and maintain it leveled. The gyro and the servo electronics are normally contained in a small cylinder next to the tracker. But we don’t have that. So, instead, Michel brought his own vertical reference gyro, and some AC servo amplifiers to build the electronic servo control loop ourselves. [Marc] We're ready to try something? [Michel] We are ready! [Marc] What have you done? [Michel] I have added the servo amplifier here, and the synchro transmitter to simulate the gyroscope output. Normally, we can adjust the pitch stabilization using this synchro transmitter.
[Marc] So you're trying to control one of the axis of the platform with this. [Michel] Normally, yes. The pitch is the inner gimbal. This is this gimbal here, that position here. [Marc] And it's complicated because you need 115 volt AC, 26 volt AC, and 28 volt DC, a synchro, an AC amplifier, in a control loop. That's a servo loop, right? [Michel] This is a servo loop.
[Marc] The servo loop is not inside this. [Michel] It is outside. [Marc] It's a part of the electronics we don't have, right? [Michel] Unfortunately, we don't have everything. We just have the mechanical part. So we need (to build) all (the) electronics. [Marc] But you brought one, that you know works.
Okay, let's try it! Let me interrupt, yet again, to explain how a synchro servo loop works. A review of our previous episode on synchros will probably help, I’ll put a link in the doodley-doo. As a summary, a transmitting synchro is somewhat similar to an AC motor used in reverse. It has 3 windings on the stator at 120 degrees of each other, and one winding on the rotor.
Normally, you power the rotor with 400 Hz AC, and an AC signal is produced on the three stator windings, that encodes the position of the shaft. To read that position, the simplest way is to put another synchro in parallel, called the receiving synchro. By virtue of magnetic magic, the second synchro will align itself on the first, as we demonstrated in the synchro video. But that only works if the receiving synchro is moving something light, like the needle of an instrument. Here, we want to use a transmitting synchro to move a heavy telescope gimbal. So, we replace the receiving side with a servo motor loop. The receiving synchro is replaced with
a Control Transformer. A CT, as it’s called, is the same as a synchro, but with the rotor winding off by 90 degrees. This time, its rotor winding is not powered. It is used as an output instead. If the CT is aligned to the input synchro, no voltage appears at the output. But if the two are offset, an AC signal appears,
that grows larger with the misalignment. It is called the error signal. We then feed this error signal to a simple AC power amplifier, and then connect the amplifier output to the control winding of an AC servo motor. To complete the loop, the output shaft of the servo motor is mechanically connected to the CT, usually through reduction gears. So the motor will turn the CT shaft until it is aligned with the transmit synchro, and the error signal is back to zero.
This arrangement is very powerful. You can move entire gun turrets with it. In our case, the control transformer and motors are already provided in the astrotracker. All we need is to add a transmitting synchro and an amplifier, which Michel has just done. Let’s see how that works. [Marc] We need the 28 volt first, right? [Michel] Yes. [Marc] So, 'output'. Okay.
And the noise we heard were the thing being freed, right? Yeah, it's free up. So I have this little super extra control box here. And I can lock it, or free it. Okay, so, that works. It's free, it has been freed up. And now, uh, maybe you control it, and I turn it on and off pretty quickly if something goes bad. [Michel] Yes, it's a good idea. [Marc] So, you're ready? [Michel] I'm ready.
[Marc] 'Enable'. Oh, it works! [Michel] Okay, it is expected because I didn't connect the generator. So the loop isn't stable, probably. [Marc] Oh, can I take picture of that? Well, you you're doing something. The average position is... Well, that's success if you ask me. That's a lot of wires, just to do that. Because you have the five wires of the synchros,
the control transformer, you had to find which capacitor to put for the motor to work correctly. So, you want to try to put the, um... [Michel] ...the generator? [Marc] ...the generator. [Michel] In order to stabilize the loop? [Marc] Yeah. [Michel] Normally, it should work.
[Marc] Alright, well done Michel! The old astro compass is coming back to life, but it is overshooting the position and being unstable. This was a known problem with direct servo loops. But fear not, there was also a known solution! We add a new device, called a tachometer generator. That’s an AC motor in reverse, essentially, that generates an AC voltage proportional to the speed of the motor. It is subtracted to the motor control voltage at the input of the amplifier, and should prevent the overshoot. This arrangement was so common, that the generator is an integral part of our motor, hence called a Motor Generator, or MG in the drawings. Here is one that Master Ken took apart and explained on his blog, link in the doodley-doo, and you can see the two rotors, the squirrel cage one for the motor section, and the copper one for the generator.
At this point, Michel and I had given up on English, so let me self-dub my own video. Warning, I am a pretty poor voice actor. [Marc] So you added the rate transducer? [Michel] Yes, somewhat at random.
[Marc] So, maybe you still need to lower the gain. There's some progress at least. [Michel] Yes, now it's dampening. Maybe I don't have enough rate feedback. Or it's wired in reverse? Oh yes, it's dampening out. (Moments later) [Marc] So, you found out what wasn't working? [Michel] Yes. The problem was that the phase of the motor was inverted, the reference coil. Oh no, the other one. I inverted the phase of the control coil.
[Marc] So, it was an anti-servo loop? [Michel] Exactly. But somehow it still managed to stabilize itself. But now it stabilizes without any overshoot. [Marc] Nice! [Michel] So, the next step is to do the same thing on the other axis, the roll axis. Then we'll connect everything to the gyro underneath. [Marc] See if we can stabilize the platform. And then it's onto the optical part.
Alright, we have our roll servo loop working! We just had the feedback wired in reverse, but once Michel rectified that, it became stable. We just need to do this again, in the right direction hopefully, on the pitch axis. [Marc] Okay, so we have one more amplifier? [Michel] We have one more amplifier, here, and one more synchro transmitter. And some more wiring. [Marc] Okay, but we don't know in which direction it's going to turn.
[Michel] Yes. It could oscillate. We don't know in which direction. [Marc] All right. It either smokes, or it doesn't. Looks like it didn't move. [Michel] Nope. The other one still works.
(More moments later...) [Marc] You found the issue? [Michel] Yes, the issue was the de-phasing capacitor, which is necessary for an AC motor. You need 90° between the control and the reference coils. It's done with a series capacitor. [Marc] Ah, the de-phasing capacitor. We hadn’t talked about that one and left
it as a mystery on the schematics until it decided to bite us in the butt. See, in our dual coil AC servo motor, we need to create a rotating magnetic field. For that, you need one of the coils to be out of phase by about 90 degrees. This is usually
done by adding a capacitor in series with one of the coils. The capacitor is usually adjusted to give about 90 degrees of phase lag. You could calculate it, but in our case we just used trial and error. Looks like we ended up with error. We need more trial.
[Michel] So, I added one, a bit of a blind guess, because the motor is much larger for the roll. There is a much larger moving mass. So, I had put a 0.39 microfarad cap, which is not enough. So if I put one, a little bigger, I'll add two, that will be about 1.2 microfarads. Now I can make it move, but it's not yet optimal. [Marc] But it's moving, it's moving! [Michel] Yes, it's servoed, nonetheless.
[Marc] You were telling me that the servo loop was a bit saturated? [Michel] Yeah. So, if you look at the scope, the output signal of the amp is the yellow trace. It's square, it's saturated. The signal after the cap is the green trace. So, if we add a capacitor, it's going to look a little bit better. But it's still not optimum.
However, it's working somewhat. [Marc] We are so lucky to have you here. Because, this de-phasing cap business, we might have found it eventually, but it could have taken us forever! Not bad, it works better now! [Michel] I can move it in two directions simultaneously. [Marc] Okay, we're getting closer! [Michel] Yep, getting closer! [Marc] You said, what surprised you, is that there are coils at 115 volts, and coils at 26 volts? [Michel] Yes. The motor reference coils are at 115 volts, but the tach generators are at 26 volts. That is really strange, two reference voltages.
[Marc] So, motor reference coils are 115 volts? [Michel] Yes, the motor reference coils are 115 volts, and the control coils are standard 26 volt ones. But for the tach generators, the reference coil is 26 volts. [Marc] I'm impressed that you were able to figure this out! [Michel] Yes, servo control loops are my specialty! [Marc] Great, now we have two axes. [Michel] Now we'll be able to hook up the gyroscope. [Marc] That's all there is to do! [Michel] I'll just prepare it and wire it up. [Marc] Alright, we have two axis servo control of our gimbal. In the real system,
the roll and pitch input synchros come from a vertical gyro assembly, so the telescope should be servoed to the gyro movements. It’s finally time to spin up Michel’s beautiful gyro. So, Michel, you brought another item? [Michel] Right. This is a vertical gyro, to provide a vertical reference. [Marc] So, the idea is that the vertical gyroscope, well, is always vertical, regardless of the aircraft attitude. And it moves the platform accordingly. That's a pretty one! [Michel] So here's a synchro. There is a synchro on each side. [Marc] Can you spin it up for the camera? [Michel] You want to spin it, to power it up? [Marc] Yes, to see it erect. [Michel] Yes I can, I just need to connect it to 115 volts.
[Marc] And a one and a two, here we go! [Michel] At first it wiggles all around... [Marc] It makes very little noise compared to mine. This is a good one. (sound of gyro spinning up) [Michel] It's at operating speed. [Marc] We can move it a bit? It's magic! So, the idea is to connect this to this. So, when we wiggle one, the other one wiggles too. [Michel] It will replace those two synchros here. [Marc] Okay! Good work! [Michel] We're making progress.
[Marc] I saw it move! Wait, let me do it, so you're not in front of the camera. In which direction? [Michel] Try pitch. [Marc] Oh, yeah! Pitch, around this direction. Impressive, very impressive! Even better, with two axes! [Michel] So the roll and the pitch [Marc] OK, Michel rules again, our gyro is controlling the astrotracker gimbal. In the real system though, , the gyro should be attached to the same frame as the tracker. So let’s move the gyro on the astrotracker stand.
(Soon after...) [Marc] Go for it. Did it start or not? What's happening? It's connected, it's fine. Do we have a short circuit somewhere? Wires that touch? Yeah! They touch there. Is it on purpose? No, that's not good here. That's not good.
[Michel] Ah yes, that's not good at all! Not done on purpose. The 115 volts was shorted out! [Marc] Okay, works better when there are no shorts! [Michel] It should work already now. [Marc] Ah yes, absolutely! In the other direction, it's harder. Yeah, yeah, it works! Yay, thanks to Michel, we have a gyro stabilized telescope platform, . Pretty cool! In the next episode, we’ll turn our attention to the star tracking optics. See you then!
2025-01-08 21:33
