# WWII servo replacement

Discussion in 'The Projects Forum' started by jonbarril, Feb 18, 2011.

1. ### jonbarril Thread Starter New Member

Feb 18, 2011
9
0
I am a volunteer at an aviation museum. I am trying to develop a solid state replacement for a WWII gun turret servo. I am trying to use the existing sensors (selsyns), but replace the servo (tubes and transformers) and motor controller (amplidyne).

The system supply is 115VAC, 400Hz. The selsyns produce 57.5VAC, 400HZ, that varies in amplitude and phase. Amplitude indicates the angular error magnitude, and phase (0 or 180) indicates the error direction. The servo monitored the selsyn output, and generated a +/- voltage difference between the inphase and out of phase sides of the servo amp. This +/- difference fed a bidirectional motor controller.

I figure I can use the same basic approach, with two sides to the solid state servo amp -- an in phase and an out of phase side. So, what I need is a circuit to handle the fixed amplitude 115VAC supply/phase reference, and the 57.5VAC variable amplitude/phase input signal.

I was thinking that some form of FET source follower for each side might be a good start for this. Each side would be supplied by one output from a bridge-like rectifier (half-wave in-phase and out of phase), and the input signal to both sides would be half rectified (half-wave varable amplitude and phase). Each side would then output a voltage proportional to the input signal and its phase (0 or 180). The rectified input signal voltage could be easily reduced using a voltage divider (e.g. 10:1 for 0-6V at the gate).

What do I do about the 115VAC supply voltage? What is a good way to reduce that to something that a simple FET could handle (e.g. 0-16V at the drain)?

What size source resistor is needed? (The voltage difference sensor between the two sides is essentially a high impedance opamp, whose output drives a bidirectional motor controller -- 0 for full reverse, +5 for full forward, and 2.5V for no motion.)

What transistor number should I use? (Precision is not required -- it's just for demonstration, but the environment could have temp extremes, especially on the high side -- closed airplane sitting in the sun.)

Any help would be appreciated.

--jon

2. ### jpanhalt AAC Fanatic!

Jan 18, 2008
5,699
912
I miss the point of doing this redesign for a museum project. A re-build/restoration would be more impressive. I hope the original is not destroyed in the process.

As for your design, can you post schematic? It would be very difficult to give advice about individual components without one.

John

3. ### jonbarril Thread Starter New Member

Feb 18, 2011
9
0
Servos amps are in short supply, especially ones that are operational. And, they are designed to work with an amplidyne as the motor controller. Amplidynes are even rarer and are hard to maintain. Nothing is being destroyed. The goal is to make the gun turrets on the last flying B29 operational (for demo purposes) by connecting the existing equipment to a solid state replacement for the servo/controller.

As for a schematic, I've pretty much boiled the problem down into inputs and outputs at the optimal points in the original system. What I am mainly after is a magnitude/phase comparator that can handle the high voltages.

Here is a crude schematic of what I had in mind.....

supply: 115VAC RMS @400Hz, 0 degree phase
signal: 0-57.5VAC RMS @ 400 Hz, 0/180 degree phase
output: 0-5VDC

D> is a diode
??? reduces voltage

115VAC
floating
supply
| |
gnd--D>-+ +-<D--gnd
| |
half rect +-<D-+ +-D>-+ half rect
0 phase | | 180 phase
| |
??? ???
| |
half rect +--- | ------------ | ----+
0/180 phase | | | |
| |-+ +-| |
signal --D>--Rg1---+->| |<--+
0-57.5VAC | |-+--- output ---+-|
| | |
Rg2 Rs Rs
| | |
gnd | |
+-------+------+
|
gnd

Last edited by a moderator: Feb 18, 2011
4. ### SgtWookie Expert

Jul 17, 2007
22,194
1,764
Just fixing your schematic using the CODE blocks

Is your bird "Fifi" or "Doc"?

5. ### jonbarril Thread Starter New Member

Feb 18, 2011
9
0
The plane is Fifi. I am working on this project thrice removed from the folks who operate the plane.

--jon

6. ### SgtWookie Expert

Jul 17, 2007
22,194
1,764
Very cool. I saw Fifi at a fly-in a number of years back. She's a beauty.

Does Fifi have 3-phase 400Hz 115v, or is it simply single phase? (3-phase, 1-phase)

And, she has 28VDC available too, right?

Your 0v-5v signal - how much current will be required from this output?

I'm just trying to get some basics down here; what you have to work with.

7. ### jpanhalt AAC Fanatic!

Jan 18, 2008
5,699
912
That is a beautiful airplane. My dad was with the 19th Bombardment Group in Guam (B29's). Years ago, I saw a flight of 3 take off from the Aberdeen, MD military field. I was low-level along I-95 right at the edge of the restricted zone. What impressed me most was their gracefulness. I don't know if it was their last flight, but I consider myself lucky to have seen it.

Sorry, I misunderstood the purpose. I thought it was for a static display in a museum.

John

8. ### nsaspook AAC Fanatic!

Aug 27, 2009
3,634
3,760
I think you need a Scott-T transformer to isolate and convert the signals first. Then a simple resistor network can reduce the levels to what you want.

Some links about signal conversions and what you might be dealing with:
http://magnecompinc.com/scottt.htm

You might have to register to see these.

http://www.analog.com/static/imported-files/application_notes/50205505838923AN252.pdf

9. ### jonbarril Thread Starter New Member

Feb 18, 2011
9
0
The main supply is 28VDC. There is a dynamotor that converts it to 115VAC, 400Hz, single phase. (The only three phase usage I know of is for the Sperry bombsight and autopilot -- but that is a different project.) The motor controller input is low current -- essentially an op amp measuring the voltage difference between the in-phase (0 deg) and out-of-phase (180deg) sides as shown in the schematic.

A transformer could be used but I was hoping that the voltages are low enough to be handled by voltage dividers and FETs. I think a triode would work, but I was hoping for something a bit more modern and easier to prototype.

--jon

Aug 27, 2009
3,634
3,760
11. ### nsaspook AAC Fanatic!

Aug 27, 2009
3,634
3,760
I think it finally clicked in my head where you are at in the servo train.

Your signal is the two phase output from the CX servo on the motor shaft that generates the diff signal from the CT servo position vs CX servo position and is then input to the servo amp.

At FIG 1.12a correct?
http://www.ddc-web.com/documents/synhdbk.pdf

If so a simple resistance voltage divider should work to reduce the levels.

Last edited: Feb 18, 2011
12. ### jonbarril Thread Starter New Member

Feb 18, 2011
9
0
These are selsyns, which I'm guessing or sort of like higher voltage synchros. I am trying to make this as simple as possible. The selsyns are geared to the sight and the turret (separate on a B29), and produce a single error signal referenced to ground. As I described, the signal varies in magnitude, and in phase, with phase indicating the direction of the error (i.e. positive or negative).

All of this original stuff seems to be reliable and low maintenance. The big problem is taking that signal and feeding it to a modern motoro controller, that wants 0-5VDC, with 2.5VDC being no motion.

Resistor dividers would be OK for the signal, as I have shown because of the FET high input Z, and might be OK for the supply, but I still need a FET or something to do the magnitude/phase comparison (the original circuit did it wioth transformers and tubes in a similar manner). So, since my skills in these areas are limited, I was hoping somebody on the forum could come up with a simple design to deal with the voltage reduction of the supply to the source follower -- I need voltage reduction. not gain, and virtually no power. I also need a value for the source resistor -- I googled a bit, there is lots of theaory, but little in the way of cookbook stuff (which is my level).

13. ### jonbarril Thread Starter New Member

Feb 18, 2011
9
0
I took a closer look at the fig and text. This is the correct location in the servo loop. The big difference I see is that my error signal varies in magnitude and phase, not magnitude and sign. This is why what I need is more of a phase comparator -- compare the signal phase (0 or 180, it is not continuously variable) with the reference phase, provided by the supply.

Apr 20, 2004
15,815
290
15. ### jonbarril Thread Starter New Member

Feb 18, 2011
9
0
The pedestal sight and the computer are on that sight.

For more detail check out http://www.twinbeech.com/CFCsystem.htm. This is the testbed that I have been using. Note: For this project the computer is not in the servo loop (i.e. for demo only, not for accurate target tracking).

16. ### jonbarril Thread Starter New Member

Feb 18, 2011
9
0

Thanks, but what you are referring to is the next generation or two. Selsyns are high voltage, synchros are low voltage (from what I understand).

17. ### nsaspook AAC Fanatic!

Aug 27, 2009
3,634
3,760
What you are looking for is a servo demodulator.

Old school design. You could feed the output into a "Differential Amplifier" to generate your drive signal.

http://www.tpub.com/neets/book15/62f.htm

400hz transformers
http://www.hammondmfg.com/226.htm

Like "beenthere" I was also on a ship (LPH-3) but in the radio and radar dept when lots of things ran on tubes and servos.

On this link the tubes in the circuit are part of the servo demod/phase detector with the relay coils contacts acting as a H-bridge. Sweet!

Last edited: Feb 18, 2011
18. ### beenthere Retired Moderator

Apr 20, 2004
15,815
290
Most ran at 120 VAC.

19. ### jonbarril Thread Starter New Member

Feb 18, 2011
9
0
....then I guess they were selsyns, not synchos.

The diagram by nsaspook pretty much sums it up. In this case I want to use a SS motor controller.

So, any ideas on a circuit that will do it with a FET instead of a triode, and with no transformers? I think mine is close, but I'm not good enough at circuits to solve for the values or to come up with the correct voltage reducer for the "???" location in my schematic.

Apr 20, 2004
15,815
290