Sperry ASI with resolvers

[arnolde]

Hi there,
I have a Sperry ASI (part no. 2594621-918) here and had a look inside, it has 4 or 5 DC motors inside, and just as many Clifton Differential Resolvers (400 Hz / 26V / 2 phase) and some transformers. I understand several people have succeeded in replacing the resolvers with pots and created a servodrive that way. I was thinking it might be easier/cheaper (once the code is written) to use the resolver, by programming a microcontroller to detect the position via A/D or even just the zero crossings (with 10-microscond accuracy)? The frequency fed in could even be less than 400Hz, depending on the tolerance of the resolver. I would hope to achieve a low-cost solution to drive the instrument without the need to change any mechanical parts, in my case I could even remove the transformers and have about 10x10x6cm space to integrate the circuit inside the instrument housing.

Has anyone tried decoding a resolver yet? There are custom IC's just for that purpose but I haven found any under 200 EUR yet.

[Mike.Powell]

I haven't heard of anyone going this direction, but in general terms it should be possible to use a microcontroller to measure the resolver output amplitudes. The zero-crossing idea would require some additional analog circuitry to make it useful as the resolver outputs are both in phase with the input. You could shift one output 90 degrees then add it to the other to produce a composite signal that you could make the phase measurement on.

Or... Why not just build an analog driver that powers the existing motors and uses the resolver outputs as position sensors in feedback loops? The biggest issues would likely be finding documentation for the ASI and the matching connectors for it.

Sounds like an interesting project.

[arnolde]

Luckily the documentation issue is on my side, since I have the complete flightsim documentation here (about 4 pallets of boxes) which should (havnt actually picked out the right folder yet) include the pinouts of the connectors on the instruments (all of them :-) ) and schematics of the input signals (=how the simulator operates the instrument via the standard plugs). Maybe even schematics of the instruments inside, I'll have to look. Incidentially the full cockpit with all instruments is also for sale on ebay, I'll have to add the documentation issue into the description since it's probably also worth a lot.

I dont quite understand your point: The outputs are both in phase with the input of the resolver? I thought the outputs would be 90° shifted because the stator coils are physically 90° shifted in the resolver? I was thinking I can just put a sinewave on the input and watch both outputs and by comparing them I can detect the absolute position? Of course I was going to drive the existing DC motors directly.

Making an analog feedback driver sounds interesting, what kind of input would that require? Sounds like thats whats already in there, with the transformers?

[Mike.Powell]

It is only the amplitudes of the outputs which are at 90 degrees to each other. They remain in time-phase with the signal applied to the rotor. The zero crossings don't move.

[arnolde]

Aha, so for example the outputs might look like this?
0°: A=input*x, B=0
90°: A=0, B=input*x
180°: A= -input*x, B=0
270°: A=0, B= -input*x
where x is a constant

EDIT: I just found another resolver-to-digital converter chip, the AD2S90. At 58$ each (digikey) still not exactly cheap. I wonder why they are so expensive?

EDIT2: And another one: http://www.analog.com/static/imported-files/data_sheets/AD2S1200.pdf only 24$, now thats more like it! Even includes a sine generator onchip.

[Mike.Powell]

Yes, something like that. However, there are a number of related devices. I would be surprised if your instrument used only one type throughout.

If you can find the instrument documentation, it may describe its operation. That would help in deciding which direction you want to take.

I think the converter chips are expensive because they are low-volume, laser-trimmed, precision products.

[arnolde]

5453
This ASI has 7 of these blue resolvers inside. Although to be fair, I havent checked they're all the same type, I just assumed.
I'll search for the docs tomorrow.

[arnolde]

Yay, found something interesting in the Sim docs.
Are those 3-phase AC RPM inputs or Synchro?
5455

[Mike.Powell]

Looks like synchro inputs to me. I think they may be using one of the resolver windings in as in a control transformer to provide position feedback for the motor control loop, and the other resolver winding for the monitor circuitry that powers the flag.

[arnolde]

I found and scanned several more instrument schematics. Maybe they're of interest to somebody. Getting the FD working completely would certainly be a challenge :-)

I can email the original sizes for easier reading if anyone is interested. Each jpg is about 1.5MB in size.

54625463546454655466

[arnolde]

54675468546954705471

[Marnus]

Hi arnolde

Im bussy decoding my Resolver(Clifton 2 phase 2 volt ) at this moment. It is fitted inside
my OBS/VOR, It is mounted on the OBS knob, Im still trying to find some data ,but no luck.

Do you have a sioc script for a resolver, and thru wich board will i interface this gadget..