Resolver to Digital Conversion

[dau]

Hello all,
we use an AD2S90 RD-Converter to measure the position of a synchron machine. Everything works fine but:
our circuit is integrated in a more complex system which switches off the power supply to our device under a fault condition. At power up the AD2S90 looses its tracking because the motor is still turning and ( presumably?) the admissible angle acceleration is too high. Because the IC has no reset input we are a little bit puzzled what to do now. It is a fly-wheel application so we must start the system under high speed.
Any suggestions?

 

[VEBill]

Add a UPS?

 

[VEBill]

Also, if it is a fast-turning flywheel, should something other than a synchro have been used? Fast RPM systems can use faster/cheaper/better ARP/ACP sensors (one ref pulse and X clock pulses per revolution).

And if it is a flywheel, you might even be able to dispense with the ACP (provided it doesn't change RPM quickly).

 

[sreid]

I would think that the R/D converter would lock up even with a turning motor. Have you talked to Analog Devices Tech Support (for these devices you need to get to the right Guru)? Forgive me in advance; you are using the absolute serial output.

 

[sreid]

An old (and perhaps largely forgotten) way to do R/D conversion is as follows; you use the R/D "Transformer" in reverse.

The sine and cosine "outputs" are used as inputs. These are driven by two signals, a sine wave and a cosine wave. What is normally the excitation input becomes the output signal. The output signal is a constant amplitude "Sine Wave" which is phase shifted relative to the sine and cosine excitation signals. The phase shift is directly proportional to the shaft position.

The implementation was done as follows (This could now be done with one small microprocessor).

First a continously running binary counter (say 12 bits) that is clocked so that the frequency of the 1024 counts is equal to the resolver excitation frequency desired. The counter is connected to an EROM which converts the binary count to digital sin and cos amplitude values. Thes values are fed to D to A converters and buffer amplifiers to drive the Resolver inputs. The output signal goes to a zero crossing detector (comparator). This signal is used to capture the binary counter count at zero crossing. This is the shaft binary phase position.

 

[dau]

@ VE1BLL
We've already considered that. It involves a lot of changes and scrapping of our cards. We thought of a battery only for the AD2S90.
The synchron machine is used by our client.
What's an ARP/ACP sensor ?
We need the actual angle position to drive the synchron machine.

@ sreid
Yes we use the absolute serial output. AD gives not much info apart from don't apply signals before VDD is up and so. We've delayed all possible signal inputs but the acceleration at switching them to the AD2S90 appears to be too high.
The reverse drive of the resolver is interesting but the whole system is already in operation and we need a quick and possibly simple solution.

A pin compatible IC which can withstand this situation would be optimal.

Regards

 

[VEBill]

Azimuth Reference Pulse and Azimuth Clock Pulse. It is a very simple type of encoder that produces one reference pulse at the reference position and a series of clock pulses (often 1024 or 4096) all the way around. Since it is counter based, the latancy is very low (often critical for high RPM systems).

ARP/ACP downsides: At power-up, it wouldn't know where it was until the rotator passes the reference position once. The simplest version cannot detect direction of rotation, but versions with I/Q clock outputs are common.

 

[sreid]

What's the RPM of the flywheel?

 

[dau]

@ VE1BLL
I'll look at these ACP/ARP devices. Seem interesting, only that the actual position at power up is critical for motor excitation.
@sreid
up to 25000/min
the AD2S90 can handle in our circuit/layout up to approx. 15000/min before clipping.

Has anyone worked with the AD2S1200? It has a reset input and a fault output activated at an agle difference between input and output greater than 5 degrees. I'm not quite sure if this reset really resets the integrator "overflow" and not the whole "IC" so that a new start would only delay the procedure and the IC sees the original high acceleration every time resulting in a kind of hick-up mode.

 

[sreid]

I have only worked with the AD2S83 but the conversion process is the same as the AD2S90/AD2S1200. My guess is that the ability of the loop to lock up on the fly is primarilly determined by the loop bandwith. This seem to be around 1 kHz for both the AD2S90 and the AD2S1200 so my guess is that the lock up performance would be the same. Only the "Right Person" at AD could answer that.

Fifteen years ago it was John Christacopoulos in England at 44-1932-266031 or

44-1932-266000 (Front Desk).

 

[dau]

All-clear signal,

we've rather accidentally solved our RDC problem. An old RDC-card with no buffer amplifier at the velocity output worked up to 30000/min without any additional tricks. It puzzled us tremendously because the layout had hardly changed.
Our new design buffered the velocity output via a non-inverting amplifier with a very high input-was our assumption. Unfortunately we added a 10k/100n low pass filter before the buffer. The AD2S90 can drive only 250µA and so the fault at power-up at high speed occured.
As experience shows: 99,9% of all faults is because of human insufficiencies.But if one is at the end of ideas one is tempted to blame the highly matured technology.
Thank you for all contributions.
Regards

 

[sreid]

One would have hoped that AD would have buffered the Velocity signal so that external loading would not have messed up what is in reality the loop tracking signal.

 

[dau]

Yes, I was pretty surprised to find such a low drive capability. From my feeling about this size of IC I had assumed at least 5 mA. But there are today so sophisticated ICs with so many (admittedly) important specifications. Then added the complexity of the circuit/system and it's sometimes
amazing that no limits are violated and the whole device works as planned.