Pico Systems Resolver Converter

[charliegolf]

I have been eying the Resolver Converter for weeks now. I have several Pacific Scientific motors that are resolver-only and several AMC drives that are not so resolver-friendly.

I might be interested in picking up one/several of these converters, but I want to make sure they'll work. Do you think that Pico System's resolver converter would work for AMC's DX15C08 connected to one of Pacific Scientific's size 21 frameless resolver?

[Al_The_Man]

I believe the resolver also acts as commutation on the original motor/drive, I have not used that model of AMC, but suspect they initially require the hall inputs if used in the AC servo mode, and of course will need it if used in the BLDC mode.
Al.

[jmelson]

Originally Posted by charliegolf I have been eying the Resolver Converter for weeks now. I have several Pacific Scientific motors that are resolver-only and several AMC drives that are not so resolver-friendly. I might be interested in picking up one/several of these converters, but I want to make sure they'll work. Do you think that Pico System's resolver converter would work for AMC's DX15C08 connected to one of Pacific Scientific's size 21 frameless resolver?

It looks like this servo drive needs industry-compatible "Hall" signals, but my converter (so far) does not produce them. It does have an absolute position output that could be decoded to provide the commutation signals, but I just haven't gotten around to making that board. I don't know if anyone else has made a converter that will do this. Probably a PIC microprocessor could be programmed to do the job pretty easily. The board produces 4096 discrete position values for a full revolution. Most brushless servo motors are 8 pole. So, you need to divide the absolute position into 24 steps that repeat through the 6 commutation codes. Of course, if your motors ALSO have the Hall outputs, you don't need to go through these gyrations, and the resolver converter will work as is.

Jon

[charliegolf]

Originally Posted by jmelson It looks like this servo drive needs industry-compatible "Hall" signals, but my converter (so far) does not produce them. It does have an absolute position output that could be decoded to provide the commutation signals, but I just haven't gotten around to making that board. I don't know if anyone else has made a converter that will do this. Probably a PIC microprocessor could be programmed to do the job pretty easily. The board produces 4096 discrete position values for a full revolution. Most brushless servo motors are 8 pole. So, you need to divide the absolute position into 24 steps that repeat through the 6 commutation codes. Of course, if your motors ALSO have the Hall outputs, you don't need to go through these gyrations, and the resolver converter will work as is.

Of the four or so Pacific Scientific motors I have, I don't think a single one of them has Hall outputs. I have found a PLC-type box from I2T (modicon) that claims to have a resolver/digital converter (datasheet). I haven't received it yet, but I'm hoping that it will work for my purpose. We'll see if I manage to let all it's magic smoke out, and I'll post the results.

As you know vast amounts more about this issue than I do, what does it take to convert an analog resolver signal to an encoder signal? The only reason I ask about going the encoder route is due to the fact that I have these drives laying around. Is there a better route to take? Most importantly, are there any servo drives that you know of that:
1. Accept resolver feedback.
2. Work for AC, brushless servos.
3. Accept Mach3 step/direction commands.
4. Output moderate power levels (15-25 amps).

Sorry for so many questions, but both you and Al are a vast warehouse of knowledge, and I need all the help I can get.

Thanks, Chris

[Al_The_Man]

Reliance Electric drives, which are owned now by Allen-Bradley (Rockwell) use resolver input fairly extensively.
The problem with mix and matching a resolver motor/drive combo is a bit harder than the average DC or DCBL as although another drive accepts resolver input, the resolver excitation signal has to conform.
If you find out or detect the pole count, I would tend to replace the resolver with a commutation encoder such as Renco are dumping on ebay right now.
Although this converts them to DCBL from AC sinusoidal, I still get good performance.
I have done this to a few motors, such as Fanuc servo's, I am just in the process of experimenting and setting one up to act as a spindle and lathe C axis.
Al.

[jmelson]

Originally Posted by charliegolf Of the four or so Pacific Scientific motors I have, I don't think a single one of them has Hall outputs. I have found a PLC-type box from I2T (modicon) that claims to have a resolver/digital converter (datasheet). I haven't received it yet, but I'm hoping that it will work for my purpose. We'll see if I manage to let all it's magic smoke out, and I'll post the results.

Yes, if the motor already has an absolute position sensor that the drive knows how to read, there is no reason for the Hall sensors.

As you know vast amounts more about this issue than I do, what does it take to convert an analog resolver signal to an encoder signal?

Well, quite a lot. At least to do it the Analog Devices way, you need to excite the resolver rotor with a clean sine wave. You need to sense the signals coming back from the two stator coils with balanced amplifiers, and then compare the amplitude and phase of these signals to sine and cosine values, and work the calculated rotor position back from that. Counting the rotor position up and down can also generate the quadrature count pulses.

But, that just gets you the quadrature pulses, which my board already does. Since the AD2S1200 chip also provides a parallel 12-bit position value, I put that out to a header location on the converter board. All we need to get to the Hall signals is to compare that digital number to ranges of values and then make the Hall signals from that code. I am doing this with another device I make, for Fanuc's proprietary encoders. I had planned to make up a little plug-in board for the resolver converter, but just haven't gotten around to it. You could hook the 12 absolute position outputs to a PIC chip, or a PLC to convert the position to Hall signals for your drive.

I don't have a good answer for your other questions. I think to use almost any commercial amp, you need those Hall signals. Oh, by the way, as far as I know, the AMC amps you mentioned do NOT take step and direction signals. (I think you could convert Step/Dir to quadrature and feed it to the aux encoder input, that's about the closest to step/dir for this drive.)

Jon

[jmelson]

Originally Posted by Al_The_Man Reliance Electric drives, which are owned now by Allen-Bradley (Rockwell) use resolver input fairly extensively. The problem with mix and matching a resolver motor/drive combo is a bit harder than the average DC or DCBL as although another drive accepts resolver input, the resolver excitation signal has to conform. If you find out or detect the pole count, I would tend to replace the resolver with a commutation encoder such as Renco are dumping on ebay right now. Although this converts them to DCBL from AC sinusoidal, I still get good performance.

I have to warn you that this is not a "given". I have run a number of brushless motors from my six-step drive (what Al calls DCBL) with good results, but I have also seen two models that did not perform well AT ALL! They vibrated severely at certain speeds. And, there is no way to predict which models will run well without actually trying it (unless the manufacturer will warn you that a particular model is not suitable for 6-step drive,

Jon

[Al_The_Man]

According to the spec sheet, the CanOpen bus can program and run it and/or it can be run from step/dir or analogue input, at least that is the way I read the drive info?
Al.

[jmelson]

Originally Posted by Al_The_Man According to the spec sheet, the CanOpen bus can program and run it and/or it can be run from step/dir or analogue input, at least that is the way I read the drive info? Al.

I'm pretty sure that Mach doesn't support CanOpen, or is this something new?

Quoting from the manual :
Torque, velocity, or position commands can be generated from an analog input, the auxiliary encoder input, a preset index table, or the CAN interface.

So, it can take a form of "steps" from the auxilliary encoder input, which would be a quadrature signal, not exactly step and direction. it would be real easy to convert step/dir to quadrature, of course, I think there are even chips that do exactly that.

Maybe there is some setting of one of the drive parameters that converts the aux encoder input to step direction, but it didn't clearly state that in the little spec sheet dx15co8.pdf

Jon

[charliegolf]

I'm pretty sure that Mach doesn't support CanOpen, or is this something new?

No, I'm pretty sure it does not. CanOpen is a whole new "can" of complexity that I'm sure the Mach team could handle, but I'm not sure I could, he he.

I have tried to "talk" to this drive using AMC's software on the RS-232 lines (as the datasheet hints at this type of communication), but it has not worked thus far.

For someone just getting into the CNC world, there sure are a lot of pitfalls and booby-traps to watch out for.

What is this 6-step drive that you are referring to?

[jmelson]

Originally Posted by charliegolf What is this 6-step drive that you are referring to?

I have built the most basic brushless drive I could make. It takes a PWM signal from the controller, plus the industry-standard "Hall" signals for commutation. Six of the possible 8 codes of the commutation signals are actually used, which correspond to power being applied to one motor terminal and ground to another, with the 3rd terminal left floating. This is commonly called a six-step drive. This is also sometimes called Trapezoidal Commutation. More modern (and more expensive!!!) commercial drives mostly use Sinusoidal Commutation, which provides current to all 3 terminals all the time. A few motors seem to really notice the difference, but most work pretty well with my 6-step drive. I have run SEM, Fanuc, Panasonic, Pittman, IMTT and Keling motors so far with good results.

Some motors, like Panasonic and Fanuc use proprietary schemes for the commutation signals, so I had to make a decoder for those motors.

I also have a "controller" for these applications. It counts encoder pulses, and produces the PWM waveform to control the servo amp. It handles 4 axes, and another board can be added for 4 more axes. EMC2 comes with a driver for this board. Mach was not designed as a servo program, so it would be harder to use it, but I suppose the Galil card could be used with it.

Jon