I need a PLC to generate pulses.

[Stephen Luft]

Peter,


I can give you a Smart-PAK PLUS with a PWM output that you can use.

If you are interested, give me a call.

God Bless,

 

[Steve Bailey]

Peter, take a look at the High Speed Counter module for the GE Fanuc RX3i, part number IC695HSC304. The manual is GFK-2441, available from GE Fanuc's website.

I don't have time to delve deeply enough into the manual to give you a definitive answer to whether or not it can output a 2.5 kHz square wave pulse train, but it looks possible. The module has outputs that can be directly controlled by the counter's accumulated value, independent of the PLC scan. It can handle 1.5 mHz pulse rates and also has an internal oscillator that can be applied to the accumulator.

I'm more familiar with the High speed counter for the 90-30 platform, and I know the best you could get with that one was 2 kHz. The specs listed the output update rate at 500 microseconds. I could not find a similar statement in the manual for the IC695HSC304. However, the specs for everything in the RX3i module are better/faster than the corresponding parameter in the 90-30 module. The only spec I could find was the "Output delay time" on page 1-3 of the manual. Output delay from On to Off is listed at 125 microseconds and Off to On is 85 microseconds. In the chapter where they describe how to set up an output to respond to the accumulator value (pp 3-18 to 3-20), they mention that it is possible to set up an output range too fine to actually switch the output and refer to those output delay times.

Bottom line, since you're looking for an off-the-shelf, documented solution, this could do it for you. Worth a look.

 

[Peter Nachtwey]

Fun in Delta land

After thinking about this some more I realized that the 400 microsecond square wave is just the beginning. The idea is is to take an encoder, do magic filtering and then put two values in a high output and low output counter. The high output counter would be on while it is counting down. The low output counter would be on while it is counting down. The two counters would alternate counting down. So lets assume one has a 1 MHz counters. If I put 200 into the high output counter and 200 in the low output counter I get a square wave that is a nice symmetrical 400 micro second square wave. Now if the speed slows down just a bit I may need to change the output by 1 count. I can just increase the preset in the low output counter to 201. It doesn't make any difference if the high output and the low output times are exactly symmetrical but it does make a BIG difference that the minimum step change is 0.0025. That means that over 10 inches of travel my error could be 0.025 inches. This is not good because we may need to track an item for 40 inches. However, I can adjust the rate to dither between a 400 microsecond period and a 401 micro period to simulate a 400.5 micro second period. If I have a 10 Mhz timer or counter then I don't need to do the dithering near as often or at all.

You can see that I truly need counters or timers with much more resolution that 2500 Hz.

We are still debating. The Rockwell HSC seems to be barely fast enough so we though if we just were willing to put up with %5 less speed it would do the job but I not can I change the MaxCount on-the-fly so that it out.

 

[RussB]

This can be used a 'standalone' module or an 'expansion' module.
http://support.gefanuc.com/support/index?page=docchannel&id=S:DO2040&actp=search

 

[Peter Nachtwey]

Funny RussB, fat chance.

You know we make motion controllers don't you? Using a competitors module? That would never fly even if it was the right solution. Our engineers would alway want to design something better.

This can be used a 'standalone' module or an 'expansion' module.

I can see it has the pulse output that our new controllers lack but the GE controller doesn't have the filtering and programming that we require.

I think the FPGA designer has won. I know it will cost more than the Rabbit semiconductor solution but perhaps I will get a PWM and Pulse output module for our motion controller. Then we can control steppers and DC motors that use PWM.

I think the key to this is realizing that the timer must have MUCH more resolution than the required 2500 HZ if the frequencys are going to be adjusted in fine increments such as 2499 Hz or 2499.5 Hz.

 

[CharlesM]

I think the FPGA designer has won. I know it will cost more than the Rabbit semiconductor solution but perhaps I will get a PWM and Pulse output module for our motion controller. Then we can control steppers and DC motors that use PWM.

If this pulse output can be setup to simulate an encoder output then you could link two controllers together using this module and and encoder input. That could be very cool.

 

[timbo_uk]

Just a thought: why are you actually looking at modifying this pulse train from the encoder? Is the encoder that is fitted incorrect? Or are you retrofitting something? Can you not add an additional encoder? Is this mod getting over complicated?

 

[RMA]

I somehow missed the start of this thread,

but as several people have suggested, the FM352-5 (not the -2) is the answer to the maiden's prayers.

The beast is much faster than the 1 µs scan time suggests, because the FPGA is a 12 pipe parallel processor. The other thing to remember is that it can also be used stand-alone, ie, you don't have to use it in conjunction with a Siemens PLC.

I found the learning curve to be quite steep, but I documented all my experiences here in a thread called

S7 - Tips and Tricks for FM352-5

The only potential damper I can think of is that the memory is fairly limited - 1200 "cells", if I remember correctly and I had to modify my original concept to get the program to fit.

However, having said that, if it's speed you need, you'll be hard put to beat it!

 

[Mammoth]

GE Fanuc's VersaMax Micro20/40/64 provide 4 64kHz pulse train outputs. It's a cheap controller. Or take the cheaper VersaMax Nano with 6DI/4DO (5kHz)