PLC or PC

I searched the forum and found some of the debate issues associated with this but here is my problem. We have a half dozen or so (could grow to 10) pulse type inputs that we want to generate. We want to do various time measurements with these pulses. Mainly, stuff like, input one was 1.3225 seconds behind input 5 and 3 was 4.5532 behind 2.... so on and so forth. The issue is time resolution. Must be able to sample all signals at 10 kHz (or 1 sample every 100 microseconds). We have been back and forth on the best way to do this. Additionally we need to display our numerical results. Now, PC's and data acquisition seem pretty staight forward and are very capable of this. Only real issue is the whole PC vs PLC in the industrial enviroment. I have been going through the plc specs and I am not convinced they would be up to the job. Automation direct has expandable high speed modules that look attractive but again I'm not sure they could handle the data. Figure 10000 samples per second of 10 inputs. Thats 100000 chuncks of data per second and various math functions have to be performed while all that is being pushed through. Additionally, the unit would have to be communicating to a display as well. So I guess my first question is simply, "Can PlC's do this. If so, could anyone provide guidance on what kind of set up would be needed.

Thanks,

PS, I put 5 bucks up that says you guys could help.

I always tell people to stay as far as possible from a PC, unles they are actualy doing computing stuph, like word processing or cruzin' the net.

Since my personal choice for PLC's is GE Fanuc, I'll start with them.

Their two long time established PLCs, the 9070 and 9030 have high speed counter modules that run at 200khz, adaptable to 800khz. I don't have information readily available for their Versamax line.

I would recommend a 9030, with a High Performance CPU module, maybe even the 374, which I think is their best one. Could be easily readable or programmable from a Horner or GE HMI/MMI unit.

The last time I figured up a 9030 control panel with a hot processor, rack full of modules, auxillary power supply, pilot devices, other goodies, wired and programmed, it was still under $5,000.00.

Siemans did have some 20khz inputs a few years ago. I am not sure what they have now as I haven't worked with them in a long time. I expect that they have hardware that will work.

I am certain that Allen Bradley also has the capabilities, but I stopped using them in 1996 on a regular basis do to cost. I could use a 9030 for $2,300.00 when an A-B would run $25,000.00, and only have one serial port. (Second port ran another $5,000.00 at the time). A-B makes good equipment, but you pay (dearly) for it.

It has been years since I used a Reliance AutoMax, but I'm sure they still have the capability. (Rockwell / A-B Premier PLC).

Entertron will do at least 1 10khz input. Beyond that, I'm not sure of their capabilities.

Easy links to various manufacturers is in the bottom of the Learn PLC's link at the top of the page.

So, I think that most major players in the PLC field can help you out.

I would stick with GE Fanuc, Siemans, A-B, or entertron, as I know the support is available.

best regards.....casey

Here's how you might be able to do it with a GE Fanuc 90-30 High Speed Counter module. The module has an internal oscillator that can be configured to run at 10 KHz and act as the count source. The module also has up to four 'strobe' inputs which, on their off-to-on transition can copy the current accumulated value to holding registers without interrupting the count. The PLC gets updated values of the holding registers every I/O scan. The HSC operates independently from the PLC.

I took a quick look at the manual for the module and it looks like one module could handle up to three of your pulses. I'm not sure about how you'd go about synchronizing multiple HSCs, though.

At any rate. if you'd like to pursue that approach, you can download the manual for the module (publication number GFK-0293c) from the GE Fanuc website .

10 capture registers will be hard to find.

The hard ware to do what you want to do will be hard to find even on an embedded controller. It may take some special hardware to use capture registers. The easiest way would be to get an off the shelf microcontroller that can generate interrupts on a change of state of an input. The microcontroller should be able to respond to the interrupt well within 100 microseconds. I wouldn't trust a PLC to do that.

I'll have to side with Peter on this one. I don't see how you could use a PLC to measure time with that kind of resolution and accuracy. Even hardware interrupts have SOME delay. 500µs is pretty darn quick, but not for THIS application. You're measuring time BETWEEN signals, so the delay would not accumulate for the most part, but even so, I don't think it will be repeatable.

If you used hardware interrupts, what happens when two or more inputs turn ON at the same time? Multiple interupts in a PLC will have a priority associated with them. There goes your accuracy out the window...

IOW, I'd be thoroughly impressed if you could accomplish this with a PLC.

beerchug

-Eric

I would use simple hardware counters as they do not have problems with interrupts etc. However this means building yourself.
the output can be read by serial into a pc or parrallel into a pc that is not big problem.
Your main problem is realtime. so make it in hardware as a processor has timing problems.
In hardware a resolution of 1 MHz is very simple.

I would use one oscillator with 1 MHz crystal.
use ten sets of counters with start,stop reset and a i2c bussystem for output. (yes this is computer based but does not affect timing. and yes it is slow.
If interested i will have a better look on this

This application seems to be out of the league of what a PLC can do.

As you say "Only real issue is the whole PC vs PLC in the industrial enviroment", then get a truly industrial PC.
My best bet would be BECKHOFF .

I would only worry about 2 things:
1. Lock it in (a panel) to avoid tampering.
2. Keep a fully functional spare. This because PC hardware and software changes too fast and you will have trouble to get spares after a few years. And when the worst thing do happen, it takes a lot of software installation and configuration before you are up and running again.

Angie, what you are asking is beyond the capabilities of most (if not all) PLCs. I can pretty much state that the AutomationDirect unit will not be capable of this. Most PLC high speed I/O can only count and trigger and is not capable of higher math or "is this input lagging or leading" type of comparison. I'm going to have to side with JesperMP and say that a industrial pc is the way to go.

This is definitely not for a PLC. I don't think its a good idea to try to force the application into a specialty PLC module. You'll only end up feeling like this: banghead

Use a PC!

If you're in a tough environment, just protect it well and it will do fine. I constantly use Dell computers in a Nema 12 cabinet. I'm in a mid-range industrial facility and have had no problems.

Good Luck!

Well this is going the way I thought it would. (Cohorts still haven't coughed up the 5 bucxs yet, lol). Seems like the PC is the way to go. Thank all of you.

I concede defeat, but would still try it myself with a PLC.

I respect the judgement of these guys, and they probably have much more experience with high speed counting then I do.

So, now that you are going with a PC, what is the best operating system to use? I can't say, I just upgraded to Windows 98 six months ago.

What is the best computer? I am using a gateway, ONLY because it was FREE.

If I were to spend cash money, then I would look at Dell, Micron, or Sager. They all seem to hold up well, even on the shop floor.

The GE factory guys told me a few years ago that some brands of computers were outstanding with their software, and some brands (Compaq!!!) just wouldn't work at all with their software. At the time, they recommended Dell and Micron. I have heard lately that Micron's desktops are still okay, but problems are showing up with the laptops.

Seems like if you can buy a computer at a large consumer discount store, that it may not be the best one suited for industrial pc applications.

Please follow up on this thread after your system is up an running and let us know how it is going.

best regards......casey

Quick and dirty results of using a GE Fanuc High Speed Counter module to measure the elapsed time between pulses.

I configured the HSC to count pulses from its internal oscillator, and set the oscillator frequency at 10 KHz, so each count in the HSC accumulator represents .0001 second. I connected three discrete outputs to the three strobe inputs for the HSC module. I forced the PLC to run at a fixed 5 mS scan.

Ladder logic in the PLC turned on each of the three outputs sequentially for one scan.

Output2      Output3

---] [----------( )-



Output1      Output2

---] [----------( )-



Trigger      Output1

---] [----------(^)-

After each firing of the outputs, I calculated the difference between the holding registers associated with each strobe signal.

The results for 1000 trials:
The range of measured time differences between events was from 47 to 52 (.0047 - .0052 seconds).
.0047 seconds, 9 samples
.0048 seconds, 69 samples
.0049 seconds, 117 samples
.0050 seconds, 563 samples
.0051 seconds, 226 samples
.0052 seconds, 16 samples

Angie, what do you think?