PLC with micro-second time base?

Does a PLC exist with micro second time base? We have an application that involves calculating time in micro-seconds.. Wondering if a PLC can measure it..

Thanks for your time

1 mico second, as in 1E-6s? or did you really mean milli second?

I'm unaware of a more or less standard PLC that could produce or count a 1 mhz pulse rate. Maybe Peter knows...

I don't know of a PLC CPU that can do it, but you might find an intelligent card that can. I know Siemens makes a CP352 card that gets down to the microsecond level, so you might look into that.

A very few, very obscure, PLCs can indeed provide micro-second time to the process from the real-time clock. (The most likely providers of micro-second timing considerations are the Soft-PLCs.)

HOWEVER... with respect to typical PLCs, any attempt to make a control decision on a micro-second basis is actually quite useless. This is because typical PLC scan-time is usually on the order of tens of milli-seconds. Trying to react/respond to a micro-second change in time on that scan basis is fruitless.

In general, this is also true with respect to Soft-PLCs because they are typically designed to develop a CALL_PLC_Interrupt routine on a scheduled basis that is very similar to a typical PLC scan-time.

However, for very small systems, that call might indeed be made on a micro-second basis (however, NOT on the 1-micro-second basis, more likely on the tens of micro-seconds basis)... although, that call should NEVER, EVER, be allowed to STEP-ON the current "scan". As always... it depends!

S7Guy said:

I don't know of a PLC CPU that can do it, but you might find an intelligent card that can. I know Siemens makes a CP352 card that gets down to the microsecond level, so you might look into that.

Click to expand...

I looked into that card recently for a project I am working on that requires very high precision and fast reaction times. I was pretty impressed with it, although I eventually used another technique to get the required speed.

1. Can be programmed in S7 with a restricted instruction set (no real numbers etc)
2. Scan time down to 1 micro-second (no typo) due to partly parallel FPGA architecture
3. 15 inputs, 8 outputs, enough for many tasks. Can't expand IO unfortunately, but you are unlikely to require lots of points that need that sort of processing speed.
4. Can run as independent or co-processor (can share memory with another processor over backplane)
5. Not as expensive as you'd think (about $1200 Australian)

Very useful for tasks that require high precision provided your mechanical equipment can make use of the extra accuracy.

Binaural,

Can you point-me to the data-sheet or, preferably, the manual on that module?

Uhhh.. after re-reading... is that a non-expandable CPU Module?

http://support.automation.siemens.com/WW/llisapi.dll/csfetch/9240171/fm352-5_e.pdf?func=cslib.csFetch&nodeid=9237692


Enjoy

Thanks for your replies..
Actually 10us would be fine. Even 100us can work, although we would prefer much more resolution.

The Siemens looks good.. do you know the approx price of it?
[edit]: Just noticed the price, does that equate to less than 900USD?!

Terry, can you give me some information on Soft-PLCs? Is this a computer with a card connected to it?

Thanks..

Cryogen,

A Soft-PLC is a program written for use on a PC. It typically requires a special card that can read Inputs and control Outputs. However, it can also be set-up that communication to various Input/Ouput Control Modules are handled by the Serial or Parallel Port.

My experience involves using special I/O cards with programs, that I developed, written in "C". As I recall (several years ago) each card was capable of handling 64-I/O. Each individual I/O point could be designated as an Input or an Output.

The basic scheme is to write a program and then install that program as a TSR (Terminate and Stay Resident) program. The same scheme is used by the Mouse, the Keyboard, and the Clock-Handler in the PLC. This occurs at the most primitive levels of the PC. The Soft-PLC program is then "called" by an "Interrupt" that is generated by the Clock-Handler. The Clock-Handler is the fastest aspect of any PC. That speed depends on the particular CPU and Motherboard. The older PC's did "steps" at least at the micro-second level. These days... it's probably at the pico-second level or maybe even smaller.

If you know what the maximum time is for the particular application, including ALL of the time consuming over-head issues, then you can reduce the timing of your "Call" to the "Interrupt".

However, while the Inputs operate independently, it takes a certain amount of time for those Inputs to react to field conditions. This means, if you can call your interrupt every 10-micro-seconds, and yet it takes a particular input 15-micro-seconds to respond to the actual condition... you won't know that the input has changed state until the next call.

Then there is the issue of feed-back from remote controllers. This feed-back is subject to the timing of the communication routine and all of the assorted over-head issues.

All in all, falling back on the PLCnet Standard answer... "It depends!"

Are you looking for microsecond response times, or are you looking for microsecond resolution of events ?

Rockwell's ControlLogix, for example, uses a microsecond timebase system clock. If you have two timestamped events in the controller and compare their timestamps, your resolution is in microseconds. Now, I/O module reaction times and system interrupt latency are generally in the hundreds of microseconds, but the actual timebase is in fact, 0.000001 seconds.

What are you trying to calculate ?

I am a little curious. What kind of application is this??

Another possibility apart from the FM352 mentioned by S7Guy and Binaural (at least I assume the CP352 was a misprint), is the FM352-5. This is a fully fledged parallel processor which can also be used stand-alone - i.e. it does not need to be a aprt of a Siemens Step7 system.

The cycle time is 1 micro-sec and it can handle input and outputs at 10 micro-second resolution. I used three of them to provide delay timimg of thyristor trigger pulses in a recent project and was able to get delay time of 1 - 200 ms with a maximum error (at 2 second delay) of just over 200 micro-secs.

You can read about my experiences with the beast here.

look at the new controllogix sequence of event modules

http://literature.rockwellautomation.com/idc/groups/literature/documents/pp/1756-pp007_-en-e.pdf