integrating IEEE card on PLC


Thread Starter

Harry Friemann

Dear 'collegues', I am looking for a solution for the following problem: I have to update a machine to measure electrolytical capacitors. The throughput rate of the machine is 6000/hour. The whole machine is controlled by one axe, which therefore makes two revolutions per second. I want to remove a large number of sensors and use an absolute encoder instead. This will be my trigger for all activities. The measurements have to be with Itronics Fuchs equipment using IEEE (possibly RS232). We want to control the machine, make settings to the measuring-equipment bij computer and some day collect the measurements. The problem I run into is: - When using a soft-PLC I do not know how to integrate IEEE into this (The books do not refer to this at all) - When using PLC I do not have the ability to use IEEE (PS: I know Philips PC20 - this does not mean I cannot use anything else) and even when I will use a PLC together with a PC I do not have a PLC which has a cycle time of 2 msec or less. - I have thought about using labview with a IO-card and an IEEE card but I have trouble trusting WindowsNT I need a reliable and fast system. When a measurement is started, after getting the result 'wrong', I have to activate an output for ejecting this product directly. Can I integrate IEEE into TwinCAT, WhizPLC/CON, etc.? Does anyone know of a PLC that has a cycle time less than 2 msec. (for about 1 kb - preferably more) ? Does anyone tell me if and how I can trust windowsNT, with labview for this problem? I know I could use Linux, but have zero experience and I am alone on this, no fancy team and so on. I hope any can help me further, Sincerely yours, Harry Friemann Assistent Technische Informatie Systemen BCcomponents, afdeling R&O Tel: +31 38 4575 317 Fax: +31 38 4575 298 E-mail: [email protected]

Michael Griffin

I don't know if you will find a PLC with a scan rate of less than 2 msec., unless it was running a trivial program. I would be surprised if any soft logic system could guarantee that either. If you are measuring these capacitors one at a time, then 6000/hour amounts to 1.67 msec per part. This is very fast indeed. In addition though, you mention that the measurements will be made with an "Itronics Fuchs equipment using IEEE". I assume this is some sort of capacitor tester. Even if the instrument itself is fast enough for your needs, you may wish to check that the IEEE 696 communications will be fast enough (if this communications is part of your test cycle). Many IEEE instruments have response delays that are much longer than the maximum time you are looking for. Also, if you have several IEEE instruments on the same bus, one slow instrument can hold up the whole poll cycle.

Curt Wuollet

Consider Labview on Linux. Faster and more reliable and IEEE488 should be well supported. unlisten, listen, my_address..... Regards cww
Why do you need a PLC with a scan of less than 2ms? The rate of 6000 parts/hour give you 600ms per part. Kind of fast, but not unreasonable. Or does the measurement have to be done within a small window of the total cycle? What are the details of the measurement - ie, is there a pass/fail range or are you just collecting data. I'm not familiar with that instrument, but many have discrete i/o connections that will accept a trigger and provide a pass/fail output. The range could be set through the IEEE port as a setup parameter. However, if the range changes for each cycle, this approach wouldn't work. Overall, what I would look at would be a PLC with a resolver interface module (I'm assuming this is a continuous motion machine?). The resolver module (such as from AMCI) has high speed outputs that can trigger the instrument at the proper point in the cycle. The PLC can read the result and use a shift register to track it to the reject station. In addition, connect a PC to both the PLC and the IEEE port of the instrument. Setup parameter can be sent from the PC to the instrument and data can be received back. A lot of this depends on the process details, so if I made a wrong assumption let me know and I'll toss up some other ideas. Don D.

Michael Griffin

I have a bit of a correction to my previous message on this. At 6000 parts per hour that is 600 msec per part, not 1.6 msec as in my previous message (1/(6000/3600) = 600 msec). I assume the test cycle is some fraction of this number - certainly less than 1/2 of a second. In this time, the instrument must be triggered, the test performed, and the result reported. This *might* be within the reaction time of the GPIB bus, but that can only be determined by researching the specifications and most likely, also by testing. The communication speed of the bus itself will not likely be a problem. However several other factors will likely provide the contraints. Most GPIB instruments do not respond instantly. Rather, they seem to have an internal operating cycle which causes them to look at the bus for incoming data on a regular basis. A 100 msec delay there would not be unprecedented, although this of course depends upon the particular instrument. In addition to this, there will likely be a computer involved in the test system (I have not heard of a GPIB interface card for a PLC). Those computerised test systems which use a Visual Basic program running on a PC with a Windows NT operating system which I am familiar with seem to have unpredictable delays and irregularities in GPIB bus polling which can be quite significant. These delays and irregularities seem to originate within Windows. These delays are generally tolerable in test systems with longer cycle times, but they may present a problem for your application. If you intend to use a GPIB controlled instrument with a PC in the application you have described, you may need to consider what degree of real time response you may need. If the instrument offers a digital I/O interface (for start test and pass/fail), you may wish to seriously consider investigating using that instead.
> I don't know if you will find a PLC with a scan rate of less than 2 > msec., unless it was running a trivial program. I would be surprised if any > soft logic system could guarantee that either. If you are measuring these > capacitors one at a time, then 6000/hour amounts to 1.67 msec per part. This > is very fast indeed. > B&R PLC's multi-task scheme allows task priorities, user definable task scan time, hard timer task as short as 1ms guranteed. Check B&R also have soft PLC running the exact same multi-task operating system on Pentium-based PC, using the exact same programming environment (LAD, IL, SFC, ST, Ansi-C). You can also find a lots of variaties of high-end motion control. We are using B&R in semi-conductor equipment which involves lots of motion control, fast reaction and third-party connectivity. Mark