AB ControlLogix is fast but it is expensive.

You may not need a faster PLC, if it is only one 'task' that needs to run that fast.
Many PLC's (don't know about Automation Direct) support interrupts that may be able to respond much quicker than the main program.
Francis

Hello;
     If you need response times in the order of a few milliseconds, most commercial PLC hardware will not work. If for no other reason the I/O cards can not respond that fast. For this type of application you will need semi-custom microprocessor based solution.

It would be interesting to know what sort of equipment Jeremy is using a PLC to control.

I don't think you should use a PLC if you need timer resolution to a 1ms. Above someone suggested a interrupt driven solution, I would look in that direction. I believe that Automation Direct does have a module to do timers at 1 ms. Look at H2-CTRIO AND THE D2-CTRINT. I agree with those above that suggest you shouldn't be using a PLC in application as you have described it. I'd use a $2.00 microcontroller and build my required I/O. Then, the sky's the limit on speed.

PLC emulates relay logic, so it was not designed for speed.

Any microcomputer/controller can outperform most PLC if you
program it in e.g. "C" or assembly.

<nbucska@pcperipherals.com>

Many control tasks really do not need sub mSec speed. PLC's if suitable are much cheaper - a $2 chip that takes a year to program is far more expensive than a $2000 PLC that takes a month.
Jeremy - what process do you want to control?

I agree that it would probably be faster to program the PLC and less demanding design skills. I use a PLC when one will fit the application. I really think that the question is whether these speed requirement are legitimate needs or wants. The $2.00 chip shouldn't take a year if it is a simple program as jeremy suggests. A month should suffice. I'm not advocating this, I'm just suggesting that ultra fast PLCs can cost upwards of $10,000 and generally are not good substitutes for embedded engines! AB's Controllogix is very fast; but, try getting in one for under $10,000 considering development product costs. Jeremy, the fast I/O interrupt modules available for Directlogic can be used to achieve very fast real-time like responses like you've indicated. Take a look at these. I also would like to here more about the intended application!

You shouldn't assume PLC is a universal solution. On my
latest project I had to solve trigonometric and cubic
equations, perform real time control with 2 uS increments
-- try to do this with PLC !

<nbucska@pcperipherals.com>

The process I was trying to control was the pressure of an accumulator tank we were using to pump product.  The problem is that I need to control this within about .1 psi to get a very consistent flow.  I was using a pressure transducer going into a 4-20 mA input to read the pressure, and then adjust the pressure using 2 solenoid valves to either put air in or relieve air out.  We have a pump adding product to the accumulator in spurts, so it has to relieve a lot during this time, and add in gradually when the pump is off.  I used a setpoint and the pressure input to determine the length of the pules to the vales and the time between them.  This would have worked very well, but I could not get the pulses fast enough to react to the changing pressure.  We have tried to use electronic pressure regulators also, but could not get any that reacted quick enough either.  I think I could make it work if I had a PLC with pulse output modulation to change the duty cycle of the outputs and then use some interupts inside the program, but I'm not sure.  Sorry this was so long, but I didn't know how to say it in less.  Thanks for all the input so far.

Jeremy

Hello;
     I would think that the limitation is in the solenoid valves, not the control system. I would think that what you need is two motorized flow modulating valves. Expensive beasties but if the application demands, then you must pay the price.
     Using these types of valves then the control logic can use a PID controller (either in the PLC or hardware). The polority of the output of the PID then determines which valve is to be controlled. For example if the output goes positive, control the vent valve, when the output goes negative control the pressure valve.
     Sizing of these valve is very important. Too large and you can not get the control you need, too small and the system will not respond fast enough. The valve manufacturers can help you select the size of the valves, it is an arcane disipline.

It still sound to me like PLC territory, but it also sounds like a 'classic' - the problem is in the physical equipment and you are trying to compensate with the control.
Can you make the accumulator tank bigger?
Or have fast and slow solenoids?
And on the control side can you use the run / stop state of the pump to provide some feedforward?
Francis

DJS and I posted simultaneously! I agree that DJS has a good suggestion.

Yeah, the situation looks like a recipe for wearing out valves quickly and prone to instability. There are several viable approaches to this, one being a barber Coleman modulating valve as described above. You didn't say that your valves were proportional; but if you were trying to use regular valves, they can't respond faster than 10 ms anyway so I don't see how that was going to be controllable. Another solution would be to put your accumalator in a large cylinder and using a linear actuator slowly change the volume to control the pressure. I've done this in a similar situation and found control to be rather easy without pushing the controllers specs at all.