PWM and electro hydraulic valve


Thread Starter

Richard Perlefelt

Hello every-body

I wonder if somebody can give me information regarding PWM control of hydraulic valves.
I use 22ohms, 24volts magnets, 120mH. Which is the best method of controlling those valves. Low frequency (100Hz to get the ripple) or using higher frequency and the add the ripple to the control signal ? Which is the best frequency if I
use higher? Is it possible to generate this signal on a PWM output on a Texas DSP TMS320LC2404A ? . Have anyone experience of EMI problem using higher PWM frequency and long cables between control system and valves, over 10m, I need to be under class B ? . And a book about this subject, did anyone knows a good
one ?
Regards Richard

There's no good book on the subject as far as I know. Also, the best way to do it is to characterize the particular brand and model of valve you are going to use. My personal preference is to use a good valve and use two frequencies, one for current control and one for anti-hysteresis.

The best chip I've used for this is an SGS L294. You can set the PWM frequency to about 10 KHz, which turns out to be inaudible for most valves. This is the current control frequency. If you still need to cut down on the hysteresis, you should add another dither on top of that. If you can get a bode plot of the valve response from the manufacturer, this will help. It needs to be above the point where the valve will respond, probably somewhere between 80 Hz and 250 Hz. The lower the frequency, the less hysteresis; but if you get too low, you may get some resonance, and it may show up in some other component in the system (like a cylinder). About 10% amplitude is usually sufficient for the superimposed dither.

If you don't want to control current, then you could use the output of the DSP to do the anti-hysteresis dither. You can get into trouble if you run long lines with the 10 KHz signal (say 100 feet), because the coil is part of a tuned circuit; but that shouldn't be a problem on the low frequency. CDf this is a mobile application, you should definitely use current control, otherwise the valve characteristics will change markedly over the wide range of supply voltage present in such a system.

If you use a really cheap proportional valve, you may have to characterize each individual valve. Usually you will only need to find where the valve cracks, and where the valve is full open. Then you can straight line the response in between those points. The values can be stored in EEPROM.

Some valves will be rock solid over time, and others will change as their springs wear. I won't put different manufacturers' names on the list, but there is a great difference between different companies' products; make sure the valve is capable of doing what you want ot to do.

Also, you should characterize the valve's performance over temperature. We characterized one valve that worked fine at 0 degrees and 90 degrees, but had a sticky place at 60 degrees (All temps in Celsius). An analysis showed that the different coefficients of expansion of the materials used caused this problem in that temperature range, and the manufacturer wasn't even aware of it.

These are a few of the things that come to mind about this problem. If you have other questions, I will be glad to help.


Willy Smith
Numatics, Inc.
Costa Rica