Insane Hydraulics

Site theme image

Dither in Proportional Solenoid Valve Control

I had to run some tests on a proportional pressure reducing valve this week to map out its characteristics (a client asked us to find a "reasonably priced replacement" to the OEM's original but pricey solution), and it occurred to me that this would be a great opportunity to illustrate some of the ways a proportional solenoid valve can be driven and what this whole "dither business is about. I'll start with theory, and, as usual, end with a hands-on experiment.

A pressure controlling valve is a mechanical device that balances forces - one pressure-dependent and the other coming from a compressed spring or, in the case of electric proportional valves, a solenoid actuator.

In theory, the force produced by the solenoid is directly proportional to the current, and, again in theory, all we'd need to be able to precisely control our pressure would be to control the solenoid current with a simple linear regulator, but in real life, things aren't that simple.

The actual physical valve is an intricate arrangement of precision machined parts that slide against each other and therefore are subject to hysteresis, caused mainly by friction and stiction, and we all want our control systems to have as little of that as possible, don't we?

Dither to the rescue! In the old days, engineers would come up to a steam gauge and tap it to get the correct reading. As the steam pressure changed ever so slightly, the needle would stay in place due to stiction, and so the tapping would "loosen" the gauge mechanism and allow the needle to settle in the correct position. This is exactly what dither does to a proportional valve - it creates a certain amount of vibration in the force created by the solenoid actuator, which removes stiction and thus improves the valve's hysteresis.

All modern proportional solenoid drivers use PWM (Pulse Width Modulation) in order to achieve effective current control. And one may ask - "Isn't the PWM, by definition, already introducing enough vibration to our control system to reduce hysteresis to an acceptable level?" And I say to you - "Good point, you are definitely on to something!" Very often amplifier manufacturers use the PWM frequency in the range of 50-200 Hz which, in itself introduces enough vibration to assure the correct function for most valves. Also, very often the amplifier will allow you to adjust this frequency in order to accommodate it even better to the mechanical arrangement it is going to "work on".

I bet your next question would be: "Hold on a second now... But what happens when we approach the 100% duty cycle of our PWM? Doesn't it make the vibrations smaller? A 100% duty cycle equals injecting DC directly into the coil, does it not? Will the valve be sticking again?" - What a great question! Once again, you are right. This is exactly why a self-respecting proportional valve is never designed to work in the 100% duty cycle regime and can reach its maximum pressure setting with a lower duty cycle (60-80% at its rated voltage). This, by the way, is often the reason why injecting the rated DC voltage directly into an unknown proportional valve "just to test it" is not the best of ideas. It can be designed to withstand this kind of abuse, or not... in which case it will literally fry if you "cook" it for long enough.

But there's also a second way to dither a proportional valve - an (arguably) better way, which uses a higher frequency PWM, with a low-frequency dither signal super-imposed over it. Say what? This is actually easier than it sounds, and in fact, I prefer not to tell but to show in the hands-on part below - because seeing is way better! But before we dive in - let me ask another question: What if you don't have a "proper" proportional valve driver with the dither thing and all and still need to test a proportional valve? Does this mean that you need to dish out several hundred bucks on an industrial amplifier card? Well... actually no, you don't.

The thing is - if your objective is to verify if a proportional valve works or not, even a simple linear current driver will allow you to safely test 99% of the valves you'll come across. With a higher hysteresis, yes, but still perfectly valid to determine if the valve functions. There's even a hidden benefit to this method - if the valve displays an acceptable level of hysteresis when working with the direct current drive - it will work even better with a dithered driver.

All you'd need for this would be a cheap lab power supply, like this one here, for example:

The main advantage of such power supplies is the direct readout of the volts and the amps and the fact that they have both the voltage and the current limitation, which means that they are protected against a short, and will easily allow you to safely detect if there's a hidden fly-back diode in the coil, without burning it out.

I also discovered that these power supplies, most likely due to their "economic" nature, have very noisy outputs, especially when in current limiting mode, which, I guess, can count as a dither of sorts. Just have a look at these scope images. In both cases you have the same power supply connected to a purely resistive load and set at 5V and 1A, only in the first case it is voltage limited and in the second - current limited.

As you can see, when the supply is in the current limiting mode, it introduces pretty high low-frequency spikes (at about 20 Hz). I am not sure, really, if this matters much, but I do use the current knob when I test the coils. And I do know that it works.

But enough talk. Let's run some valve tests, shall we? The test plan is the following: I have a proportional pressure reducing valve with a 10-ohm coil, that will be fed with a constant 40-ish bar in the P port, and I'll drive it first with the lab power supply, and then with a proper driver (the Vickers EHH-AMP-702-D-20 amplifier plug), equipped with an adjustable dither setting (of the more advanced type I mentioned earlier), and then see what happens to our pressure when the control input goes up and down, It will be fun, I promise!

This is the test set up and "supplies":

This is what I got with the direct current drive from the lab power supply, compared to the PWM plug with the dither setting set at minimum:

Current, A Press. Up Press. Down Press. Up (PWM) Press. Down (PWM)
0.1 1.0 3.5 1.0 2.0
0.2 1.5 8.0 1.4 4.2
0.3 4.1 11.0 4.0 8.2
0.4 7.2 14.7 7.7 12.0
0.5 11.1 18.2 11.8 16.1
0.6 16.6 21.8 16.2 20.6
0.7 19.7 23.6 20.9 25.5
0.8 24.5 29.6 25.2 29.5
0.9 28.8 35.1 29.1 34.0
1.0 32.6 38.5 32.6 37.5
1.1 36.0 41.3 36.8 40.2
1.2 39.7 42.5 39.6 41.9
1.3 42.0 42.5 41.7 42.5
1.4 42.5 42.5 42.5 42.5

As you can see, the hysteresis is pretty high. Just have a look at what happens at 0,4 A input with the current drive - you get the 7,2 bar and 14,7 bar - that's a big difference! However, the 1,2 kHz PWM plug, although performing a little bit better, still allowed for significant hysteresis.

Let us try to improve this, and see how the valve fairs with the dither set at 25%, 50%, and 100%:

A P Up, Dith. 25% P Down, Dith. 25% P Up, Dith. 50% P Down, Dith. 50% P Up, Dith. 100% P Down, Dith. 100%
0.4 8.5 10.5 8.7 10.5 XXX XXX
0.5 12.5 14.7 12.6 15.0 XXX XXX
0.6 17.2 19.2 17.5 19.3 17.4 18.5
0.7 21.9 22.8 22.4 23.7 22.3 23.4
0.8 26.2 27.2 26.6 27.8 26.6 27.2
0.9 30.5 31.6 30.3 31.7 31.2 31.5
1.0 34.5 35.5 35.0 35.2 35.0 35.1

As you can see, increasing the dither does decrease hysteresis significantly, but there's a "catch", so to speak. Do you notice that I didn't manage to get any pressure control below 17 bar with the dither set at 100%? This happens because when you introduce the dither "of the super-imposed kind", you also introduce the limits to the minim and maximum current you can achieve, thus reducing the width of the "current window" you can work in. Here are the maximum and minimum current and pressure values I managed to achieve with different dither settings. (Note that I measured the current with a digital multimeter, which measures the average current over a sampling period - very convenient):

Dither A min P min A max P max
25% 0.15 1.0 1.64 41.5
50% 0.28 5.5 1.41 41.6
100% 0.59 17.4 1.21 40.7

If I had to use that particular valve with this particular plug, I would leave the hysteresis adjustment at 25%. Gives an acceptable control and allows me to use the full range of the valve. It's curious that in some places it performs even better than the 50% setting.

Now, let us have a look at the PWM. Here you can see the scope shots with the 5V and 10V control input with hysteresis set at 0%, 50%, and 100%, and I used a 12V rated 5-ohm coil for this (you'll see why in a minute):

You can clearly see what is meant by "super-imposing lower dither frequency over the 1,2 kHz PWM". You can tell that the dither has the 8 ms period, which equals 125 Hz - a pretty standard dither frequency.

An interesting thing to notice here is the fact that I only got to the 50% duty cycle of the PWM with the 5-ohm coil that I used in the scope test. When I connected a 20 ohm 24 VDC rated coil - I could easily get to the 100% duty, even with a lower than 10V control input. This tells me that the plug takes into account the immediate current that flows through the coil, and adjusts the PWM accordingly to limit it. This is very important because this means that this plug will adjust the PWM to the coil resistance change as it heats up (and coils can get very, very hot) - thus making our control system more precise!

In all - I am satisfied with the Vickers plug - it performs as it should. I do have other plugs and cards lying around, some of them broken. Maybe I'll recover a couple of them and make a post about it in the future...

Important points to take from all this:

An aside about hot-wiring proportional coils:

I get it - when a man needs to test a valve - he tests it with what he's got. But if you really need to perform such a test - at least use a multimeter first to measure the resistance of the coil to see what you're up against before connecting the wires. And only test for short periods. By the way - automobile lamps are a great current-limiting device. Last resource of course. Especially in this day and age when you can get a decent current and voltage regulated power supply for about 50 bucks.