Servo Valve and Dither Frequency

The only use of the term "dither" I know of is in data sampling technique. In this form a known random noise level (now there's an oxymoron) gets added to an over sampled signal so that averaging can produce a higher resolution than normally available. I have no idea how this concept can be applied to a servo valve.

When an actuator is used as part of a load control the actuator can develop excessive static friction. Dither is applied to keep the actuator in a constant state of motion (small) so that static friction does not occur.

Dither is a low amplitude, relatively high frequency periodic electrical signal sometimes superimposed on the servo-valve input signal. Dither is generally used to improve system performance by keeping them in constant motion to keep them from sticking at null (reduces stiction).

The flight simulation industry used to use dither prior to moving to hydrostatic actuators. I don't recall what the dither frequency was but I believe it was about 150 Hz. The amplitude was quite small (1% or so).

Here's a link to a technical bulletin by Moog.

Thanks for replying. I'm familiar with the article. It mentions dither amplitude, but doesn't say much about dither frequency or the merits of lower frequencies (60 to 150 Hz) versus higher frequencies (200 to 400 Hz).

That's because it is highly dependent on the application. If the natural frequency of your system is high enough (in the 60 Hz range) and you use a 60 Hz dither signal, you will be exciting your system. So it's best to have a dither frequency well above that of your system.

usually dither frequency has to be higher than max possible servo frequency. Servovalves can today go as high as 200 Hz. If the dither falls in the possible working range it can have a disturbing effect combining with the working signal. The spool amplitude to maintain a stick slip free motion can be very small. Dither has also a second effect in the zero region. In order to reduce leaks spools are manufactured with a small positive overlap. This without dither leads to dead zone around zero. In presence of a dither with an amplitude about the overlap the spool+sleeve combination work as without overlap thus no dead zone. If there is a negative overlap the flow/pressure gain around zero is the double of the gain in the rest of the stroke and this could lead to instability around zero. If the overlap is too big (all is relative) the control loop requires a signal to open and correct position but such a signal is given by an unwished movement of actuator. As you see dither has not only the friction reduction effect but much more contributing to quality and stability of control loop.

It's all used to reduce the static friction (stiction) as stated previously. If you keep the spool constantly moving in a very low amplitude, high frequency, it does not have to overcome the stiction, and can be shifted faster and easier.

Many valve manufacturers have specific frequency ranges of dither to be used with their product, the specific information can always usually be found on the datasheet. Most hydraulic valve controllers have a dither adjustment on the valve card driver, to "tune" the valve for optimal performance. Different geometry/pressures/flows/seals etc work best at different frequencies.

Nice post! ;) Dither is applied to keep the actuator in a constant state of motion (small) so that static friction does not occur.

NO, dither should NOT have an influence on the actuator but ONLY on the servovalve spool. It helps to better the loop quality but should not change the actuator behavior.

Stick slip in servo-actuators has to be reduced by use of adequate seals. A broad proposal from the seal manufacturers is available. First steps have been done in airborne systems by the development of combined seals : one material for the contact at sliding surface and another as seal static and spring for the first in order to maintain a minimal contact pressure at the wall.

Not completely true.

I just this week had a customer who needed to keep two cylinders in mid stroke position, at the ready to instantly move when commanded (without any stiction on the cylinder seals). This was achieved by using an on off valve running at a high-ish (secret) frequency. You could not see the actuators moving... but they were, ever so slightly, you could feel them, and the on off pulses of pressure kept the cylinders in perfect balance with their load in mid stroke. It worked very well using a dither on the actuators themselves... It of course was all controlled by a magic computer box.

But I agree that 99% of the time it's not the desired result to have any influence on the actuator, but simply on the spool itself.

Of course this can be done and the best way is to use 2 x 3/2 valves actuated at high frequencies so that the flow integral is nil although there is flow in and out. I used once this approach for a hydraulic system. It is very often used for the control of pneumatic rams since compliance being higher it is easier to do it. It is also possible to use 2 x 2/2 valves coupled with 2 constant resistors in a bridge with the piston on the diagonal. In fact the servovalve spool is the piston of an actuator which plays the role of a power amplifier for the rest of the circuit. I know a "proportional valve" which was designed with a similar system of command 2- 2 x2 + 2 resistors. The valves did not close completely due to short period time so that they only modified a resistor value shifting the bridge and the main spool. If in your case the piston did not move (for the eye) because of the very high frequency a lot higher than the system proper frequency and because of the compliance of the 2 volumes between valves and piston. I also think that due to very high frequency the valve did not open completely so that the flow was thus reduced.

I thought a bit over what you said. It is not difficult to determine which frequency you used. In the 1st phase the spool is hold in position by the springs which are prestressed so that it stays in the middle. Directional valves have an overlap in order to reduce leaks.As long as the magnetic force does not override the spring preload the spool will not move. When the magnetic force becomes big enough the spool starts to move and here is the problem since the magnetic force increases roughly with the square of the displacement and the spring force only linear (spring stiffness). In general the wish is to commute fast so that the force increase versus stroke is quite small. If you shut the supply early enough the spool will start its movement and at same time (almost) the force starts to decrease so that the spool is again centred by the springs.The limit frerquency is 1/ t1 where t1 is the time between voltage "on" and spool movement start . A frequency slightly lower will satisfy the requirements. If it is too low then it could lead to a full movement whic is not wished. In fact you work on the overlap range as bridge with 4 resistors all of them variable. It is a good idea it only asks for a very fast current decrease and this is not always as easy to do as it could be assumed.

If you are trying to optimise performance, you should investigate relating the dither to filtration and particle sizes within the oil. In some safety applications, directional (not servo) control valves have to stand idle for a long time (months or sometimes years) but operate reliably when triggered. For these applications dither was applied to prevent the valve from "silting up" with particulates present in the oil. In the applications I am referring to, the filtration was much finer than in modern aircraft servo applications, so the short term effects of silting must be present in the systems you are attempting to optimise.