Force Control Feed Forward

I am creating a LV interface to the RMC tools and would like to limit the feed forward and rate feed forward controls to a reasonably safe interval, but can’t find any documentation on this? I would also like a range for the differential, but may just leave that zero. We are going to be moving very slowly, probably 200 pounds per second going to about 5000-10000 pounds, and a slow overdamped response is fine.

On a side note to this, can anybody tell me when I write to these registers, is there a way to download them to the 150 controller like in RMC tools, or will the controller see the change as soon as I write to the register? I have it set up to write to flash after the tuning is done, but am not sure if I will even be making changes without some sort of “download to controller” command after I write to the register?

I think you will see the change as soon as they are sent.

What you could do is write to variables then copy them over when needed. Have the varible set up as retain so it will be saved and you would also need to write the variables over at power up.

Yes, the gains will be appplied as soon as you write to them.

The range of “safe” values for the gains and feed forwards varies widely between systems. The only way to determine that is to see how various values affect your system.
The values usually depend on the system gain (how fast the controlled value changes when it receives 1 volt of drive) and response, and the scaling. For example, if some system scaled in pounds-force has a feed forward rate of .1, that same system scaled in tons-force wil have a gain that is 2000 times greater.

Thanks guys, appreciate the help so far. With the velocity feed forward terms there is an equation to generate an initial value, are there similar equation for force and force rate feed forward? I’m more than a little nervous to leave the values wide open.

I currently have my proportional gain limited to one, does that sound reasonable, or is it not possible to genralize force systems? I will do some trial and error before I hand it off, but would like a reasonable starting point. Thanks again.

A limit of 1% is reasonable for the pressure or force proportional gains assuming the units are PSI. For the most part pressure gain is just about dithering the valve to let a little oil in or out using about +/2% of the valve travel. However, the gain changes dramatically depending on whether the system is calibrated in tons or our pounds. Obviously the gains would be different by a factor of 2000. If the system was calibrated in tons then the gains would look a lot more like the position gains where we have units of inches but are controlling to 0.001 inches. If you calibrated the system in tons are were controlling to points, 0.0005 tons, the gains would be similar.

The feed forwards will be about 1000 to 10000 times less than the proportional gains. I can provide you with a formula but making use of it is difficult. It involves knowing the spool constant, cylinder data, current pressures in the cylinder, supply pressure and the current piston position. Are you ready for that? There is no simple formula for the feed forward and ideally the feed forwards should change as the conditions change.

The system is scaled in pounds force. I do not know much of anything about the valves or cylinders, but I can probably gather any information necessary. We are going to be loading one specimen, we will not have much time for tuning, and if we somehow fail this specimen, like send it into resonance, we cannot replace it. I will have the opportunity to try the system out and tune it while loading a fixture that is very UNrepresentative of the actual test article, but I’m not sure how useful this will be in setting the PID paramaters.

The good news is that we can move VERY slowly. If there are equations that can get me started, I would be more than happy to research the hydraulics to try and work them out.

When you say a limit of 1% is reasonable (although I will be using lbs force, not psi) does this mean if my maximum force is 10,000 lbs, a prop. gain of 100 wold be reasonable?

Are you ready for this?
dF/dt=βKvSpool*(Acapsqrt(Ps-Pcap)/(MinVolcap+xAcap)-Arod*sqrt(Prod-Ptank)/(MinVolrod+(CylLen-x)Arod));
Where
F is force
β is the bulk modulus of oil
Kv is the flow constant. This comes from Q=Kv
sqrt(ΔP)
Spool is the spool position. -1=full P->B A->T, 0 is off, 1 = full P->A B->T. Any value between -1 and 1 is valid for servo valves
Acap is the area of the cap side of the piston
Arod is the area of the rod side of the piston
Ps is the supply pressure
Pcap is the pressure on the cap side of the piston.
Prod is the pressure on the rod side of the piston
Ptank is the pressure at the tank assuming the t port is the same as the tank port pressure
MinCapVol is the oil trapped between the piston and the valve when the piston is retracted all the way.
MinRodVol is the oil trapped between the piston and the valve when the piston is extended all the way. This value is typically larger than MInCapVol because of the transfer piping.
x is the position of the piston. Full retract is 0.

Let me know if you can make use of this. As you see the answer isn’t simple. The gains will change as a function of position within the cylinder and the current pressures and the pressures are always changing. However, all is not lost. I have found that good average estimates help significantly and most applications meet obstructions and go into pressure at the same position each time so presses are relatively simple. Test machines are more difficult when the pressure ranges change a lot.

As far as an appropriate starting gain, from my experience, especially if you are worried about ruining the part, start at a very low prop value, like 0.001 or even smaller.