High freq + dither vs req freq alone

[SJ Industries]

Hello, new to the world of Guide, I started to learn about 6 months ago, it has been a very steep learning curve! I am a diesel fitter by trade so have never required this sort of thing in my working life and you could say I have bitten off more than I can chew.

I have a project which requires control of a high pressure relief valve, I have tried various settings and combinations of High freq plus dither (config 5) freq (config 4) also initially used a stand alone amplifier which used high freq plus dither. 

The results with setting 5 are not as good as I would like. Is there a way on config 4 to control req freq so I can continuously change the freq inline with my desired current. I have written a program to do this, it compiles and works, but when I look at the current, duty and Hz on a scope I can see the control must switch the output off momentarily when it changes the frequency, it looks even worse if it is also changing the current output at the same time. I have changed the program to change current and freq separately and it has smoothed it out a little but I don't think it would be very good once live on the hydraulic system. It would have to ramp very slowly.

Thank you, Ben

[FluidPowerTom]

Good morning.  It sounds like you've got things pretty well figured out - especially for a beginner.  In my experience setting the PWM frequency (config 4) or dither frequency (config 5) at an ideal value is important, but I've never seen a situation where it's necessary to change it at different current values.  I'll venture that you'd have to be measuring performance very closely to see an effect.  How much are you needing to change the frequency?  My experience is that 10-20 Hz doesn't make a significant difference in most valves. 

My initial feeling is that if the system performance needs are barely being met to the degree that you're having to optimize dither frequency for different currents then you might just need to use a better valve or make changes to the hydraulic system.  Many times I've been asked to use tricks with the electronics to make up for a subpar hydraulic and/or mechanical system, and those situations typically become a long trip down the rabbit hole.

[SJ Industries]

Hi Tom, thankyou for your reply. It appears I may already be down the rabbit hole. I have hundreds of data logged tests on this system using a standard amplifier and the results were ok, tjere would sometimes be differences that i have struggled to determine the cause. We changed to an MC050 controller to gain more functionality on our test rig to automate some of the tests and to gain more repeatability.
If I were using a standard buffered guage or display to read the results i may never have gotten this deep. Our recordings have about 36k data points for each test and the freq issue really starts to show.
In my opinion high freq plus dither may not be the best option at all as it creates a high level of "harshness" i suppose you would call it. On some systems it is not as noticeable on on others it makes a huge difference.
I have come up against a problem with the 4000Hz setting that when the pump rpm and its frequency lines up with the frequencies of the control it sends my relief completely unstable. I have to try freq alone next time as i ran out of time to keep changing setting.
I will write up more test results once i am on a computer and not a phone.
I fknd it interesting that a lot of info out there is pushing the high freq plus dither as the best option. It has led me on a big adventure that is for sure.
On the varying frequency while running, i started at 100Hz and moved to 350Hz at 50% duty cycle, on a scope it looks really good electrically except for the contrl turning on and off the output during step changes. I think if it were possible to do in a live system it would give better results than both the current options, my opinion only.

[FluidPowerTom]

What valve are you using?  What function is this proportional relief valve performing?  Is the intention for it to just jump between a few setpoints, or is it's setting varying proportionally with some system parameter?  Is it being used in a closed loop (pressure transmitter as feedback and PI loop)?

The pump RPM having anything to do with the signal frequency to the valve seems highly suspect.  What type of pump is it?  I'm guessing it's a gear, vane, or piston pump.  I'm not a pump expert, but I know some pumps do produce small pressure ripples as they turn.  However, those ripples aren't necessarily occurring at the pump's RPM because, I think, in one pump shaft rotation will expel flow from several vanes/pistons/gear teeth.

[oiltronic]

If open-loop, have you tried creating a non-linear profile (using the profile function) to translate the desired relief setting to PWM output current?  How was the relief response not correct when using a fixed frequency?

[SJ Industries]

Sorry for the lack of responses guys, I have been flat out. I did type up a big reply and lost it so here is the short version, I may have to be a little vague on some aspects due to confidentiality agreement so I cant put down everything.

It is a main relief with an external vent with a proportional controlled SUN valve controlling its max pressure setting.

I have since carried out some more tests on a different system with different pumps and did not experience the issues I have on the pump. Both pumps are variable displacement piston pumps. On my latest checks I did both config 4 and 5. This not closed loop as far as there is no PID or pressure sensor, simply the current feed back built in to the controller.
 
Config 5. worked much better on the different pump, it still displays some harshness to the logged results, the hysteresis is higher overall than just fixed freq, which makes sense as the two frequencies are added together, the amplitude has to be kept at 50mA otherwise it has very high hysteresis when the pressure is kept at a set point. It also appears more jerky than fixed freq when changing pressures.

Config 4 on the different pump was good, Sun recommend 140hz but I found 160 to 180hz the best, not sure but I think it is due to the fact the main relief is amplifying the proportional frequency? I left it at 160Hz on that project.

I haven't had a chance to look at the original pump with the problem with fixed freq config 4. I will set it up for the next time at 160Hz and see. 

Tom, I know the pressure pulses from the pump combining sounds a bit crazy but it is definitely there, if you graph the 2 freqs of 4000hz and what ever dither and have a look at the combined wave forms there are some really bad combinations, if you then overlay the freq of pump pulses it gets much worse at certain RPM ranges. The weird thing is I can see the same patterns in the logged recordings. I can see each pulse from the 9 pistons every 5milliseconds on the recordings. it appears as this pump arrangement with its fully mechanical regulator is very susceptible to harmonics, especially if they are high hour units with wear in the internal regulator linkages, the same settings for this set up on the other style piston pump and it is fine?

Oiltronic, I am not sure what you mean by the first part? If that is the exponential profiles in the work function library I haven't tried them because I haven't paid for that part of the licence. I think they may be useful for a future project. As I said above the fixed freq is fine on the other pump, actually better I feel, I will try it on the original pump soon.

https://www.desmos.com/calculator/mp7qmv1igx

If you look at the link to the graph calculator you can see what I mean about the combined frequencies, press play on column 9 (simulates change of RPM) then slide "c" on column 6 to 4000, you can see this is similar to what I see on the data logger. 
 

[FluidPowerTom]

Thanks for the reply; that does shed some light.  The 140 Hz that Sun recommends is really just a starting point.  Some systems work best down as low as 80-100 Hz, and indeed some work best up at 200 Hz.

Is your dither amplitude at 50 mA?  That seems extremely high.  I always run it at 5 or 10 mA.  I'd strongly recommend dialing that down and seeing how things go.

[SJ Industries]

The amplitude always seemed higher than it should be but i was going off of recomendations, initially I didnt question it, then I tried going higher as I have seen people using 100mA?
I will definatley give it a go and let you know how it responds. You havent seen any performance/response issues to changes in signal with having that level of amplitude?

[FluidPowerTom]

I've never used dither amplitude anywhere near that high.  I may have had one application where we went up around 20mA, but for all other applications I do 5 mA or 10 mA dither amplitude.  I remember on that application it was a PVG, and you could even see the handle on the valve vibrating pretty heavily. 

The purpose of dither is to keep the valve armature moving for reasons of kinetic friction vs static friction, so dither only needs to induce a slight vibration.

[SJ Industries]

Well Tom, that fixed the majority of the problem! tried 10mA dither and it was much better. It must be something to do with the mechanical feed back on the HP control on the pump regulator that makes it extremely sensitive to the various frequencies, cycles seen in the high pressure circuit can be seen in a separately supplied pilot control circuit, feeds back through the main pump swash plate servo. I had questioned the 50mA dither suggestion and was told that it definitely needs to be set that high?  Led me on an adventure.

[FluidPowerTom]

That's good to hear!  I'm wondering if the sensitivity to those frequencies has gone away now that you've got that lower dither amplitude.

[SJ Industries]

I have checked the pressure recordings, the sensitivity is reduced to a tolerable amount but the effect is still there. It has reduce to a 3BAR cycling to the pressure trace @300BAR.  This is down from almost 10BAR. Also the hysteresis is approx 30BAR @ 300BAR system pressure down from almost 60.
On a different pump design i am getting only 9 BAR Hysteresis.

I tested different dither frequencies as well. The most stable options are whole numbers that divide equally into 4000.
Example. 160Hz 200Hz 125Hz

If these 2 freq are stable together the system seems less sensitive to any other introduced harmonics. A 3rd freq that goes through like i had with that pump feedback pressure and the 2 conflicting freq (4000 + 140) really goes funny in the right conditions.

I am not sure what the effect would be outaide of my testing as i really have an isolated system with no other system interaction but it is definatley something i will be considering on all my set ups from now on.