LabVIEW
Would like to hear opinions about the capabilites of a mechanical system
@Ben wrote:
Would anyone care to offer any suggestion on how to go about tuning two PIDs each running one of the hydraulic servo valves?
I can hear our process engineer say: "PID controllers are the worst kind of controller you can come up with. They're very general applicable, but terrible at what they do".
It would be better to design a custom MIMO system, but that requires process engineering knowledge (I tried, but not my cup of tea).
Synchronizing two parallel PID's would almost certainly involve hacking and tweaking 'until it works'.
Search LabVIEW like a graph! 
Application Story
We were testing helicopter transmissions under extreme load.
Strain gauges were carefully hand laminated to the gear teeth and a giant bundle of wires snaked out to the DAQ hardware.
A big motor connected to a huge reducer would drive the transmission against an air brake. The brake would go from 0 to around 20 kNm, but it had to ramp and hold steady as quick as possible because the wires pulled in as the gear turned. Only one rotation was possible.
Not only was this a major programming challenge, but all the DUTs were basically priceless prototypes.
Fortunately, LabVIEW for the win on this one. Setting up the delayed PID I needed to manage delay in the operating the air brake would have been pretty much impossible otherwise.
The only near disasters we had were once when brake was run "dry", which made it extra grabby. Fortunately the emergency stop caught it in time and nothing was damaged.
The one that made us all jump was when someone forgot to bolt the brake back down. It was pretty heavy, but not 20 kNm heavy. Fortunately I caught the movement out of the corner of my (good) eye and mashed one of the big red buttons before something really bad happened.
The durability test on this one was fun too. Drain all the oil and spin the transmission way past redline.
Had some hydraulic lines blow during these tests, but the DUT never did explode, even though there were precautions for just that. It sure made some crazy noises though. (...like how I imagine the soixante neuf drive would sound....extra points for anyone knowing that obscure reference )
Same disclaimer as Bob, of course 😄
I assume it's not you doing the PID, but that hardware has itself a PID incorporated which processes the control signal to get the hardware to the setpoint. As a first step, I would re-scale the two graphs for matching average amplitudes in order to better display the time-lag between them (currently, the black has a lower amplitude, causing the signals to cross). Ideally, you should be able to shift, scale, subtract so only the high frequency remains. The shift will tell you about the time lag. (Or just do a FFT and look at the phase difference of the dominant frequency)
The amplitude of the oscillations itself seems to be proportional to the amplitude of the control signal. They seem easily 4x larger near the peaks compared to the valleys. Do these oscillations remain if you keep the drive signal at a constant value? Do they disappear if the controls signal is zero?
Do you have some real data to play with?
