Using a Lynx 6 robot for a light industrial project (very light) changed out standard servos (HS-475HB) with new HSR-5990TG motors. I am experiencing motor jitter. I did in fact eliminate the standard motor power supply and powered up the controller using separate supplies. The servo motors use a 6-6.5 VDC supply capable of 3.5A. I still experience motor jitter at rest and during moves. Do I need a still bigger supply with these motors to smooth out motion? What does anyone suggest? 
I am also experiencing this jitter. I would like to know what it is caused by, in case anyone has thought about this over the year.
I suspect that the jitter may be due to supplying the servos with a voltage that is on the high side of their normal voltage range.
I donāt think itās oversupplying voltage, as their range is 6V-7.4V. I try to keep the voltage right at 6 because I have other servos in the mix, but it often drops down to 5.8. I have also tried about 6.5V, and the jitter has the same characteristics. Itās not constant jitter, it depends on the load. Itās more of an occasional twitching.
One thing I can think of is the whole HMI thingā¦ somebody in the forums mentioned that HMI requests interrupt regular servo operation. Iām not doing any HMI stuff with it, and the character of the twitching changes depending on the position of the arm (so the load changes).
If it would help, I can try to get some video of it tomorrow.
If youāre powering the SSC-32ās VL (logic) from the VS (servos) supply, the SSC-32 may actually be resetting at times when the load is greater. Because the SSC-32 is constantly receiving position commands while in motion, if it were to reset, the servos would briefly go limp, then return to holding position. This would happen very quickly and be perceived as jitter. If this is what is happening then the solution is simple. Add the 9vdc battery clip to the VL terminal red +, black -, and remove the VS1 = VL jumper. 8)
If the jittering is pretty much constant, then you may want to see if you have the capability to lower the gain on the amp in the servo.
Iāll have to double check, but I think when I was programming the endpoints of the rotation on the servo, I lowered the gain to the lowest available.
Also, I am powering the servos on batteries independent from the logic and have removed the jumper.
Great ideas, and thanks for responding, but I donāt think weāve hit it yet.
Ok, I have finally recorded some video of the jitter that I am experiencing. The 5990 is the āelbowā joint servo, and you can see the jitter as the arm bounces when itās supposed to be sitting still, or if you have good hearing, there is a high-pitched whine that briefly cuts out whenever the jitter occurs.
One other thing in my implementation that I have forgotten to mention (Iāll have to test some things) is that I am querying the SSC-32 servo position for all of the 6 servo channels at 20 Hz. Could that be causing the jitter?
I just tried turning the sampling rate down to 2 Hz, and that didnāt seem to affect anything.
Anyway, hereās the link to the video:
lynxArm5990Jitter.wmv
3.7 MB
It looks more like an oscillation than jitter. Do you have some springs that you can add on the rear of the arm? It may prevent it from happening. You may want to look at modifying the servo hinge part of the arm to see if you can eliminate the ball bearing. Iām currently experimenting with adding mechanical dampening to prevent these oscillations as we design stronger arms.
Also, Iāve seen oscillations before, but not with such a strong servoā¦
So, this oscillation happens even when the arm is just sitting still?
I can see how that would beā¦ correct my reasoning here.
The servo has reached its commanded position, according to the pot, so nothing is happening. Gravity acts on the arm and puts a torque on the servo head. This ever so slightly moves the pot and the servo moves to correct the error, and then maybe overshoots just a little and stops. Rinse and repeat.
Is that the idea? If so, I will definitely try playing with the servo gains, as I do have the HMI programmer cable. I donāt have any springs, and Iām worried about removing the bearing, as binding at that joint is what fried the previous occupant of that location. I suppose as long as Iām extra careful it should be alright.
In the end, it doesnāt bother me too muchā¦ itās just a curiosity, but I wouldnāt complain if I could make it disappear.
You have it exactly.
Well, adjusting the gains changes the characteristics a little, but doesnāt get rid of it. I was just using the windows GUI tool to change them, and they offer 3 settings for the D gain and 3 settings for the P gain.
Do the servos support more settings for the gains (and the GUI doesnāt) or is that all I can do?
I donāt know the answer to this. Sorryā¦