I'll start out with the solution and save the gory details to the end.

>>>>>

I was having some noise on the LVTTL inputs that I was running with 3v generated from the 5 v logic supply originating on the 5 v lines on the 50 pin interface connector off the SnapAmp.

 I was also using the opto inputs running on an external isolated 12 v and saw no nuisance noise problems with those inputs

I was using a Winford 50 pin ribbon terminal board which had space to install some filter caps on the input lines.  I installed  0.047 mfd disk caps on each LVTTL input to ground on the Winford board.

This helped immediately for a while then I saw the noise issue come back and it seemed to get worse as I ran (enabled) the SnapAmp longer.  The longer I left the amp disabled the longer I could leave it enabled  before the noise issue came back.

To solve input noise interference, the final solution was to install a ferrite tube 1.6 in long and 0.6 inch OD , 0.2 in ID. with both motor leads run through them plus an inductor in each leg of the motor (22mh each) between the ferrite and the motor.

It looked like the noise reduction offered by the ferrites alone would be enough to prevent noise triggering of the inputs but inductors had a significant added benefit.  Inductance of 5 or 10 mh each probably would have been sufficient but I didn't have that value available so it wasn't tested.  The other motor (axis2) already had an inductor in each arm lead  (? mh) so that motor didn't seem to need any added inductance because it didn't show any increase in noise when enabled.

The result was very quiet operation.  Both audible noise and electrical noise was dramatically reduced so there was no interference on the Inputs at all.  I left the input filter caps and resistors installed and didn't test without them.

The other benefit with the inductors was that the axis motor position offset was dramatically reduced from about 80 counts to about 0 to 2 counts on the axes with the inductors.

My motors are quite small using 0.1 amps to run no load and maybe 1.0 amp fully loaded.  I don't know what the motor inductance is but I'll check with the manufacturer. The motor leads are all about 2 feet long with the shield in the motor leads grounded to the chassis, (Not the 0v ).  I have wall power ground connected to the chassis.  Motor bus power is 24 VDC isolated from the 5 v supply.

 I was using a Winford 50 pin ribbon terminal board which had space to install some filter caps.

So I installed  0.047 mfd disk caps on each LVTTL input to ground on the Winford board.

This helped immediately for a while then I saw the noise issue come back and it seemed to get worse as I ran (enabled) the SnapAmp longer.  The longer I left the amp disabled the longer I could enable it before the noise issue came back.

The ribbon cable between the breakout board and the SnapAmp started out at 16 inches long but reducing it to 6 inches didn't help the noise issue noticeably.

I tried inserting 22K ohm resistors in series with the input leads that had the caps  but that didn't change anything.   My scope showed the noise was everywhere on the input circuitry.   

When  I installed a set of tubular ferrites ( 0.6 OD, 0.2 ID, 1.6 inch long )on the motor leads ( both motor leads through each ferrite on each axis) This significantly reduced the noise and I was able to see the individual noise spike generated from each axis on the scope.  I tried disabling / enabling the axes with Kmotion Axis screen and I saw that one axis was generating the noise that was interfering with the inputs.  

That axis spike eventually grew to more than 5 times the spike of the other two motors.  I tried swapping everything to narrow down the actual source.  Nothing I swapped changed the source of the spike and the spike grew with the length of time the axis was enabled.  It turns out that the spike was on Axis 1 of the system.  (SnapAmp output pins 2, 3.)  I was using SnapAmp outputs 0, 1;  2, 3;  6, 7. 

 I was also experiencing some significant position offset when the axes were holding position.  Axis 0, and 1 were showing position offsets of 50 to 80 counts with no load.  Enabling the integral gain reduced this somewhat but not below 30.  This offset only showed up on Axes 0 and 1,   Axis 2 had a different motor on it that had inductors on each motor lead near the motor.  I don't have the inductance rating of that motor.  The other motors were spec'd at 9.35 millihenry

I added two inductors on each motor 22 mh each between the ferrite and the Motor.  This had a significant effect of reducing the audible switching noise to near silence and the electrical noise in the system went down to less than 0.25 volt. and the postion offset was 0 to 2 counts.on all the motors.

Any ideas on why SnapAmp outputs 2 and 3 were the ones generating the most noise and why running longer increased the noise intensity?

AZ

Hi AZ,

Wow so many strange effects.

I assume these are small brush motors?

Its hard to imagine how filtering the inputs with an RC of 22K and 0.047us wouldn't stop any noise.  That would have a time constant of 1 millisecond.  Were those Ceramic capacitors?  Close to SnapAmp?  Grounded close to SnapAmp?

Strange there would be audible noise.  SnapAmp PWM's at 33KHz which should be inaudible.  The only thing I can think of is that in DC Servo mode a digital current loop is used running at 33KHz in the SnapAmp FPGA to regulate to the current to the specified level by varying the PWM.  If this control loop went unstable then that might result in an audible current oscillation.  I would expect that to be obvious in a current plot on the Step Response Screen.  There are programmable gain settings for the current loop.  But we've never known of a case where they needed adjustment.  But these may be highly abnormal motors so that might be something to try.

I can't think of a reason the noise would slowly grow.  The only thing I can think of would be temperature.  Do the motors run hot?  Motor resistance will change a small amount with temperature.  Resistance changes might change the current loop stability.

Having a constant position offset with any I Gain should be impossible.  The Integrator will continuously ramp the command current until the position offset drops to zero.  Again a current plot should make it obvious what is happening.

Sorry for no clear answers...

Regards

TK

 

I'll start out with the solution and save the gory details to the end.

>>>>>

I was having some noise on the LVTTL inputs that I was running with 3v generated from the 5 v logic supply originating on the 5 v lines on the 50 pin interface connector off the SnapAmp.

 I was also using the opto inputs running on an external isolated 12 v and saw no nuisance noise problems with those inputs

I was using a Winford 50 pin ribbon terminal board which had space to install some filter caps on the input lines.  I installed  0.047 mfd disk caps on each LVTTL input to ground on the Winford board.

This helped immediately for a while then I saw the noise issue come back and it seemed to get worse as I ran (enabled) the SnapAmp longer.  The longer I left the amp disabled the longer I could leave it enabled  before the noise issue came back.

To solve input noise interference, the final solution was to install a ferrite tube 1.6 in long and 0.6 inch OD , 0.2 in ID. with both motor leads run through them plus an inductor in each leg of the motor (22mh each) between the ferrite and the motor.

 

It looked like the noise reduction offered by the ferrites alone would be enough to prevent noise triggering of the inputs but inductors had a significant added benefit.  Inductance of 5 or 10 mh each probably would have been sufficient but I didn't have that value available so it wasn't tested.  The other motor (axis2) already had an inductor in each arm lead  (? mh) so that motor didn't seem to need any added inductance because it didn't show any increase in noise when enabled.

The result was very quiet operation.  Both audible noise and electrical noise was dramatically reduced so there was no interference on the Inputs at all.  I left the input filter caps and resistors installed and didn't test without them.

The other benefit with the inductors was that the axis motor position offset was dramatically reduced from about 80 counts to about 0 to 2 counts on the axes with the inductors.

My motors are quite small using 0.1 amps to run no load and maybe 1.0 amp fully loaded.  I don't know what the motor inductance is but I'll check with the manufacturer. The motor leads are all about 2 feet long with the shield in the motor leads grounded to the chassis, (Not the 0v ).  I have wall power ground connected to the chassis.  Motor bus power is 24 VDC isolated from the 5 v supply.

Gory details:>>

 The Motor bus was a 24 VDC supply isolated whose common was isolated from the 5 v supply common.  I also had  5000 mfd cap on the motor bus supply within 4 inches of the Snap Amp.

 I was using a Winford 50 pin ribbon terminal board which had space to install some filter caps.

So I installed  0.047 mfd disk caps on each LVTTL input to ground on the Winford board.

This helped immediately for a while then I saw the noise issue come back and it seemed to get worse as I ran (enabled) the SnapAmp longer.  The longer I left the amp disabled the longer I could enable it before the noise issue came back.

The ribbon cable between the breakout board and the SnapAmp started out at 16 inches long but reducing it to 6 inches didn't help the noise issue noticeably.

I tried inserting 22K ohm resistors in series with the input leads that had the caps  but that didn't change anything.   My scope showed the noise was everywhere on the input circuitry.   

When  I installed a set of tubular ferrites ( 0.6 OD, 0.2 ID, 1.6 inch long )on the motor leads ( both motor leads through each ferrite on each axis) This significantly reduced the noise and I was able to see the individual noise spike generated from each axis on the scope.  I tried disabling / enabling the axes with Kmotion Axis screen and I saw that one axis was generating the noise that was interfering with the inputs.  

That axis spike eventually grew to more than 5 times the spike of the other two motors.  I tried swapping everything to narrow down the actual source.  Nothing I swapped changed the source of the spike and the spike grew with the length of time the axis was enabled.  It turns out that the spike was on Axis 1 of the system.  (SnapAmp output pins 2, 3.)  I was using SnapAmp outputs 0, 1;  2, 3;  6, 7. 

 I was also experiencing some significant position offset when the axes were holding position.  Axis 0, and 1 were showing position offsets of 50 to 80 counts with no load.  Enabling the integral gain reduced this somewhat but not below 30.  This offset only showed up on Axes 0 and 1,   Axis 2 had a different motor on it that had inductors on each motor lead near the motor.  I don't have the inductance rating of that motor.  The other motors were spec'd at 9.35 millihenry

I added two inductors on each motor 22 mh each between the ferrite and the Motor.  This had a significant effect of reducing the audible switching noise to near silence and the electrical noise in the system went down to less than 0.25 volt. and the postion offset was 0 to 2 counts.on all the motors.

Any ideas on why SnapAmp outputs 2 and 3 were the ones generating the most noise and why running longer increased the noise intensity?

AZ