Random thoughts...
Only one end of a shielded cable is typically grounded to avoid ground loops. The AC neutral is eventually tied to earth-ground at the star-point of the transformer and then you have the machine's PE (protective earth)....possibility of ground loop?
In my case I never have shield connected at both ends.
I do not have Neutral connected to GND at the power supply. Neutral connects to the - side of the transformer. I can easily tie them together, but it's odd this has never presented an issue and is only blowing up the LS7366. But if it's best practice to tie neutral and GND at the power supply that is easy to do.
Ah, sorry, miscommunication...I wasn't suggesting that you'd grounded both ends of your shielded cables, that was meant as an example of standard practice to avoid ground loops. For me, the 12v supply is the prime suspect here. :nerd:
Can't tell from the pictures, is there anything in your circuit to prevent reverse current flow when the supply changes over during power fail and restore? (Assuming the whole chip is killed and not just a / some pins) As to the extra counts first mentioned, random noise might give different values each time. Are the quadrature signals filtered at the input to eliminate any power line or motor switching noise picked up in the 10' signal cables prior to the 7336 input pins?
Oh, does this happen on one or multiple sites? Residential (US) systems require exactly one ground/neutral bonding point in the main panel. Not sure what your sites are like, maybe one is different from another or your bench system.
Headed out to LA now to see the system that broke. Taking a logic analyzer so I can get a count from the analyzer and compare with the counter and see what’s going on.
Frank, the encoder and counter only run on battery. There is no change over. Random noise should not accumulate in one direction only but somehow it is.
Ok, missed the battery only for the click and encoder. Went back to your drawing. Noticed the prop and the click board are both on phs A and B of the encoder. If the prop still tracks the encoder, does its count agree with the click board or could the prop be doing something with these pins that would appear as pulses to the click?
Prop is not connected to encoder a/b. Only a MAX 7311 IO expander connected to Prop. No counting. Helps to be able to diagnose if pins toggle to trouble shoot all inputs in field.
No. The motor is turned off while they are testing by hand. Freewheel. Clutch disengaged from motor. No noise from motor or clutch PWM.
Holy moly, they can make it happen with no power to the motor. Presumably this means that it's not a long term drift and that you could disconnect the charger altogether and do the same test.
No. The motor is turned off while they are testing by hand. Freewheel. Clutch disengaged from motor. No noise from motor or clutch PWM.
Holy moly, they can make it happen with no power to the motor. Presumably this means that it's not a long term drift and that you could disconnect the charger altogether and do the same test.
I've got lost here, is this now a replaced Encoder, or are these tests on the suspected damaged unit ?
Here is the installed shaft encoder. Obviously some issues here.
What that a mechanical pause (tho the rate seems stable?)
If it is not at a pause, that's unexpected, but those lone pulses should cancel, with a ++/-- pair ?
I think this is solved. The shaft encoder is bad. As shown. The other encoder test (mounted on rear of motor) that tested to accumulate as well really wasn’t. I like to run the Clutch at 50% when idle and I was slamming the motor back and forth against a hard stop by hand causing the clutch to slip when the motor came to an abrupt stop. I set the clutch to 100 and now the motor mounted encoder is consistent. Thanks guys for the suggestions! This particular issue is handled. Who knows about the bench issues.
Just curious, but does your design have a software or hardware limit switch for an e-stop function to avoid hitting an end stop? Guess virtual troubleshooting, kibitzing or whatever makes good mental excercise. Besides, it's the suburban substitute for the old sidewalk supervisors seen looking through cutouts in the fence around hi-rise projects.......
Here is the installed shaft encoder. Obviously some issues here.
What that a mechanical pause (tho the rate seems stable?)
If it is not at a pause, that's unexpected, but those lone pulses should cancel, with a ++/-- pair ?
The pulses only happen during shaft rotation. A pause would be two flat lines at whatever state. The two streams are 90 degrees out of phase (count/direction).
That particular condition of one channel flat-lining can happen when you have a fractured conductor in a moving conduit/cable chain. You get a make/break during movement of the axis.
Comments
Only one end of a shielded cable is typically grounded to avoid ground loops. The AC neutral is eventually tied to earth-ground at the star-point of the transformer and then you have the machine's PE (protective earth)....possibility of ground loop?
I do not have Neutral connected to GND at the power supply. Neutral connects to the - side of the transformer. I can easily tie them together, but it's odd this has never presented an issue and is only blowing up the LS7366. But if it's best practice to tie neutral and GND at the power supply that is easy to do.
Oh, does this happen on one or multiple sites? Residential (US) systems require exactly one ground/neutral bonding point in the main panel. Not sure what your sites are like, maybe one is different from another or your bench system.
Frank, the encoder and counter only run on battery. There is no change over. Random noise should not accumulate in one direction only but somehow it is.
Holy moly, they can make it happen with no power to the motor. Presumably this means that it's not a long term drift and that you could disconnect the charger altogether and do the same test.
I've got lost here, is this now a replaced Encoder, or are these tests on the suspected damaged unit ?
I believe that the 12v charger is not suited for this application because its -ve is tied to the AC neutral.... just feels wrong.
Coincidental with adding the 7366 is the addition of this charger.
The fact that other components haven't been damaged, well that's nothing that we haven't seen before.
Motor power issues have just been ruled out.
If they can lose/gain counts by hand rotating, then the battery alone should be able to prove the point.
Has this unbalance/creep issue always been there, or is this new ? (after some stress event ?)
by appx how much, and is it the same each time ? eg if you attach vice grips or similar to get a known up/down span, is it consistent.
In theory any noise impulse on quad encoders should self-cancel, with an inc/dec pair, but maybe their filters are not perfectly symmetric ?
This is quite bizarre.
What that a mechanical pause (tho the rate seems stable?)
If it is not at a pause, that's unexpected, but those lone pulses should cancel, with a ++/-- pair ?
Edit: Oh wait, that's the "bad" encoder.
The pulses only happen during shaft rotation. A pause would be two flat lines at whatever state. The two streams are 90 degrees out of phase (count/direction).
That particular condition of one channel flat-lining can happen when you have a fractured conductor in a moving conduit/cable chain. You get a make/break during movement of the axis.