Strange Motor behavior causing COM's to crash

[Beau Schwabe]

We have a setup where a 5V Serial com controls a standard 48V DC H-Bridge brushed motor .... the same Serial com controls a couple of Stepper motors.
We can send signals to the Steppers all day long and tell the steppers what to do, but as soon as we send a signal to the H-Bridge brushed motor, the motor controller receives the command just fine and continues with the last received command, but kills the com line so that no further commands can be received by the H-Bridge or the Steppers.

Last night we connected a completely separate power supply directly to the brushed motor (no motor controller, just power to the motor) a good 2ft away from the controller and if the power connection is steady we have Serial com. If we intermittently apply power (just a manual make/break connection), we loose Serial com. So because it works in steady mode, it does not appear that the brushes of the motor arcing are causing a noise issue.

Now for the really strange part .... if we connect ONE terminal of the motor (the other is left open) to the H-Bridge controller and send a signal to the motor, it kills the serial com. THIS is just a capacitive load like driving a antenna. The motor contacts are isolated from the motor case, so there is no shorting going on there, the resistance of the motor terminals are about 1.6 Ohms and the motor works fine on the lab bench with the H-Bridge controller.


We have tried all sorts of filtering techniques: (The motor is a 48V DC brushed motor)
1) Delta cap configuration on the motor with 100nF caps ... i.e. across motor terminals, and then from each motor terminal leg to ground
2) Separate power supply to the H-Bridge
3) We have shielding on all of our serial com lines ... the greatest distance of a serial com line is about 2 feet

4) Another thing ... right now we are using 5V RS232, but we have tried a RS485 configuration as well with the same results. There appears to be no behavioral difference between RS232 or RS485.

The only thing we are not doing is shielding the H-Bridge controller which we plan on testing today.

Any thoughts would be greatly appreciated

Thanks

-Beau

 

[MaxHeadRoom78]

Sounds to me you may have earth leakage somewhere, motor etc, if the system is not properly earth ground bonded, you could be getting all kinds of ground currents, especially if it occurs with one motor lead connected
You should have equi-potential earth bonding of all parts of the system, resulting in a star ground point which includes the service GND.
Max.

 

[Reloadron]

Depending on how things go you may want to consider a duplex opto-isolated RS232 interface if the setup allows. Would also be nice using a scope if you could actually look at your data lines for any glitches or spikes.

Ron

 

[Beau Schwabe]

Turns out it was a shielding problem to electrical ground on one of the the COM lines that was missed. making the shield act ore like an antenna.
Still very strange behavior.

 

[MaxHeadRoom78]

Still very strange behavior.

EMI is not uncommon if not precautions as per #2 are taken!
In the CNC forums, there are those that often report spurious apparent tripping of limit switches caused by the same type of EMI.
Max.