Novice. I am controlling a Delta B2 24vdc Servo drive with a 5vdc controller through a 2N7000 transistor(s) All part of a much larger picture inside a control of a large CNC router. The attached diagram is as built from what I thought would work but during operations the servo drive is receiving uncommanded pulses. The most recent example was after 6 positioning moves, when it went back to zero the pulse count should have been zero but was 388 which is considerable, and counts for almost a full half turn. Also to note, when the other motors were working, with this one stationary, I could see the pulse count rising by about 2 per minute. So its gaining pulses from somewhere.

Everything is shielded and I have ferrite cores on the conductors. The machine is earthed to the nana.

I have some PC817 optocouplers here in my workshop I was wondering if they would be a solution? Or, is there a way I can clean this circuit up to remove the non commanded pulses? Like I say, novice with enough knowledge to do damage ha

 

My first guess would be lack of a "Star-Grounding" scheme,
an/or excessively long distances between units.
.
.
.

 

Novice. I am controlling a Delta B2 24vdc Servo drive with a 5vdc controller through a 2N7000 transistor(s) All part of a much larger picture inside a control of a large CNC router. The attached diagram is as built from what I thought would work but during operations the servo drive is receiving uncommanded pulses. The most recent example was after 6 positioning moves, when it went back to zero the pulse count should have been zero but was 388 which is considerable, and counts for almost a full half turn. Also to note, when the other motors were working, with this one stationary, I could see the pulse count rising by about 2 per minute. So its gaining pulses from somewhere.

Everything is shielded and I have ferrite cores on the conductors. The machine is earthed to the nana.

I have some PC817 optocouplers here in my workshop I was wondering if they would be a solution? Or, is there a way I can clean this circuit up to remove the non commanded pulses? Like I say, novice with enough knowledge to do damage ha

Do you have a 24V pull-up voltage on the drains of your 2N7000 MOSFET's, which you would get if you link pins 17 and 35 on the Delta controller? (See page 3-28, drawing C3-1).
What exactly is your "Sig Generator". Maybe your MOSFET's need pull-down resistors on the gates?

 

Hi yes I do. I was unable to fix this, and noticed I am getting noise from other portions of the control circuit. I have the grounds of a 5v and 24v tied together, and noticed the step count on the Delta drive move even when I hit the enable. I have moved away from the 2N2222 circuit and installed a couple of PC817 opto couplers which gave a slow signal, but perfect position. I am going to use this whilst I have the parts to make a line driver.

 

I see no pull-ups on the collectors of the opto devices, so that might be part of the problem.
The big question is which are the long exposed wires for the pulse and direction signals, That is not clear from the drawing.
One possible source of grief is in the connections of shielding to "ground". The problems come because not all "grounds" are equal. Connecting the shield to the wrong point will introduce a lot of the wrong signals is the thing I discovered years ago.

 

I second post #2, normally a star grounding is set up that includes all power commons and ensure the service earth is also connected.
Remember the PC signals into the BOB are also ground referenced in tower/table type PC's.

 

The "Star" arrangement is the choice for power distribution, no question there.
For signal connections there is an additional guide, which is that ONLY the signal current should flow in ALL of the signal loop. With that is the requirement that there exists an actual LOOP that contains only the signal current. The deadly "Ground Loop" is the section where another current, often a power return current, uses the same conductor as the signal loop..

 

The term Ground-loop also describes a one or more earth ground points where a potential exists between any two grounded points, one reason for equi-potential bonding as outlined in the Siemens publication.

 

The 2N7000 is a 5 ohm NFET and needs about 1k ohm pullup. The wires should be STP cables with earthed shield at one end only.

 

I went over to the latest view some time ago for servo installations, When equi-potential bonding is carried out, the preferred method of grounding the shield each end can be done (see ch6 of the Siemens man.)
I also use one of the best cable versions out there when working with servo's and CNC. Belden 9891. but costly!
https://www.techtarget.com/searchnetworking/definition/shielded-twisted-pair

 

The term Ground-loop also describes a one or more earth ground points where a potential exists between any two grounded points, one reason for equi-potential bonding as outlined in the Siemens publication.

Certainly if a voltage is present between two connected points then there must be a current flow between those two points. And if that current is flowing because of any other reason than it is the signal current, then the situation is exactly what I described. Max has described the same thing from a different point of view. Good Job, MAX.

 

Intermittent errors are difficult to analyze without lots of measurements and details of same.
e.g. grounding STP at both ends may create faults when low Z grounds are not 0 V due to ground filter currents. But in other situations when ground noise is a higher Z and higher frequency then PE shielding may be earthed at both ends. There are many articles on this.

USB faults are very common when say servo controller has a different SMPS even floating to PE than the computer with PE gnd due to switcher transformer C coupled noise. Then good PE gnds are necessary for all peripherals.

 

I would NEVER use a USB connection in an application where a communications failure could cause damage or a serious malfunction.

 

I would never use a SMPS for servo's, even the manuf. recommend against using such.

If a communication failure can cause severe problems, then extra steps (software/hardware etc) should be taken in order for it to be prevented.