BOB's input interference, help please!

[Walky]

Hi

I'm having a hard time with my last cnc router. I just finished building it and just needed to wire the limit switches to the breakout box, which is optoisolated. Today I did just that and found myself in a horrible situation.

I think I'd better make a list of facts, so it's easier to know my situation:


- Input pins 10, 11, 12 and 13 are optoisolated.

- Input pin 15 is not optoisolated and is wired directly to que parallel port.

- The isolated and not isolated parts of the BOB have different grounds.

- With the steppers connected (doesn't matter if 1 or all 3), the isolated inputs have constant spikes that won't even let me move the steppers.

- With the steppers disconnected, the isolated inputs work fine.

- With the steppers connected, the non isolated input works perfectly if I connect a limit switch into it (the same switch that had spikes previously).

-I tried joining different grounds (not only the BOB's) in different ways to no avail, the problem is still there.

-If I join the grounds from the isolated and not isolated sides of the board, with a switch connected to the isolated side and another one connected to the non isolated side, the isolated one still have the same frequent spikes, while the non isolated one (which, as I previously said, usually works fine) starts having spikes, but not so frequently as the other switch.

-If I disconnect the USB plug (used only as a 5V supply to power the optoisolators), the interference stops and all the switches from the isolated side work perfectly, but I can't move the steppers correctly (they can move, but quite inconsistently since the optoisolators are not working with the proper 5V; curiously, they seem to be getting power from the DIR signals under this circumstances). In other worlds, the optoisolators seem to handle the limit switches fine when they aren't supposed to, but seem to be unable to do so under normal circumstances.

-If I use the BOB's integrated power regulator (an alternative to the usb power) I get the same spikes.

-All the switches I mentioned are installed in the machine and are wired near the steppers' cables side by side. They are all wired NC. I really, really don't want to replace the wiring since it took me a lot of time to finish it (it goes through holes and cable carriers), looks really clean, I'm very happy with the soldering job I did at the XLR connectors and I'm not even sure if that would fix the problem at all.

-If I wire an external switch to the isolated side, the spikes stop.




All those points make me think that there is probably some interference from the steppers' cables, but at a level that the parallel port can handle directly but the BOB's isolated side can't (the switches working when the optoisolators are not properly powered up makes things even weirder). I don't know if the optoisolators can't "copy" the signal accurately enough when the spikes are present, or if the problem lies in the BOB's way of handling ground.


I read someone suggested using 0.01uF capacitors parallel to each limit switch (between each input and ground) to filter the signal, I'll try that on monday if I still can't figure out what's going on by then.

The other option I can think of would be to build a relay board so I can drive the switches at 24V, but the 24V would be from the same psu I'm using to power the isolated side of the breakout board, would the interference would still be there?

This is my BOB:
http://www.probotix.com/manuals/PBX-RF_manual.pdf




Does anyone have any theory about what could be probably be happening?


I don't join all the switches and the e-stop to the only usable input since I would be left without hardware homing and touch probe. I don't want to use a "Y" parallel cable (or even a second parallel port) to feed the inputs directly, since the control box is really nice (made on my other CNC) with the limit's connector included in it and even "Limits/E-Stop" v-carved over it; it would just not look good after the 3 months I took designing and building this machine.

Thanks in advance!

[neilw20]

Take some photos so we can see exactly where your wiring, including power supply wire come and go to.

It is not sufficient just to connect grounds together, wherever they are.
This needs some planning, and what you have is symptomatic of some wrongly shared ground leads.

[Al_The_Man]

Is your PC power supply 5v common connected to ground? Usually this is done internally by ground buss screws.
By ground, I mean earth ground.
Using the term ground for various commons can get confusing.
If so I would take a Ground wire from PP common to the central earth ground plate.
Make sure the service ground is also connected to the central plate.
Also ensure that all metalic parts of the machine and motors are bonded back to the Earth Plate.
See this past post also Serious noise issues
Al.

[Walky]

Is your PC power supply 5v common connected to ground? Usually this is done internally by ground buss screws.
By ground, I mean earth ground.
Using the term ground for various commons can get confusing.
If so I would take a Ground wire from PP common to the central earth ground plate.
Make sure the service ground is also connected to the central plate.
Also ensure that all metalic parts of the machine and motors are bonded back to the Earth Plate.
See this past post also Serious noise issues
Al.

I checked the grounding and it seems fine (PP- 5Vgnd, PP-earth ground, 5Vgnd-earth ground). I noticed the BOB's PP case is not connected to the non-isolated side's ground; y tried connecting it to the isolated and non isolated grounds but it didn't make a difference. Maybe I should modify the BOB to bypass the opto and feed the inputs directly to the PP, but I don't want to screw up and kill the board, since I've already spent too much in this project.

The machine uses awg22 wire for the steppers and awg26 for the limit switches and e-stop, they run parallel all the way (the router is wired independently to make it easier to replace). I know I should have used shielded wire, but it was the only flexible wire I could find locally that could handle the motor's current (actually, I had to modify the XLR connectors to make the wires fit in it). The design makes it difficult to add extra wires (it would be way too tight) to ground the motors and metallic parts (luckily, most of the machine is made of MDF).

I'll try to add some pictures soon.

[Al_The_Man]

The design makes it difficult to add extra wires (it would be way too tight) to ground the motors and metallic parts (luckily, most of the machine is made of MDF).

I'll try to add some pictures soon.

Actually that is a case where it might be more important to add ground bonding, as with MDF, motor frames can become isolated from earth ground much easier.
Al.

[Walky]

Actually that is a case where it might be more important to add ground bonding, as with MDF, motor frames can become isolated from earth ground much easier.
Al.

Sorry, my mistake. I though of the metal structure as a big antenna and forgot it would be grounded.

BTW, I grounded the motors cases to the AC ground, and I'm still getting spikes

It seems I'll probably have to replace the wiring (or try the relay board's approach). Would shielded 24awg be enough for my 6-wire (originally 8-wire) steppers at 41V, ~2.5A each?

Do you think there's a chance that it won't fix my problem?

[neilw20]

The wrong termination of ANY of the power wiring AND powers supply components can cause this fault.
As I said, take some pictures so we can assist.
Using a guessing game, has not much chance of resolving your problem.
A connecting wire as short as 1" in the wrong place, especially in the power supply low voltage terminations can cause this.

[Walky]

Here are some pictures (the inputs' connector doesn't appear since I took the pictures after some experimenting).

http://img268.imageshack.us/img268/6...6784grande.jpg
http://img833.imageshack.us/img833/8...6778grande.jpg
http://img834.imageshack.us/img834/5...6779grande.jpg
http://img830.imageshack.us/img830/4...6780grande.jpg
http://img833.imageshack.us/img833/3...6781grande.jpg
http://img834.imageshack.us/img834/7...6782grande.jpg

The psu at the left is 24V and powers the fans (two pairs in series) and the breakout board. The other psu is 36V running at 41V by the built-in potentiometer, specifically for the steppers.

[neilw20]

I see the switchmode supply has is one of the ones that has 3 negative and 3 positive terminals.
Try not to use common wires, like for the negative and positive.
Reason: Each of the drivers can be drawing different currents at any one time, and the voltage drop on the common wires changes all the time.

From each of the positive power terminals, run 3 red wires to each fuse individually.
I see you have a common negative bus. Good try, but the white wire to the supply negative will have varying voltage drops.

Run each of the negatives from the driver straight to the supply negatives, and 1 wire to the ground common busbar.

Return single wire grounds all back to the bus bar.
Don't daisy chain ANY wires.

Your limit switch wiring not in the photos. Are they connected yet?
You can run all of your limit wires in a shielded cable, like an old printer cable.
The shield on your cable should ONLY connect to the busbar.
The other end of the shield should not be terminated at the machine.

From the small supply run individual pairs of wires to where they are needed.

Keep the mains wiring separately routed away from low voltage wires.
Try to rout the stepper wires away from the limit switch wiring.

Run a single wire from the common negative point to the busbar, if you need one. Try putting a 100 ohm resistor in series with this wire.
It should run COLD, with no voltage drop across it. If not that should lead you to where the fault currents might be going.

I assume the fifth wire from each stepper is an earth.
Connect each to the busbar.

The whole strategy here is that any voltage drops in the wires do not interfere with another circuit.

Allow the router earth to be done through it's own cable, or is it double insulated? If turning it on and off does not cause resets etc, that's OK.
Beware that you may need an earth from the router, to the nearest stepper will probably be OK, so that static electricity doesn't cause glitches.

You might earth the guide rails too.

Small voltage drops in wires can be quite significant compared to the low voltages used in the limit switch circuits.
The spikes quite often can be larger than the signals, and are very hard to capture. The above rules makes mostly them irrelevant.

When wires are run to limit switches in pairs, and not commoned anywhere in the machine, then any interference introduced on one wire appears on the other too, and they subtract from each other (call common mode) and interference cancels!

One hidden source of trouble is currents running back to the PC through the 0v common. This will cause havoc, as evidenced by unplugging the USB.

You need to make sure currents don't go through your signal commons.

Machine looks nice. I hope this helps.
As you are finding, disconnecting circuits can make the problem move, but while currents (including interference spikes) flow through common signal wires you will have problems.

[Walky]

Thank you very much for your advice, I'll rewire everything as you said.

My limit switches wiring appears in the pictures as a bunch of wires next to the steppers' ones, coming out of the back of the machine (with a DE9 at the end of it, and a miniplug intended to use with a touch probe). The controller's part of this connection is missing in the pictures, but it's just a male DE9 with some short wires going straight into the BOB's input (I removed it meanwhile so the cables don't stay in the way while testing the inputs).

Run a single wire from the common negative point to the busbar, if you need one. Try putting a 100 ohm resistor in series with this wire.

What do you mean by "common negative point"?


I made a little test a couple of minutes ago:
I just removed the Z axis switches' wires so they don't run near the stepper wires and then connected them as a single input (from the bob directly to the Z assembly): the spikes are still there! I even tried disabling the Z motor and it made no difference. This pair of cables is originally joined together by heatshrink every 5 inches or so, but not twisted.

It seems that the problem might indeed be in the controller box or its surroundings; I sort of hope so, since rewiring the whole machine would be a horrible task.

[Walky]

I've just found this:

E-Stop, Probotix PBX-RF, and Mach3

It seems I'm not alone in this, go figure.

I sent a mail to Probotix, I hope they can help me with it (their customer service have been great so far).

BTW, I just noticed the isolators ICs on the board are not optical but RF-based.

[neilw20]

Thank you very much for your advice, I'll rewire everything as you said.

My limit switches wiring appears in the pictures as a bunch of wires next to the steppers' ones, coming out of the back of the machine (with a DB9 at the end of it, and a miniplug intended to use with a touch probe). The controller's part of this connection is missing in the pictures, but it's just a male DB9 with some short wires going straight into the BOB's input (I removed it meanwhile so the cables don't stay in the way while testing the inputs).



What do you mean by "common negative point"?
The negative of the 12v power supply.


I made a little test a couple of minutes ago:
I just removed the Z axis switches' wires so they don't run near the stepper wires and then connected them as a single input (from the bob directly to the Z assembly): the spikes are still there!
How are you measuring the spikes? Relative to which point?
I even tried disabling the Z motor and it made no difference. This pair of cables is originally joined together by heatshrink every 5 inches or so, but not twisted.

Disconnect the wires at the BOB and if the spike problem is still there then the noise is coming in through the supply leads.

With multiple circuits it is often easiest to get each circuit going one at a time.

It seems that the problem might indeed be in the controller box or its surroundings; I sort of hope so, since rewiring the whole machine would be a horrible task.

If you haven't changed the power wires to the drivers, then nothing will have changed.