Limits/Noise problem

[RichardV48]

Guys, I need some help with a limits/ noise problem. I ran this by Art with Mach 3 and he suggested I check with you. I wired the X, Y, and Z limits in series rather than individually. Everything, limits and motors have shielded cable which is connected to a common ground. The limits work correctly as far as being able to manually operate a switch and get the led to light in Mach 3, a limit switch triggered indication and a restart indication. The problem is that about every 20 seconds the limit led will flicker and I get a reset condition indicating a limit switch was triggered. I assume this is a noise spike. When this spike or led flicker occurs the Xylotex controller is turned on and the motors are locked up, but nothing is being cut so they are not running. I disconnected all motors and powered up the controller and the flicker does not occur. I plugged in one motor, powered up the controller and the limit led flickers. This happens regardless of which single axis, X, Y, or Z is plugged in.
I started with a debounce setting of 2000 and worked up to 7000, which Art said is equal to 280ms in time and it has no effect on the problem. There is a debounce interval and a index debounce setting. How should each of these be set?
Art indicated a capacitor across the line may correct it but wasn’t sure what size to use.
I would like to find out what is causing the spike and correct it if possible. What is the best way to track this down?

[Torchhead]

Is this feeding directly into the parallel port or through some form of isolation? I assume the switches are wired in NC format and that when a switch opens it lets the pin go high and MACH is set to active High. If the limit LED flickers than I would suspect a bad element in the string or ground problems. It is, after all, clamping the input pin to ground and noise would have to be on the ground itself. Also make sure (with an ohmmeter) that there is no connection between the switches and their ground to the table. If you have grounded your table, power supply or other parts to AC safety and the table itself too then you might have a ground loop situation. The Xylotex is referenced to the same ground as the PC and the Motor DC is the same. With everything going to the same ground it presents an excellent opportunity for ground loops. I doubt a cap will get rid of it. If an input can go positive enough to turn on the input and it's clamped to ground the noise has to be coming FROM the ground!. The fact debounce has no effect on it means it's not noise spikes or even PWM noise. It's something with a much lower frequency component.

A ground loop is defined as having more than one ground for a signal to travel on in a system. Worse grounding technique is a daisy-chain ground. Next worse is having a component that is grounded from two different directions. Either take all grounds to a central "master" grounding point (star ground) or use galvanic isolation (separate grounds and voltage sources) between systems.



Tom Caudle
www.CandCNC.com

[kkilst]

Gentleman,
May be we are facing the same and this irritating problem.
Retrofit 3-axes milling machine using Rutex R2010 drives and MACH3.
I am facing this unsolved problem, that is one the drives (mostly Z-axis, sometimes X-Axis)
tripped in the middle of program run, with "External E-Stop Requested" appearing in MACH3 status screen.
Debounce Interval 15000, and all limit-switches and E-Stop with 0.1uF ceramic capacitor installed.
X,Y,Z limit-switches (parallel connected) are Active Low, and E-Stop can only be set to Active High.
This problem originates from Rutex R2010 drives or MACH3 setting ?
Comments and ideas is appreciated
AssoProf. Lie S.T. (Nanyang Technological University)

[RichardV48]

Tom,
Thanks for the input.
There is nothing between the parallel port and the controller board and the switches are wired in a NC format. Everything is tied back to a common receptical ground. The shielded cable is grounded on one end only and they are all tied to a common point. I'll do some checking with the meter and see what I come up with.
Thanks,
Richard

[Torchhead]

Originally Posted by RichardV48 Tom, Thanks for the input. <snip>Everything is tied back to a common receptical ground. </snip> Richard

So every ground in the system (table itself, motors, PC chassis, Shields, Motor DC neg) is all tied back to that one receptical ground?

And where might that receptical be grounded? 20ft away? 60ft? maybe not until it gets to the light pole outside?

The grounding and shielding for an system with high current waveforms (like the motor PWM from the drivers) is a bit more complicated than for 60HZ AC. Switching high current at higher frequencies injects noise on the attached conductors. The higher the frequency the shorter the length of conductor taht will "load up". Even poorly grounded 60HZ components that have two or more ground paths can set up a loop condition that can cause problems.

Unless you can confirm the validity of a ground (golden rod driven to the center of the earth (:-)) every point in a system is above ground by some amount. If you make everything in a system tie to the same point even if it is above ground the net voltage differential will be closer to zero.

When you see weird problems in signals you have to think noise. When conventional things like filter caps, shielded cables and debounce numbers above 2000 don't fix it you can start to suspect it is a ground related issue.

[RichardV48]

Tom,
I disconnected all the grounding on the machine except for the power cord ground to the controller and the pc. I believe that would have eliminated any ground loop conditions, and got the same indication. With all motors disconnected and the controller powered up, there is no limit tripped. As soon as I connect a motor from any axis the limit led flickers and I get an estop.
If the spike was coming in from the local ground wouldnt it trip a limit indication without the motor hooked up?

[mr_mocap]

Just read this thread with interest as I seem to have the same problem (or similar) after getting my machine running for the first time - limit switches trigger without being hit when homing / running cycles. So I wondered, did the problem get resolved in this case, if so, what fixed it? I think I can make everything reliable but any pointers I get in the meantime would help greatly...

[kreutz]

What kind of PC interface are you using? Isolated or non-isolated?

What is your PC LPt port output voltage for a logic High? 5 Volt or 3.3 Volt?

Were are you getting the logic voltage attached to the limit switches from? From the PC´s power supply or from the Controller´s power supply?

[mr_mocap]

Originally Posted by kreutz What kind of PC interface are you using? Isolated or non-isolated? What is your PC LPt port output voltage for a logic High? 5 Volt or 3.3 Volt? Were are you getting the logic voltage attached to the limit switches from? From the PC´s power supply or from the Controller´s power supply?

Thanks for replying! I measured the voltage at the port, and it is 5V. I also have a digital hand held scope purchased recently, I believe it can operate in storage mode and this should help isolate problems of this nature (but I haven't yet had the time to investigate this functionality). All voltages are supplied by the PC at the moment, unisolated as it happens. The limit switches are NC in series, so they are grounded when unoperated and the input floats high when the switches operate. The cables are shielded with the shield connected to domestic earth, rather than digital ground. I was a bit surprised that there was a glitch of this nature as the port inputs should effectively be hard grounded when the machine is away from the limits, but there is obviously some noise somewhere. I have considered using a relay to the port pin - this would relocate the digital signal to the proximity of the controller case, rather than having it extended all over the machine. I also suspect the problem will be worsened when the spindle is running, which it isn't yet as I am trying to resolve problems in turn...

[kreutz]

Originally Posted by mr_mocap Thanks for replying! I measured the voltage at the port, and it is 5V. I also have a digital hand held scope purchased recently, I believe it can operate in storage mode and this should help isolate problems of this nature (but I haven't yet had the time to investigate this functionality). All voltages are supplied by the PC at the moment, unisolated as it happens. The limit switches are NC in series, so they are grounded when unoperated and the input floats high when the switches operate. The cables are shielded with the shield connected to domestic earth, rather than digital ground. I was a bit surprised that there was a glitch of this nature as the port inputs should effectively be hard grounded when the machine is away from the limits, but there is obviously some noise somewhere. I have considered using a relay to the port pin - this would relocate the digital signal to the proximity of the controller case, rather than having it extended all over the machine. I also suspect the problem will be worsened when the spindle is running, which it isn't yet as I am trying to resolve problems in turn...

If you are supplying the +5V from the PC, then use the +5V Power ground and disconnect the ground on the controller's side from pins 18 to 25 of the DB25 connector ( you can leave them tied together), that will eliminate a possible ground loop. Check that all your power ground (return for +5Volt power) connections are solid, a false contact on the ground connection used by the limit switches will appear as a limit switch trip.

[mr_mocap]

Originally Posted by kreutz If you are supplying the +5V from the PC, then use the +5V Power ground and disconnect the ground on the controller's side from pins 18 to 25 of the DB25 connector ( you can leave them tied together), that will eliminate a possible ground loop. Check that all your power ground (return for +5Volt power) connections are solid, a false contact on the ground connection used by the limit switches will appear as a limit switch trip.

I misunderstood your question earlier re. the 5V supply. There is no 5V supply as such; I am using the properties of the parallel port to represent a logic 1 when the port is (effectively) disconnected. It is my understanding that there is an internal pull-up that ensures this is the case? In this event the ground reference I'm using is of course the DB25 ground pins. However, you may tell me that this is not a good way of doing things in which case I'm all ears!

Really I would prefer not to extend the PC's 5V supply unless there is a benefit in doing so, as it makes the setup a little messier than I would like. If 5V is actually necessary for some reason I would normally derive this from the controller's existing 12V supply, using a 7805 or similar positive fixed voltage regulator.

Having said all of that I am having unexpected problems, obviously, so any further advice or opinions are very welcome!

[kreutz]

Originally Posted by mr_mocap I misunderstood your question earlier re. the 5V supply. There is no 5V supply as such; I am using the properties of the parallel port to represent a logic 1 when the port is (effectively) disconnected. It is my understanding that there is an internal pull-up that ensures this is the case? In this event the ground reference I'm using is of course the DB25 ground pins. However, you may tell me that this is not a good way of doing things in which case I'm all ears! Really I would prefer not to extend the PC's 5V supply unless there is a benefit in doing so, as it makes the setup a little messier than I would like. If 5V is actually necessary for some reason I would normally derive this from the controller's existing 12V supply, using a 7805 or similar positive fixed voltage regulator. Having said all of that I am having unexpected problems, obviously, so any further advice or opinions are very welcome!

Don't trust the pull-up on the LPT port, there are many flavors of them. Limit switches (micro-switches in general) need a minimum current flow in order to keep the contact resistance low, normally it is around 5 mA. You will need a means to provide it in order to get reliable results, and so the need to tap the PC's power supply. 1Kohm pull up attached to the +5 volts is enough. Use a 0.01 uF ceramic capacitor between the 1Kohm resistor and ground, it will serve as a filter and also help keeping clean the contacts when they open and close.