[Fastest1]

A few years ago I built a Hobby Cnc Pro board (which besides this particular issue has performed very well). It has controlled the steppers on my mills, lathes and or printer when asked to do so. When I assembled this controller I did install it in a metal enclosure instead of the plastic box but other than that pretty much just like the instructions said to do.

As I learned more I decided I wanted limit/homing switches. I had run the machines without them just fine but recovery was not possible if I crashed, over fed or broke a tool.
First I ran mechanical microswitches. They were great for repitition and find a place to call home. Sure enough during long runs and sometimes during short runs the inevitable "limit switch triggered" would pop up for no particular reason. I blamed it on cheap switches, vibration etc.

Then during reading about these noise issues I read about installing optical switches. Again I learned enough to assemble the switches. During testing of these switches on the bench, they performed flawlessly. I get a reading of -.01v if the gate is unobstructed and 4.69-4.7v if gate is obstructed. Thinking my problems were over i installed these only to find the noise gremlin still there.
I do know that my original controller box might not have been the best wiring or organizational attempt (especially compared to some of the absolutely beautiful examples provided here on a regular basis). No reason to chase any further til this was cleaned up or so I thought.

After much procrastination, I dove in. I undid all of the wiring. During the rewire, I twisted all pairs that powered any other power supplies or peripherals. I used a star ground and grounded everything there, the pc case, the machine, and all drains from the shielded wires used for steppers and the limit switches. I tried to makes sure no wires ran parallel to other wires. I attempted to make all wire intersections 90 degrees. All shields are connected at the controller end only.
I have tried capacitors across the input pins, raising the debounce levels, seemingly everything I have read. Still no luck.

Today during the reassmbly I still had my limit switches powered by a wall wart. With my controllers power off but my limit switches powered up and Mach open, my limit switches perform flawlessly. No random flickering or nonsense just an led when the gate is obstructed.
Then I turned on the controllers power all of a sudden this random flickering shows up on the diagnostics/hardware page. It flickers with no rhyme or reason across pins 10-13. Sometimes it is only pin 11, sometimes all 3 or 4 pins. It is happening often and debounce doesnt seem to have an effect. I have not even turned on a spindle or moved the steppers to cause this and doing either of these doesnt make it worse oddly enough.

I guess if I had an oscilloscope and or had a clue how to use it I could most likely see a spike causing this.

What is my next logical step in solving this issue? The power supply is a combination of a transformer, rectifier and capacitor (the parts HCNC recommended) and it tests for voltage as it should.

[Al_The_Man]

Did you try taking the P.P. ground pins to the star ground point, also did you do the same with the DC -ve of of the P.S.?
Is there any P.s.'s that are floating, such as wall-warts etc?
Al.

[alan_3301]

I am by no means an expert, just tossing ideas out there.
Is it possible there is a problem in the rectifier or capacitor of the power supply? Can you check for AC voltage on the DC lines? Im not sure what an acceptable level would be. Can you get your hands on a bench power supply or computer PSU to power the driver board? See how it acts with a clean power supply to rule it out. I had a hobbycnc kit in the past, and the power supply I built was pretty crude. just a transformer rectifier and capacitor.

[Fastest1]

Al, I did not connect the -ve to the star ground just the chassis itself of the 5v supply. How would I ground the power supply of the transformer/rectifier/capacitor combination? Do I just connect the negative wire from the capacitor (42v output) to the star ground also? That was going to be my next question. Is it OK to connect the ground of the HCNC input pins there too? Which would be connecting the PP thru it also, correct? The different terms though similar are very confusing, ground/-/earth.
I dont have any floating groundsat the moment though I was testing both ways with no changes. The culprit appears the moment I power up the transformer.

Did you try taking the P.P. ground pins to the star ground point, also did you do the same with the DC -ve of of the P.S.?
Is there any P.s.'s that are floating, such as wall-warts etc?
Al.

[Al_The_Man]

In the case of Power supplies, as long as the AC secondary side is not connected to ground then you normally can take the DC-ve to Earth Ground.
Likewise with any 5v supplies etc.
Normally your PC P.S. is grounded to the case via the mother board, but I normally re-reference the P.S. common when it leaves the P.C. (P.P. etc).
I have usually always used the term 'Ground' as Earth Ground and any P.S.-ve's etc as Common.
Al.

[Fastest1]

Ok, so I connected the -ve of the 5v ps to the star ground. I also ran a wire from the - terminal of the capactor (of my power supply) to the same point on the star ground. I still get trips. However my debounce now does have some effect but the underlying problem is still there. I then added a .1uf ceramic disc capacitor between each of the input pins for the limit switch and common. Still no love.
Btw Al, I am not blaming you for using different terms. It is in my understanding of the terms where the problem lies. Are you saying it is OK to connect the commons and the -ve's to the star ground? Also my pc is being powered from an auxillary plug within the same enclosure and voltage supply source if that makes any difference.
Not sure about the secondary side of the transformer, is this the step down voltage prior to rectification? (still AC at this point?)

[Fastest1]

Same power supply most likely. As I built it from their parts list and suggested suppliers IIRC.

Regarding the checking for Ac on the DC lines, is there a way to do this with just a multimeter?

The bench supply was my next approach if all else failed. I dont have one at the moment but it seems they are always needed.

So I just depowered the HCNC board. All limit and E stop signals still pass thru and are verified of the hardware/diagnostics page within Mach3. I get no tripping unless I obstruct the gate. It definitely makes me suspicious of the power supply side of the equation but then again I have been wrong almost every time in the analyzation of this issue. But a Keling or similar power supply would allow for more room inside the case!

I am by no means an expert, just tossing ideas out there.
Is it possible there is a problem in the rectifier or capacitor of the power supply? Can you check for AC voltage on the DC lines? Im not sure what an acceptable level would be. Can you get your hands on a bench power supply or computer PSU to power the driver board? See how it acts with a clean power supply to rule it out. I had a hobbycnc kit in the past, and the power supply I built was pretty crude. just a transformer rectifier and capacitor.

[Al_The_Man]

Are you saying it is OK to connect the commons and the -ve's to the star ground? Also my pc is being powered from an auxillary plug within the same enclosure and voltage supply source if that makes any difference.
Not sure about the secondary side of the transformer, is this the step down voltage prior to rectification? (still AC at this point?)

I usually Earth Ground all commons, it is rare, but sometimes the secondary side of a transformer is grounded so it pays to check, unless you built it, in that case you would know.
Al.

[Fastest1]

I used another power supply (24v 4.5A) with no change in behavior.

On my transformer there are 5 wires, 2 of 3 are used direct to the rectifier (1 is unused and covered) and the other 2 are the supply voltage. There is no ground within the power supply that I can trace. I have the incoming ground wire attached to the star ground. Though I just disconnected it and it made no difference. I get the error regardless if the -ve is connected to the star ground or not.

I again disconnected the home made power supply from the HCNC board but left everything else connected including the 5v power supply for the limit switches. No errors, Then I turned on the power supply (though this time it is not connected to the HCNC board) and the errors come back immediately. I tried it with no ground connected and no change then either. I cant believe how long this problem has been an issue. I dont feel any closer to a resolution. It seemed obvious before trying to rewire/clean up the rats nest. Just baffling.

[mike_Kilroy]

1) Regarding the checking for Ac on the DC lines, is there a way to do this with just a multimeter?

2) So I just depowered the HCNC board. All limit and E stop signals still pass thru and are verified of the hardware/diagnostics page within Mach3. I get no tripping unless I obstruct the gate.

1) yes, just put ur meter on AC setting and look across the DC terminals u want to check - any ac riding on top of the dc will be measured.

2) What does power to HCNC board do?

of course u should ground stuff and shield stuff to reduce noise getting into stuff, but u also should try to find just WHERE the noise comes from - that can help to know what additionally needs to be done. most often the noise is caused ONLY when the drives are turned on (moving? sitting still? enabled?) - is that ur case? 2 drives? 3? have u turned them off and watched for the noise on the switch input sigs? then turn on 1 at a time? u never know; u may find the problem is CAUSED by only 1 of the 2(3?) drives.

r ur drives servo or stepper?

assuming u find the noise only happens when the drives are on, have u used shielded motor wires between the drives and motors? that itself helps some folks reduce the noise.

think of it this way.... there is noise. it can be big or small. ur inputs blink and turn on with the noise when they shouldnt. the inputs need a certain amount of noise to false alarm on u. how much is the question. ALL systems will have noise. ur job to prevent perpetual noise issues is the reduce the noise to a level sufficiently below that turn on point to stop it. so say on scale of 0-10, any noise above 5 say causes ur switches to turn on at wrong times. u need to make ur noise below 5. make it 4.9 and u will continue to have random problems. make it 3 and u r done with it for good until something changes to bring it back up to 5. make sense? so consider that each drive probably is adding 2 right now; 3 axes, means u r at 6 - problem. maybe changing to shielded motor leads will bring each drive noise down to 1; hence 3 total and ur problem is gone.... each time u fix a grounding wire issue, u may take the background noise level down .5 - see?

we used to ground the dc bus negative on our servo drives but found in time that it really never ever reduced system noise so stopped doing so many years ago. If someone had a 42vdc ps for their drives and major noise, of course trying grounding this point to the star point is worth it but usually does not help from our experience.

any chance u ran those noise input wires close to the motor leads? or 42vdc power leads? or ac power leads? those are no-nos that will allow them to pick up the motor noise..... if so, u need to seperate them - see how much that reduces ur noise ceiling...

I am going to assume ur HCNC board thing power was used to supply enable sigs (outputs) to ur drives and so w/o it on, the drives are all off, so ur noise issue went away - did I assume correctly?

hope this helps some.....

[Fastest1]

Mike (and to all who contribute) thanks for any input.

This is a stepper system. Hobby Cnc Pro is the board I am using. All drives are combined and not able to be individually turned on or off. That was why I thought it was odd that it faulted when the 34v power supply was energized but not connected to the board. The 5v supply was still on powering the limit switches. Any triggering physically of the limit switches was verifiable on the Mach diagnostics/hardware page even though the board wasnt powered up. This leads me to think the signals pass thru the PP regardless if power is present.

All wires to steppers or limits are shielded and are seperated from each other. They do not run parallel to any power or other wires. All intersections in wiring were done at 90 degree angles.

I did connect my multimeter to the DC side of the PS measuring for AC voltage. With no power it reads .000 (2v scale), upon powering up, the scale briefly jumps to .4 though this wasnt repeatable to the exact number. Then it quickly (less than a second?) settles to a .002 reading with no change from then on. Would this small amount of AC cause a problem?

[alan_3301]

2mv AC is no problem. Did you solder your board together or did you get it assembled? what board and revision is it? I know it might not be likely, but you might check every component on the board. Possible you soldered in a capacitor backwards?

Sorry if you said already, but have you tried with the steppers disconnected from the board? If all is well, power down, wire up 1 stepper, power up, and repeat with each one. maybe something in the board itself is causing the interference. It sounds like the power supply has been ruled out.

Anything on the hobbycnc support group to give you any clues?

[Fastest1]

alan, Thanks on the voltage. Yes I did assemble the board myself (a first for soldering boards). It passed all of the tests just fine and I even had a friend who is an electronics specialist review it prior to connecting it up (2-3 years ago). He thought it looked very good, (thanks to Youtube soldering tutorials).

Upon further testing this morning. It appears I can actually power the board up as long as the stepper wires are not connected to the machine. This leads me to think it is within the stepper wiring or shields. I have the debounce on 0, the board is powered up and I can jog any axis (remember the steppers arent connected during this test), or all 4 at the same time and turn on and off my 2 relays and no errors are occuring! I am going to try disconnecting each shield, 1 at a time until it goes away. Hopefully.

This problem has plagued me from the beginning with this controller stepper combo. Though I had workarounds so I could enjoy some of the journey. Oddly enough the only thing that hasnt changed is the wiring to the steppers. Embarassing if my new suspicions are correct. Resolution is the objective not appearance!

[Fastest1]

I just disonnected the shields of the wiring to the steppers and the problem is gone! (this is the one thing that hadnt really changed over the years). No debounce, no blips on the diagnostics page. Excellent. The steppers are very quiet at idle. I would always notice a change in the stepper volume at idle during connecting or disconnecting various shields. Subtle sound change but I was suspicious of the change I just didnt know what it meant. My shields were always only connected at the controller end and I was under the impression this was to drain the noise thru the star ground. This was the first thing I did during the build to help eliminate noise issues was to use shielded wire and to tie the drain to the star ground.

Thanks to any and all for your input.

[Al_The_Man]

It would be interesting to see what the effect of grounding both ends of the shield, this is also presuming you have a ground bonding conductor to the frame of each motor.
Al.

[Fastest1]

I think I will let sleeping dogs lie on this one. If I need to extend my stepper wiring in the future so be it. I will check then. The wiring was soldered and heat shrink on each connection. If I were to do it again I would probably go with crimp style connectors. I have been moving the controller around and wiggling all the wires inside with no errors today. Now of course I want an LED for each axis to let me know when a home switch has been tripped. I have one that works physically (the LED lights up when the gate is obstructed however I havent measured the voltage change) but I am sure a resistor value will need to be changed in my optical switch circuits.(chair)

It would be interesting to see what the effect of grounding both ends of the shield, this is also presuming you have a ground bonding conductor to the frame of each motor.
Al.

[Fastest1]

Al, running extra ground bonding to the machine and controller made no difference. Now or before.

[mike_Kilroy]

glad to hear u got it much better. two small comments - food for thought.....

u keep saying ur debounce is set to 0; i assume this is on the inputs that have had issues. why? debounce is a good thing. ok for turning off for testing, but then on these kinds of non industrial setups especially, you really need debounce to stop these little noise issues. i see no reason to not have debounce set to values that are equiv to say 100-200 mseconds to make the sigs that much more rugged.

second comment relates to al's suggestion to tie motor cable shields at BOTH ends. not sure if ur last post means u tried it or not.... SIGNAL wire shlds ARE tied to ground at one end only - to drain outside electrostatic noise off and keep it OUT of the wires inside. motor shielding should be tied at BOTH ends - different applicaiton than sig wires - it is to keep the inside electromagnetic noise INSIDE the shield and hence tie both ends to let it flow across the shield. again in a noise problem area, both ways are usually tried but the industry std today is to tie both ends to ground on motor lead shields. also, today's drives switch onoff the transistors or fets in sub microseconds, making very high frequency noise, so hooking a 6" little # 18 wire to the shield and that to ground is almost useless - that 6' pc of wire wil have high resistance at the noise freq, and may even be a 1/4 wavelength at the noise freq and hence an infinite resistance so same as not even there... also, the noise running thru the shld is on the outer surface only - called skin effect (google it) and so today's way is to peel back the pvc cable cover exposing the shield and then wiretie it directly to a metal ground point - at both ends. THAT makes the ultimate motor cable shield.

just a couple last food for thought comments.... again, congrats on getting there after all this time!

[Fastest1]

Mike,
Thanks for the comments. If I read it, there is a chance I will misunderstand it.

I had never read about connecting the shield to a motor wire at both ends. That is interesting. I always noticed if tying down the shield to a signal wire on my controller the hum/buzz would change in sound. I tried quite a few things over the years with no luck like jumpers between the machine/controller/computer etc.

Regarding the debounce, I had tried many different values over the years with varying amounts of success. I do have a few G540's on other machines and have much less if any problems in this area. Probably accidental as I have been prepping the stepper wires the same way regarding the drain at the controller end only. I would be totally happy with a debounce as low as 100-200. It would trip with any value before, regardless (this is what let me know there was really a problem and not errant noise). At the moment I am not running any additional preventitives, capacitors, resistors or ferrites. I did clean up the wiring considerably either way and I had been procrastinating that.

I just want to thank all those who contribute. Even those with little or simple suggestions. It is usually the obvious that was missed. I also know in my case that during an attempt to help others with or thru a problem it clarifies my understanding immensely. If it doesnt usually a more knowledgable person does! It is a true pleasure to have access to so many intelligent and experienced people who contribute their knowledge freely. The exchange of information 24/7 keeps so many things out of the FOR SALE ads. Thanks again, John

Oddly enough I installed custom home theaters and similar peripherals, automated window coverings integrated into projector controls. Lots of Crestron stuff and the like (not a programmer). The concept of signals running either on the outer layer or thru the core always baffled me. It has been explained to me by very knowledgable people and I believe them I just dont get it. Of course I didnt get Electronics in high school and dropped out. If I only would have paid attention then!

glad to hear u got it much better. two small comments - food for thought.....

u keep saying ur debounce is set to 0; i assume this is on the inputs that have had issues. why? debounce is a good thing. ok for turning off for testing, but then on these kinds of non industrial setups especially, you really need debounce to stop these little noise issues. i see no reason to not have debounce set to values that are equiv to say 100-200 mseconds to make the sigs that much more rugged.

second comment relates to al's suggestion to tie motor cable shields at BOTH ends. not sure if ur last post means u tried it or not.... SIGNAL wire shlds ARE tied to ground at one end only - to drain outside electrostatic noise off and keep it OUT of the wires inside. motor shielding should be tied at BOTH ends - different applicaiton than sig wires - it is to keep the inside electromagnetic noise INSIDE the shield and hence tie both ends to let it flow across the shield. again in a noise problem area, both ways are usually tried but the industry std today is to tie both ends to ground on motor lead shields. also, today's drives switch onoff the transistors or fets in sub microseconds, making very high frequency noise, so hooking a 6" little # 18 wire to the shield and that to ground is almost useless - that 6' pc of wire wil have high resistance at the noise freq, and may even be a 1/4 wavelength at the noise freq and hence an infinite resistance so same as not even there... also, the noise running thru the shld is on the outer surface only - called skin effect (google it) and so today's way is to peel back the pvc cable cover exposing the shield and then wiretie it directly to a metal ground point - at both ends. THAT makes the ultimate motor cable shield.

just a couple last food for thought comments.... again, congrats on getting there after all this time!