CNC Router – Chasing ghosts in the machine. The dreaded e-stop

Bottom line up front - We have a problem with electrical Noise in the CNC Router. Noise doesn’t come just from the spindle and steppers on the router, but from other equipment in Hive13 as well.

As CNC Warden I’m calling for volunteers to join an electrical noise mitigation project for the CNC Router. We need CNC users, wannabe CNC users and electronics makers for planning and hands on help. We can break this down into sub-tasks and share the work. See the outline below.

When: Planning to start in mid-January. We need to sketch the improvements, buy parts and plan the install. The dollar cost is probably low. Perhaps under $400.

Back Ground: Most everyone who has used our big CNC Router has encountered a random e-stop. It happens so often, restarting after an e-stop is almost standard procedure. Groan. E-stops are bad, but the noise level is so high that occasionally the hall effect end stops are ignored.

The Solution: Suspecting noise, I asked Lorin for advice, which resulted in this set of steps to mitigate noise in the CPU:

  1. Assure star grounding for all components of the CNC Router. Create a grounding buss bar connected to a chassis ground for the control module on the wall. Follow best practices for independently grounding the steppers, spindle and chassis to the ground buss. This buss can be a metal bar with tapped holes, ring terminals and 12 ga wire.

  2. Shorten and repair or replace all external signal and driver wiring with shielded twisted double pair wiring. Attach shields to ground for all external signal wiring. Because current wiring shields are not continuous and do not connect to ground they are antennas. Add ferrite loops to signal lines to damp out spikes.

  3. Add a separate power supply for logic level voltages 5v & 12v. Stop feeding power from USB to logic boards. Use the separate power supply. USB is not well shielded.

  4. Shield the microprocessors in individual metal boxes.

  5. Low priority – Check/filter RF noise from spindle motor. Add filters on stepper motors. This is probably redundant to stepper drivers.

None of this is planned out in any detail though we can sub divide the tasks as follows to make each job easier for a team.

This isn’t guaranteed to solve all problems, but the noise we tracked on the oscilloscope is guaranteed to cause intermittent problems. We saw a 1 to 12.5 signal to noise ratio on one e-stop lead. Wow.

Steps 1 and 2 should have big payoff. Steps 3 and 4 are further protection.

One obvious team division is inside and outside of the plastic shield on the wall.

The controller team(s) will gather parts, make the bus, power supply mounts and shield boxes in advance, then do a get’er done Saturday change over to minimize down time.

The external wiring team would assure shielded/grounded cables on the table. Add grounding to the spindle, steppers and carriage so they are all electrically neutral to one another. The external wiring can be done over time before and after the controller changes.

Let’s discuss it here, form up teams and make this happen.

Have a Merry Christmas, See you all in January.


Friendly reminder from someone who deals with shielded wire a lot…

Connect ONE end of the shield to ground. ONE. Not both. ONE.


Yes, the end at the chassis / earth bus to be specific. Star grounding at it’s best. Also know the difference between earth and ground. :slight_smile:

That was decent.

Hello All,
Hope your holiday break was good.
Unfortunately the CNC Router chose the break to go from having random eStops to a non-performing crisis.
SO, I’m bumping this post up to remind everyone of the CNC eStop issue and planned cures.
We need to gather materials and implement the basic steps faster. Preferably in the listed sequence as these are thought to give the greatest improvement for the effort.

I’ll try scrounging materials from Hive13 stocks and post what we have and what we need. If you have parts that could be contributed, please post here.

This is an OPEN INVITATION to All Hive13 Members. Saturday 1/12/19 will be a CNC Router rejuvenation work day. Your help would be appreciated. RSVP’s are appreciated.

You don’t have to be a CNC router expert to participate.

Thank to all.

I will try and come down and help with electrical stuff.
I might also remind you that the Hive owns a HackRF Blue. While SDR is often used for communication, I’ve found it to also be rather useful for measuring and quantifying electrical noise. If you indeed suspect that the cause of the e-stops might be related to RF interference, capturing or observing the frequency of RF noise might give some insight into the source of said noise by its primary frequency. From there, it might be a lot easier to come up with a targeted solution to mitigate it.

If you have the tools, use them.
-Dave B

Based on what I was seeing last week when it gave up, I think the issue is that we have part of the system isolated from ground, so the ground reference is floating and different parts of the system or referencing different voltages as “ground”. I’ll try to be in to help, though I’m not 100% sure I don’t have a schedule conflict.

I traced the electrical connections to ground this afternoon. It's a daisy chain until the chain breaks. As far as I can see the shielded cables running all over the machine are not grounded at all. None of them. We need to add a common ground BUS and tie all the stuff together. That can start with the cables in place, but will require some care to connect shields across breaks in cabling and continuity checks on shields to assure it's grounded at one end. I couldn't find A suitable ground buss at hive so we need buy or get a donation.

I'm now going to look at best practices to assure grounding on spindle to chassis connections.

I’ll try and swing by after work tomorrow, should be able to help before the meeting.

What sort of current does this BUS need to carry? What’s the cumulative load of all the moving thingies? I have some copper bus bars around here somewhere, but it may be overkill for this application.

When Dave and I went over the wiring 3 weeks ago, I saw the same thing. There’s a lot of capacitive coupling and reactance making noisy voltage vomit (crossstalk, emf, etc). The majority of the crosstalk and noise is low current and simply needs to be properly tied to a star ground rather than a Spaghetti Reactance Obfuscator. I’d say high quality, unbroken shielded cables (not cut up and spliced together all over) should be the biggest priority. The hefty-ness of the ground or earth bus is then simply a matter of current load (is it that high?).

I’ll free up my calendar. I will guarantee that we’re going to need some new shielded cable. Many cable runs have reached a point where it would be easier to simplify and rerun them than to face the challenge of splicing dicing.

Maybe I’m misinterpreting the bus bar thing… Regardless, should be fun!

The grounding terminal strip is more a matter of convenience than amps. There are many branches to motors and sensors that all need to reach ground. There are also long stretches of naked conductors spread to multiple terminal strips. These too need to be shorter.

I'll get more specifics on amps and guage tomorrow.

OK - Wire for the CNC re-do requires some trade offs. Ideal wire has more than 7 strands per conductor. All high flex 18/4 AWG Shielded cables I can find are larger diameter in the 9.2mm OD range. That wire won’t fit the gantry wire drag chain which is 15/15 mm internal and must carry the spindle, Y and Z steppers and a 4 conductor sensor wire for z-end stop and zero sense.

The wire chain can’t be easily increased in size as the gantry has limited space for the guide to pass below. Going wider interferes with the screws assembling the gantry. So, the trade-off in wire that will fit the guide in 18/4 AWG Shielded cable is 7 stranded with foil and drain wire.

One choice possible is BELDEN 5302FE or similar. - I’ll need to place an order soon to get delivery by Friday. Any inputs?

Come one come all.

The CNC Rewiring extravaganza is ON for Saturday January 12, 2019 I’ll be in Hive13 around 10:30 am

I’ve also probably forgotten something(s) in the task list below. Please help fix those omissions.
There most certainly will be a lot of decisions to be made on the fly, Bring your brains and prepare to remain caffeinated.
Your help is greatly appreciated in getting the CNC router back on line.

Tasks to do - Many of these can be done in parallel. Yes there will be something for you to do.

  1. Prepare the machine to be rewired. Remove old wiring. Move the cable tray and wide wire chain from the aisle side to the other side. It will shorten the cable runs.

  2. Add a ground buss bar to the Control panel. Reroute Ground to the bus. Replace odd color coded grounds with Green wire. Run a separate ground line out to the chassis and on up through the gantry to the Z carriage and Motor spindle.

  3. Remove the board with the Terminal Strips from the end of the chassis.

  4. Prepare to mount a directly wired harness to the control boards. This will eliminate multiple connections and assure continuity of shielding. Add strain relief(s) for motor and signal lines to wall.

  5. Drill a hole in the Gantry to allow wires to be routed through the extrusion and up through the flex wire chain. Add strain relief

  6. Separate signal from motor wiring on panel by at least 10cm. Do not cross the motor and signal wiring. Rewire to add shield to the signal wiring. Reroute motor wiring as required.

  7. Layout / route and cut the new wire.

  8. Check shield to assure continuity and that there is no ground loop, then wire into the panel.

  9. Solder on the motor connections.

  10. Mount the end stops, estops and limit switches.

  11. Quality check it.

  12. Fire it up for a test run.

Wire ordered is Belden 5302FE - 18/4 Shielded. 500 ft should be delivered Friday. We can use it for signal too.

We have a buss bar for a star ground
The only thing missing is you.


I added a calendar item from 10:30-4:00

I am planing to help, see you Saturday.

Depending on the upcoming snowfall . . . I’ll try and be there.

I’m going to try to be there Saturday.

One thing that I remembered today is that foil shielding is not great for high flex situations. Sorry to bring this up at the last minute but it’s worth considering before rewiring the whole thing.

Thanks for the link on shielding. There are so many trade off’s to be made. The cable selected is one of these. The limiting factor is the size of the four wires in the wire path up to the spindle. Unless we massively rework the mechanical design, the wire flex chain is 15 mm square.I could not find braided shield, high flex cable in anything less than 9.5mm OD. The wire chain guide would handle at most two of those four wires. So, the foil was chosen and will arrive this afternoon.
The article didn’t quantify failure or flex as a function of time, The bend radius is gentle. I’m going with the compromise.
When this shield fails at some point then we can have another rewire party,


Snowpocalipse? Nah. Just a ploy to sell bread and milk.

Let's get this CNC rewire job done today. See you at 10:30 or when you can make it.

Thanks, Dave

I don’t think it is a snowpocalypse, but I’m more worried about the awesome Cincinanti drivers on the 40 minute drive.
I’ll see how things look later today, us Floridans don’t like the snow. Feel free to dock my pay. :stuck_out_tongue_winking_eye:

Happy to do some work later.