So I have a bit of a strange problem. My CNC machine is in my shop that is about 90 feet from my house. I had struggled with getting internet connection to my shop until I put a TP-Link power line adapter in to run the internet connection through the cable connecting the shop to the house. Everything is great until I start the spindle on my cnc, then the internet drops.
I have the cnc machine on a separate circuit from my powerline adapter, and it is across the shop. Is there an easy way to fix this problem? I have thought about using a surge protector to plug the spindle PSU into, but do I need to do more?
How is the VFD powered? Are you in the US or similar place with split-phase power, or do you have three-phase mains power? If in the US or similar, 120 or 240V to the VFD? If in a three phase residential country, three phases to the VFD?
The general case of the answer will be the same no matter what, but the specifics of how to deal with it will be different depending on the power feeding your VFD.
You will need to make a device called a ācommon mode chokeā that involves exactly all of the power wires feeding your VFD and nothing else. You will be winding them together around one or more ferrite toroids, probably in the manganese-zinc family. The type of toroid to use depends on what frequencies are actually causing trouble.
I would start with a mix 31 ferite toroid and as many turns as possible. FT240-31 is a common name across multiple vendors for this part. I would buy two of them and stack them. To be complete from the start, and considering that powerline Internet might be using higher frequencies, I would add a mix 75 choke for the next higher frequency band. That would be a separate stack, separately wound; you donāt sandwich different types of ferrite toroid in one set of windings.
Anyway, when you provide the details, I can be much more complete on how to build this device. Itās not particularly difficult.
Actually, you probably just want to start with a ābrute forceā AC EMI filter. These come with multiple different chokes including a common-mode choke, and shunt capacitors that make them more effective filters.
I canāt link to one without the details, though. US mains power or something else? If US power, 120V or 240V?
VFD? I donāt think I do, things are pretty simple on my setup. I have a 500W spindle with the power supply and a speed dial - very similar to this:
I am in the US and have 120 power. This all sounds very complicated, but I am up for the challenge.
You have seen most of my setup before, but here are a couple pics if it helps.
OK, I see lots of different systems here so I canāt just remember what everyone hasā¦
Snapping those filters onto the power cord is a common-mode choke, but a fairly weak one, and unspecific mix so you really donāt know what frequencies it might attenuate. If you have one you can try it, but I doubt it will solve the problem. I wouldnāt spend a hot cent buying one with hope that it will solve this particular problem.
The brute force EMI filters combine a more effective common-mode choke with shunt capacitors; they filter a lot more noise.
Not remembering what you have meant I didnāt want to give the wrong easy solution.
I see what looks like hundreds of these in the $10ā$20 range on amazon. Many are identical-looking no-name units with no significant review history. Hereās one with a bunch of good reviews:
You can see on that image that it has two common-mode chokes inside (the wiggly bits on the schematic) as well as capacitor shunts to filter impulse noise (probably a significant source of your problem).
I suggest putting one of those on your spindle power supply. Itās the first thing I would try if I had that problem.
Ground itself isnāt filtered, so thereās only one ground connection. Just make sure to connect the line where it tells you, on the side with ground, and then connect the ground from the same terminal on to your spindle power supply.
Definitely right at the spindle power supply. Closer is better; any wire between the spindle power supply and this filter is an antenna and is radiating the noise, and the longer the wire between the spindle power supply and the filter, the better than antenna will couple with the antennas in your wall that you want to have a clean internet signal on.
A common mode filter between the mains and the power supply will definitely help but it is a remedy against the symptoms not a solution to the source of the problem.
I would use the filter that Michael advised but also put a filter in the lines to the motor. The red and black wires in the Amazon photograph. A 100nF capacitor between the red and the black and 2x 100nF, each from one of the wires to the motor housing. Also put a ferrite bead (shown earlier) around both wires but keep the capacitors between the ferrite and the motor.
@hennep Oh, I was assuming switching noise from the SMPS dominated, and that serial inductors in the SMPS would filter high frequency commutation noise from the spindle.
Thank you! Since you appear to have direct experience here, Iād like to defer to you on one more thing. If I were doing this, I would use HV capacitors, not cheap bulk ceramics, because some of those might only be rated for (say) 50V. If this is a 90V spindle, with spikes on top of that, I could imagine even 100V ceramics might be not enough margin. But maybe Iām wrong and bulk ceramics are good enough here?
Now Iām at my computer and can check easilyā¦ The unit linked above as āsimilarā has a 100V operating voltage and appears to be a DC brushed, commutated unit. 90V and 100V rated units are common in this 500W spindle size in my experience.
So my suggestion is to see whether the EMI filter is good enough to solve the symptom before suggesting to buy a kit of 100nF HV capacitors to add to the motor. Iām all for solving the problem at the source, but the EMI filter also addresses SMPS noise which is definitely a real thing too. And the spindle leads probably arenāt as tightly coupled in the near field of the rest of the AC wiring as the AC wires from the SMPS to the mains power, so from an antenna standpoint it also makes sense that it could resolve the symptom. Itās been a classic effective solution to this class of problems for a while.
Now we know that it is a brushed motor, I think you have found the culprit. These brushes always cause sparks when they switch the coils of the motor and the wires between the motor and the power supply will happily act as an antenna.
It would be nice if the common mode filter would solve the problem. but I expect that it will reduce the noise a bit. I hope @Fitfo_Buchanan will let us know the result next weekend
When you say itās on a different ācircuitā are you saying itās on the same āphaseā(L1 or L2) but just itās own breaker? If it is on the same phase as your powerline adapter then you can probably just swap your CNC machine circuit breaker with one adjoining(side-by-side) to switch phases.
If not up on US 120/240 power phase setup, power comes in on 2 lines named L1 and L2 which are 120VAC each relative to the white Neutral line but they are 180Deg out of phase with each other. Because they are 180 deg out of phase, when you use L1 and L2 then you get 240VAC.
Your breaker panel typically has L1 and L2 coming in the top(or bottom) with bus bars which have āfingersā such that even numbered fingers are L1 and odd numbered fingers are L2 so that each individual breaker is just 120VAC but 2 side-by-side breakers make up a 240VAC circuit.
This is why if you find your powerline adapter and your CNC are on the same L1 or L2 phase, by swapping the CNC breaker or the powerline breaker with the adjacent breaker will swap the phase and might isolate your interference.
Yāall are talking on a level so far above my head I donāt know if I am in an electrical engineer forum or a cnc forum. WOW the level of ignorance on my part is staggering.
The part from Amazon is delayed and supposed to arrive tomorrow, so we will see. As far as the phase issue, I attached a picture of the sub panel in my shop. The outlet that feeds the powerline adapter is on its own breaker (annotated PLA). The breaker that feeds the outlets that the CNC machine uses is also annotated (CNC)
Those inexpensive brushed-motor spindles generate a huge amount of electrical noise, and the power wires between it and the power supply are notoriously effective antennas. The solution is to suppress the noise as close to the motor as possible, just like @easytarget and @hennep suggested. Suppression caps are a good idea, but Iād suggest 200V-rated parts. Between the terminals Iād add a bidirectional transient voltage suppression (TVS) diode rated for 2X the maximum spindle voltage. These spindles typically use 48VDC max, so a 100V TVS should be fine.
A brushed DC motor is electrically modeled as an inductor that uses mechanical switches (the brush connections to the commutator) to be energized and de-energized. Inductors oppose sudden changes in current, so when the āswitch" turns on the energizing current ramps up rather slowly (compared to a short circuit). As current builds up, energy is stored as a magnetic field. Then, when the switch opens, the magnetic field collapses because thereās no current to support it. A collapsing magnetic field in an inductor produces a very high, transient voltage that is opposite in polarity to the energizing voltage, and unless it is suppressed it will arc across the āswitchesā. This is why we see sparks where the brushes and commutator connect. This turns the motor into a spark-gap radio transmitter, and the motor wires act as antennas.
Contrary to inductors, capacitor current increases rapidly as voltage changes, so they create an electrical path for the transient currents. We donāt want the motor wires to act as antennas, so we want to do the suppression as close to the motor as possible.
Again, yāall are speaking at the 30,000ft level and I am in the 10,000ft range.
I have seen a lot of talk about capacitors, but where exactly do they go and how? I have a short length (~10ā) of 14g wire running from the spindle power supply to the spindle. I then have the power from the wall running along the gantry down to the switch box for on/off. So I can understand, where is the wire acting as an antenna and where do I need to introduce the parts to stop the interference.
Just so we are all clear, the interference is only affecting the WiFi signal and there seems to be no problems with the operation of the machine.
If you have a multimeter, you can measure the AC voltage from the āPLAā to āCNCā breaker. If it is about 240V, they are out of phase; if it is about 0V, they are in phase.
Take a picture of the top of your spindle motor. Look at the one that @easytarget posted for his motor; take a good clear picture like it. Or a few. Then it will be easier to show where they go on your motor.
They will need to be soldered on. Do you have soldering tools and knowledge?
I agree.
The linked part above says 100V. Without measuring or a dataplate itās hard to be sure.
Assuming this is at least a 90V, maybe 100V power supply, if the link is a close match, means that I definitely wouldnāt want typical 50V rated capacitors. I find 1kV rated capacitors more easily on amazon than 200Vā¦
Given capacitors here, why a TVS? Iād think that the capacitors would be a sufficient snubber?
Here, I labelled your breakers just randomly picking which is L1 and which is L2 on that main breaker(80A 240V). You can see the tongue at the bottom right which has no breaker attached.
Thatās the thing which connnect the input power to your breakers.
From the picture you can see that your CNC and your PLA are on the same AC input ālegā.
So, if you turn off the PLA breaker, left(rotate) it up from the side where the black circuit wire is connected you can pivot it off of the L2 bus bar tongue and then you can move it down one space into that open L1 bus bar tongue and press it down so it latches onto the tongue securely.
Throw the breaker and thereās a good chance you will not have any noise problems or will at least have much less.
You could also swap your CNC breaker with the one above it and then the CNC would be on the L1 bus/line and your PLA would remain on the L2 bus/line.
