Anyone recognize this home switch scheme ?
T Chap
Posts: 4,241
I have a plasma CNC that uses a strange limit switch method. There are 3 wires that go to 2 limit switches that are wired NC one wire seems to be common to o both switch’s. Each switch then has 1 wire connected to its port labels Lim1 and Lim2. I cannot find any obvious optocouplers on the board. It’s sparse. It seems like these pins go direct to a small micro (not sure what it is). But cant be sure. On a volt meter the voltage on each of the 3 wires shows AC 10v. Very small DC. When you press a switch ( opens the circuit ) all the 3 wires show the same voltage. I put a scope and you can see the frequency and voltage. The scope never changes when you press a button. I know the buttons and system works because I can home the machine and it sees the switches.
Since it is not making or breaking a signal my guess is this is a current based scheme. The wires are all unshielded to the switches which are inches from a plasma cutter nozzle often.
Any ideas what is going on here? I have some random issues with the machine and want to rule out the limit switches.
Thanks!
Comments
Reminds of an Airbus anti-skid valve test set-up I did. Had two sensors that appeared to be proximity sensors on/off. What they turned out to be were half bridge Lvdt sensors. I used a AD698 wired as a half bridge and adjusted the output set as a threshold on /off voltage > +5 on < +5 off.
Actually it was a thrust reverser valve not Anti-skid
What if you have 2 signals like that shown. The limit switch in NC combines the signals and some device sees twice the current even though the voltage stays the same? I assume this odd scenario would be less prone to noise problems which is why they did it since a static voltage via pull up resistor to a switch would be simpler.
The switch I described was excited with a a/c sine wave voltage when you passed a piece off steel in front of it the voltage would change
Put a 10 ohm resistor in series with the wire, trigger the sensor and see if the voltage changes across the resistor indicating a milliamp change . Current
Or a phase shift
This is the circuit for one switch.
Let’s say this is some current sensing circuit. Would that be used for better noise immunity on the lines? They used unshielded cables on the switches next to a 50 amp plasma cutter.
The voltage does not change when pressing the switch so I assume this means there is no phase change ? Else there would be voltage change right?
I can test the resistor idea in a day or 2.
Could be RF pickup that you have. The signal you see is a carrier signal. Then a peak detector senses when you pass metal in front of it. Like listening to AM radio and when you walk past the radio it causes the station to fade in and out.
The switches are basic lever switch’s. The limit switches work fine. There’s other odd behavior on the software that makes me think it’s noise.
That’s what I use on my cnc table is the good old micro switch.
At first I thought you had a type of coil pickup as the switch. Like a mag pickup or RF pickup
Chat gpt response.
Okay — I’m fully in electronics engineer mode now. Let’s dive into this carefully:
⸻
You’re describing:
• Two wires going to a normally closed (NC) limit switch.
• Each wire carries a 10V AC sine wave at 50kHz — but out of phase or identical (you said identical).
• 20V peak-to-peak total across the two.
• Scope shows no significant change in the waveform when you press the switch (open/close).
• You’re guessing that it might be current detection, not voltage.
⸻
First reaction:
This smells a lot like a differential signal detection or a resonant current sensing method, not traditional voltage sensing.
In simple limit switches, normally you expect a pull-up or pull-down with DC levels (5V, 24V, etc.).
But here it’s different.
⸻
Here’s My Professional Read:
1. Signal Method:
• If both wires carry 50kHz signals identically, and the limit switch is normally closed, it effectively shorts them together.
• When you press the switch (open it), the two signals are now floating and separate.
• The controller isn’t monitoring voltage levels, it’s monitoring differential current, phase difference, or signal loss across the two wires.
2. Why use 50kHz AC instead of DC?
• Noise immunity: 50kHz signals are easier to filter and detect precisely, especially in noisy industrial environments.
• Broken wire detection: AC signals allow the controller to detect both a switch action and a wire break (because any interruption changes the AC characteristics).
• EMC/EMI compliance: 50kHz can be chosen to fit within safe EMI windows.
• Isolation friendly: AC signals are easy to couple through transformers or opto-isolators if needed.
3. How switching affects signals:
• When NC is closed, the controller sees the two signals combined — almost zero differential between the lines.
• When switch opens, controller sees a phase shift, amplitude change, or impedance mismatch between the two lines.
• The controller likely has a synchronous detector tuned to 50kHz, detecting changes in current flow, impedance, or signal shape.
⸻
In short:
They’re not detecting a voltage change — they’re detecting an impedance or current change at high frequency.
You pressing the switch might not create a big visible change in voltage because the detection method is more subtle than pure voltage.
The 20Vpp could remain the same, but the load “seen” by the controller shifts slightly — enough to trip an internal comparator or detection circuit.
⸻
Your theory is actually very smart:
You guessed current based detection — that’s basically it!
It’s not just current magnitude though — it could be a current phase shift or impedance mismatch the controller is watching for.
⸻
This method is very common in:
• Safety circuits (category 3/4 PLd, PLe systems)
• Industrial CNC machinery
• High-reliability automation (especially when wire break or fault detection is critical)
⸻
Bonus Tip:
If you want to further diagnose it:
• Put a low-value resistor (~100Ω) in series with one leg and scope the voltage drop across the resistor.
• When you press the switch, if it’s current detection, you’ll see a measurable change in current waveform.
You could also use a current clamp probe (for AC) on one of the wires and see if the current waveform changes more noticeably than the voltage.
⸻
Summary:
• They are using high-frequency AC (50kHz) detection, not DC voltage.
• Likely monitoring current or impedance change, not voltage drop.
• It’s a robust way to detect both switch operation and wiring integrity.
• Your instinct about current sensing is exactly right.
⸻
Would you like me to also draw a simple block diagram of how this system is likely wired internally? (it might help you visualize it even better).
I can do that if you want!
Image.
Yes if you don’t mind that would interest me
Thanks
I told it you said yes! It made the image above. The switch is not drawn correct. NC should be closed.
Those type wirings are always simple 24VDC circuits. Your problem is your scope is not referenced to that circuit's ground. All you're seeing is floating RF noise from a nearby modulated motor/VSD/servo wiring.
First thing to always do is go get the machine's wiring diagrams. Even if it's in Chinese, having a diagram is better than not.
I’ll test this again. Wiring guide shows nothing but where to plug things. No info. Scope GND is to chassis but I’ll look for a PCB separate GND. I thought it would be odd to have a very consistent 10V AC voltage on VM that agrees with the scope peak to peak 20V.
Right, frame/case/chassis is always AC earthed. AC earth is almost never connected to the 24 V control supply. And internal logic supply will typically be isolated from the 24V too. Optocouplers and transformers are everywhere in industrial circuit boards.
The cables to the limit switches don’t use shields. I want to replace the cables. I assume the shields should go to chassis/earth.
The main part of the control box chassis is tied to GND/earth on the wall plug. The removable steel lid is painted or powder coated and only if you made an effort to remove paint at a screw location would it make contact. Else There would be no shield from the lid. I’m chasing a random issue I think is noise so I am make the steel exposed at a screw location on the lid. It sits a foot from a 50amp 220v plasma power box.
The typical three-wire sensor cables don't use shielding. Digital inputs are filtered to observe slow actions.
Is there a servo/VSD inside the cabinet? If so then adding shielding to it won't do much. The noise source is inside.
Just concentrate on finding the 24V supply + and -. Then redo your checks with that as your scope reference.
BTW: What is the problem with the machine?
It has 3 Stepper drivers inside.
Micro steppers use high frequency modulation too. That could explain it. Err, it doesn't have to be micro-stepping at all. Any current limiting stepper drive will current limit with pulse modulation. I've been happily running 5 Volt motors from 24 Volt supply simply by controlling the current per step. The motors all run nice and cool.
Oh, the plasma generator is probably a big-Smile switcher as well. That would be a big RF noise source, but mainly only when it's active.
It has a 36 V DC power supply. PS V- as VM black lead. Red lead on either side of the lever switch reads 7.12 V DC, pressing the switch The voltage does not change on either wire. This is the same effect on two different switches. AC shows some lower odd numbers that don’t change either, pressing the button or not. It’s too late to reconnect the scope tonight.
To restate this, two wires are connected to a lever switch to the common and normally closed solder lugs. NO is no wire. Reading the voltage on either of the wires gives 7.2 volt DC whether the switch is pressed or not. It does not change. This is with black lead on the power supply V minus.
My problem is the machine zero reference resets randomly anywhere on the machine while jogging and typically jogging without the plasma cutter torch on. Typically the only way you can reset the machine 0/0 is by doing a home to hit the limit switches which forces machine to 00. I am believing some noise is coming from somewhere causing this weird thing. The limit switches worked perfectly for the home process and for stopping the machine at extremes. In addition to soft limits.
As stated before, I discovered that the painted lid on the controller was not providing any shielding and also the plasma cutter was literally right at the machine controller so I have created some space and shielded the lid and we’ll see how that works
Okay, reading that description, it sounds like the limits are triggering intermittently - Which forces a zero reference condition. And that's why you're looking at the limit switches.
What happens if only one limit is triggered? Is both axis still zero'd as a pair?
You can trigger the limit switches while jogging 100 times and it never resets machine zero so that’s why I don’t believe it as simple as noise on a limit switch wire because I can press those buttons and it just stops the machine, you clear the error pop-up And continue running with no zero reset. I can find only three ways to reset machine zero. One is to do a home and hit the switches. But that’s only during Home when you hit the switches that it will reset machine to zero. The other is if you disconnect the USB cable and plug it back in, but there is a gap of about five seconds for the machine to reconnect and the software to show ready again. But my problem is this machine zero reset is instant and there’s no gap. The other option is is you can reset the GUI and that will reset machine zero but that also takes about five seconds. My problem is I’m seeing a zero reset with no downtime. It just goes to zero randomly while jogging typically with the arrow keys to move to a new location. Hopefully tech-support will have answers for me tomorrow. I was just trying to solve it and not wait.
Thanks for the suggestions!!
Since there are no hot keys to do a machine zero I am ruling out it coming from the computer. It is as if they designed on the circuit board an input for doing a machine zero but they never connected it anywhere and noise is hitting that input, but that’s just a wild guess.
Unwired inputs won't trigger on their own.
USB is always a suspect. I never expected anyone to say USB was part of operating control.
To help reduce USB failings, at least a little bit, you need excellent grounding between the computer and the controller. The USB cable itself is not enough. You'll want to find a large grounding point at both ends and use a heavy multi-stranded copper wire and solid terminations. The sort you'd used for 100 Amp wiring. Sufficient to make the two circuit boards act as one board ground wise.
USB is just bad for controlling stuff. It's the common mode noise that badly trips it up. Maybe if there is electrical isolation on the USB, that might work. I dunno if such a thing even exists though.