ALWAYS disconnect power before working on your printer

Laser8302

Horror story...

In my efforts to do dual extrusion, I had a jam in the heat break. So I had to disassemble the hot end. Nozzle, heater carriage and temp sensor out, remove everything clean and reinsert. Well I like to use my LEDs to light the printer which means +24V is live at the heater, and there's an active ground connection through the temp sensor... zap.

Now my SKR board is non responsive. So got one on order.

So from now on I'll disconnect the temp sensor or just turn off the printer and use a work light...

In short. DA HAI is right. ALWAYS unplug your printer before performing any work.

Laser8302

I DID IT AGAIN...

I was going to change my nozzle out while printing some face shields. Going from a 0.4 -> 0.8mm
My wrench was apparently long enough to reach over to the thermistor and short it AGAIN. A 2nd SKR 1.3 is dead.

I'm going with a MKS SGEN L V1.0 this time... Similar price and same processor. Maybe I'll have better luck? I would think that just grounding a thermistor input wouldn't cause an issue like this...

Stephen Todd

You are shorting the smoothing capacitor to ground and likely making a spark as I think the capacitor will be 100 uf and charged to 12 volts

Normal electronics design would have a resistor or choke where Vout is, and a second smoothing capacitor, and maybe also one to three small value capacitors to take out any high frequency - that would stop any surge or noise getting past this section

Laser8302

So I know for a fact that the SKR v1.4 has improvements in that area:

I'm just done with BTT/SKR...

Laser8302

Seems Duet has found this problem and solved it:

How to destroy a Duet board:
Create a short between a hot end thermistor connection and an extruder heater or fan connection. On many electronics - including RAMPS and Duet 0.6 - this will kill the processor. However, this type of short is surprisingly common in 3D printers, so you're out of luck because we've taken precautions against it. The most damage you are likely to do is to blow the VSSA fuse. Although that fuse can be tricky to replace, it's quite easy to bypass with an external fuse. So if you are trying to destroy your Duet, blowing the VSSA fuse doesn't really count. Oh, and to spoil your fun, in PCB version 1.02 and later we use a self-resetting fuse.

Stephen Todd

If you could find a schematic diagram showing the thermistor input circuitry of the Duet, it would be easy to make a small circuit board and add it to any controller board between the thermistor leads and the controller board input plug

Laser8302

Found it. (Thanks Duet 3d for keeping the open-source idea alive!)

https://www.belfuse.com/product/part-details?partn=0ZCJ0020FF2E


One side of the thermistor is connected to MCU (through filter capacitor, 4.7k/10k voltage divider) then they each get their ground though this one little fuse. a 200ma PPTC fuse that resets after the offending voltage is removed. 0.2A constant rating, 0.4A trip. They make radial lead PTC fuses, I would just be worried about the time to trip vs the SMD version. The SMD version is open circuit if 8.0A is present for 0.1s vs the radial lead version trips with 1.0A present for 2.2s. The physical size of the devices would also tell you because the axial lead versions are much larger than the SMD.

Wish I could find the circuit diagram for the SKR board. They aren't available as far as I can tell from their github page.

Laser8302

More insight.
SKR v1.3:


SKR v1.4:


MKS SGEN L:


Of these three, the SKR V1.4 has "better" protection with a varistor, but that will only protect if voltage is above 25V or an ESD event, not a short circuit to 12/24V and at that, only on the input line. (The MKS GEN L V1.0 has the same circuit as the SGEN L... I'm guessing the ATMEGA chip is better at rejecting over-current or over-voltage events since it runs on 5V vs 3.3V for the LPC1768.)

So that all being said, assume there is a 10kohm resistor on the MCU input that will reject the +24V down to 2.4mA or less. (Not sure why the MKS SGEN L goes through a 100ohm resistor bank...) We are only concerned about the ground side of the circuit, which isn't really protected in any of these cases, it is connected directly to ground vs the case of the Duet wifi board going though a PTC.

TL:DR
If you are working on your hot end, be careful of the thermistor shorting out or swap it to a cartridge style
Alternatly, you can put a PTC in-line with the GROUND side to protect your control board.

Stephen Todd

Lots of good info there - will study it tomorrow, but pretty much as I suspected - all passive, and no voltage following/feedback, which is normal protection

Stephen Todd

This now looks like a completely different problem than what I thought it was

Looks like the "ground" is not machine ground which would sink the short voltage - but a floating ground, and the stray voltage goes though the floating ground to destroy various input/output pins of the processor that have a path to ground

Surprised about the use of a fuse - in electronics, a fuse is not normally fast enough to prevent damage - it is to prevent "additional" damage being caused to the other components when a component fails

Laser8302

On the SKR 1.3, the ground is just on the other side of an inductor.
On the SRK 1.4, it's behind FB3, which I'm having a hard time finding the BOM to tell what the device is. (Still, not ideal)

The fuse in this case is a PTC fuse. If you've ever shorted out your USB port, it has a PTC in it as well. It is fast enough to trip and prevent any damage to your PC USB port, so it should be sufficient for this purpose. The benefit to a PTC is it will limit the current going through it to the nA range until the output is shut off, then it will reset itself without user intervention. If we used a standard board-mount fuse we'd have to open the chassis to change the fuse every time the thermistor shorts out.

So that being said, I'm going to take the same PTC fuse from the duet board and put it on a fiber board with a wire inline with the thermistor ground. From mouser.com, the fuse is $0.11 and the board is $1.09 with $7.99 shipping... So I ordered 25 fuses. Anyone wanting to do this mod could just solder to the wire, I'll send the fuse via regular mail for $1 each

Stephen Todd

Sounds like a good and simple solution

I can't think of an alternative

Soldered inline and heat-shrinked is what I'd do, but an inline plug-in adaptor board would be better for a lot of people so they can just plug it in between the thermistor lead and the board

Laser8302

Here's what I did: 

Seems to work. I had to disconnect the pin trace from both sides of the PCB.

Samuel Pinches

Nice!!

Laser8302

So I found one more item that was concerning.

When I upgraded my heater cartridge from 30W->40W I thought it didn't matter where + and - went. WRONG. I measured +24V on the heater block when the block was heating up!!! (black probe on "-" input voltage on the control board, red probe on the aluminum heater block.)

Not only is this bad for the block, it could potentially short the power supply should the bearings ground the heatsink to the case ground! It was also the probable cause of the thermistor shorting out to the +24V rail when I went to change nozzles. Swapping the heater wires at the board now gives 0V at the block.

Samuel Pinches

glad that one was caught!!

Stephen Todd

Now that is a logical explanation - I did consider the thermistor wires shorting out to a positive voltage on the block, but discounted it as not possible

So, sounds like one wire of the heater cartridge is connected to the metal case of the cartridge

Stephen Todd

I tested 5 heater cartridges I bought as spares from aliexpress - the wires are not connected to the cartridge metal case, but on 200 mega ohms resistance range of my digital meter I did detect some leakage - and it varied from cartridge to cartridge - but I do not have and specific leakage detection equipment

As the JGaurora uses plastic parts in the gantry, is the head not being grounded? - if not, then some low current voltage could be on the head when the heater cartridge is getting power

Gandy

On my JGA5, the printhead is electrically isolated from the chassis. The Z-rods and spindles do have contact with a few kOhms but the plastic X-gantry parts provide no contact.

I have recently installed a E3D-V6 with a E3D heater cardridge and checked for isolation of the wires from the cartridge housing/heater block. On the scale of my digital meter (supposedly <=80MOhm) there is no detectable leakage.

I also checked the cartridge on the now shelved original JGA5 printhead which also has no detectable leakage. I did notice however, that the isolation on both wires coming out of the cartridge is broken, exposing the bare wires. This could easily introduce leakage currents.

Stephen Todd

Problem is, the thermistor input section of the controller board may not be grounded to the printer chassis - it could be a common/floating ground - i.e. various components can be grounded together, but they may not be "earthed" to the printer metalwork and the wall socket ground

You could measure from the board components to the heater block - I have not done it, as I assumed until very recently that controller board was grounded to the printer metalwork, but that may not be the case for all the negative/black wires connecting to the controller board

Laser8302

So my A5 is one of the old school ones with the metal CNC'd gantry supports (I know, I'm super lucky right?) But my X-axis tension has to be done by removing the pulley...

That's why I was concerned with grounding through the linear bearings.

Richy_T

I managed to kill my A1 mainboard by grounding the thermistor. I'd been messing around with the PTFE tube to get some broken filament out and was pushing all the wires back into the cable wrap and pop. Seems like the design is not great and it's really easy to contact the wire bits of the thermistor to the grounded hot end and there you go. I wrapped it with a little kapton tape this time but I'm still not happy.