Marlin 2.0 bed PID: autotuned, but won't heat up

I'm trying to enable bed PID in Marlin 2.0. I have PIDTEMPBED enabled in Configuration.h, and autotune yielded these constants:

  #define  DEFAULT_bedKp 206.42
  #define  DEFAULT_bedKi 19.09
  #define  DEFAULT_bedKd 557.90

However, when I dial in a bed temperature, the heater never switches on.

I also have PIDTEMP, PID_EDIT_MENU, and PID_AUTOTUNE_MENU defined, but this only provides access to the hotend PID values, not the bed PID. I also tried enabling PID_PARAMS_PER_HOTEND (even though I only have one) to see if that'd help. It didn't.

I'm currently using a Re-ARM and RAMPS 1.6 combo, though I had been using an Arduino Mega knockoff until yesterday. ISTR trying to enable bed PID in the past on Marlin 2.0 and having problems, though I once had bed PID working on Marlin 1.1.9.

I tracked down a general bed-heating problem to a power-supply issue. I've been running my A8 off an ATX power supply. I ran the printer off the 8-pin ATX12V connector (and a Raspberry Pi off the 5VSB output, but that isn't relevant here). With the stock bed, all was well, but I think the connector was overloaded when I upgraded to a 200W bed. On an attempted test print today, I saw that the bed-heater idiot light on the RAMPS was lit up, but not the corresponding indicator on the MOSFET board. A voltmeter across the MOSFET signal input showed the expected 12V, but I was only seeing about 3-4V on the MOSFET power input. Even when the MOSFET's idiot light came on, the power input only had maybe 10.2V on it. That's a sizable drop.

I swapped in an HP server power supply, with a PCIe breakout board that previously powered a mining rig. When I disconnected the old power supply, the wires going into the ATX12V connector were a bit scorched, and the connector broke apart when I tried to separate it. I now have a 6-pin PCIe cable with three pairs of 16-ga. wire feeding the MOSFET and one pair of a second cable driving the rest of the printer. Voltage drop at the MOSFET is no more than 200 mV or so.

Also, with the new power supply in place, bed PID is once again working as expected. I was about ready to blame a possible bed-power regression in Marlin for my problems, but that turns out to not have been the case.

Now I need to fab up a mount for the new power supply. (A new mount for the electronics is already printing.)

Quote
salfter
I tracked down a general bed-heating problem to a power-supply issue. I've been running my A8 off an ATX power supply. I ran the printer off the 8-pin ATX12V connector (and a Raspberry Pi off the 5VSB output, but that isn't relevant here). With the stock bed, all was well, but I think the connector was overloaded when I upgraded to a 200W bed. On an attempted test print today, I saw that the bed-heater idiot light on the RAMPS was lit up, but not the corresponding indicator on the MOSFET board. A voltmeter across the MOSFET signal input showed the expected 12V, but I was only seeing about 3-4V on the MOSFET power input. Even when the MOSFET's idiot light came on, the power input only had maybe 10.2V on it. That's a sizable drop.

I swapped in an HP server power supply, with a PCIe breakout board that previously powered a mining rig. When I disconnected the old power supply, the wires going into the ATX12V connector were a bit scorched, and the connector broke apart when I tried to separate it. I now have a 6-pin PCIe cable with three pairs of 16-ga. wire feeding the MOSFET and one pair of a second cable driving the rest of the printer. Voltage drop at the MOSFET is no more than 200 mV or so.

Also, with the new power supply in place, bed PID is once again working as expected. I was about ready to blame a possible bed-power regression in Marlin for my problems, but that turns out to not have been the case.

Now I need to fab up a mount for the new power supply. (A new mount for the electronics is already printing.)

Interesting. I was already looking into PSU's the other day and found out that mining rig PSU's are becoming cheaper in the 2nd hand market now. These can usually pump out 700-1000 Watts and seem to be candidates for higher-power beds. I didn't buy one eventually since I found out that my old PC had a really quiet ATX PSU that can deliver 450Watts and this should be enough for the machine I'm currently trying to operate. I also rigged the other old ATX supply I found in my storage for a pen plotter so I now have more than enough to work with.

Oh, before I forget, my two running 3D-printers both use Hitachi HS400-08C-S PSU's which are also server PSU's. These work brilliantly.

Edited 1 time(s). Last edit at 06/10/2019 04:55AM by Ohmarinus.

---

http://www.marinusdebeer.nl/