search

bigtreetech / BIGTREETECH-SKR-PRO-V1.1

Aiming at some problems existing in 3D printed motherboards in the market. Bigtree Technology Co., Ltd. launched a high performance 3D printer master board with STM32F407ZGT6 as the core controller, BIGTREETECH-SKR-PRO-V1.1.
275 stars 222 forks source link

Limit switch and TMC2209 driver #47

Open theWituch opened 5 years ago

theWituch commented 5 years ago

Hi, I have SKR PRO board with TMC2209 v1.1 stepper drivers and when i connect driver into board, suitable limit switch seems stop working. Without driver in slot, led shows switch state (led is on when switch is opened, and off when closed). With driver in slot, nothing happened.

I checked schematics of board and TMC driver and i think TMCs GND pin is connected to limit switch ouput signal. This signal is connected to uC too, so pin is always pulled down and will never work.

Im right? Is any solution for this issue?

Meddleglitch commented 5 years ago

Aha! I've been having this issue too, and I hadn't considered that my 2209s were the problem, but now that you mention it, I did happen to notice that the limit switch for my E2 driver (which doesn't exist / isn't installed) works fine, while the other five won't work even if I use an external pullup (oh jeez, I sure hope I didn't break anything! >_<)

..........wait.......... hang on a minute... Are WE the problem, here!? image Is THIS why they tell us to castrate our poor 2209 drivers? o_o Capture

omahena commented 5 years ago

Hmm. I have the TMC5160 drivers. I had to add the stallGuard pin to actually be able to use the sensorless homing feature. However I only get into trouble when I use a stallGuard feature AND try to use the endstop switches as well. The endstops worked fine if I disabled sensorless homing even with the diag pin of the steppers connected to the board.

But... TMC2209 might be a little different and I suppose they wouldn't dedicate a part of the manual to tell you to cut that pin off.

Btw... You don't have to cut the pin. You can also just heat it up to melt the solder and push it through the board so it doesn't make contact with the socket below it. When you need it again... push it through again to the bottom side. Not sure how many cycles of this modification your connector pin can take though.

Personally I find the stallGuard based sensorless homing overrated. Or maybe it just doesn't work well on my system. With a lot of tuning of the sensitivity parameter I still get grinding. Especially for example when homing an already homed hotend. I only keep it enabled because it was so hard to enable it and it refused to work for weeks.

Meddleglitch commented 5 years ago

Yeah, I don't think they can work together unless you can somehow assign two different pins to the same endstop, because it seems like the TMC drivers connect that pin to ground, and whichever way I set up my endstops before I realized what was going on (or rather, before I remembered what I had read about weeks in advance and then forgotten to actually do when the time came) I couldn't get the mechanical ones to trigger/un-trigger with that extra ground wire connected, even with a jumper cap between 3.3V/GND & the sense pin. I saw the value rising and falling when I checked it with a voltmeter, but only by about 0.3V in either direction -- not enough to trigger anything.

I suppose if you really wanted to use both at the same time (assuming that it's not currently possible to configure a whole extra set of endstop pins and wire the mechanical ones to them instead), you could have the standard normally-closed endstop (+pullup) signal pass through a logic-level PNP MOSfet before draining to ground (it shouldn't matter which side of the endstop switch it's on) with the normally-grounded TMC StallGuard pin (clipped/pushed through/not connected to the board) controlling the gate, so that when either it goes high or the switch gets pushed, the circuit breaks just like it would with only one or the other connected.

(I doubt the few nanoseconds of rise (wait, fall?) time would significantly affect the StallGuard's accuracy, but then that's the whole reason for the endstops in the first place.)

AFAIK, that should properly isolate the two switches while keeping them on the same circuit, although I'm not entirely certain whether or not the StallGuard would kick in & mess with the repeatability of the mechanical endstops during a normal home operation... but I do strongly suspect that the spring-action switches would trigger "long" before the motors would ever actually encounter enough resistance & stall out (as long as Marlin's endstop noise filter isn't active). Like, entire microseconds beforehand... Which is basically forever, if you're a microcontroller.

I might just try this myself sometime, if the day ever comes where I actually have nothing better to do & don't have a dozen other neglected projects cluttering up my desk. <_<