search

bdring / Grbl_ESP32_MPCNC_Controller

A Grbl_ESP32CNC Controller for MPCNC
86 stars 25 forks source link

more flexible support for stepstick modules #2

Open davidelang opened 5 years ago

davidelang commented 5 years ago

In the next revision, could you provide support for more advanced stepper modules?

this would require that the three config pins have a way to set them individually or have them connect to spi (for the most advanced features)

If this could support the tmc2130 modules, you could potentially eliminate limit switches.

I'm thinking of a 3 pin jumper for each of the config pins, high, low (one side connected or removed) or spi (other side connected)

bdring commented 5 years ago

I have other Grbl_ESP32 boards with SPI driver support. It can use a lot of extra I/O. You need a Chip Select pin for each driver.

I have a test board that uses one CS pin for all drivers. It works, but requires modifications to the standard library. Those mods are write only at this time. It does allow you to setup all features.

bdring commented 5 years ago

Here are some photos of that controller.

20190731_165601

20190805_134430

dvisser-zz commented 5 years ago

@bdring can you share the artwork for this board variant please? possibly as a new branch called "trinamic" or "spi" or similar? even just the schematic would suffice, as I dont use diptrace (mainly KiCAD or Altium).

dvisser-zz commented 5 years ago

and are you still daisy-chaining the trinamic modules? (ie. all 4 of them use a single CS line?)

bdring commented 5 years ago

OK. I am out of town, so it will take a day or so.

I have a few extras assembled. I would be willing to part with one, but only to seasoned programmers, that can help move this forward.

dvisser-zz commented 5 years ago

I dont expect a handout of hardware mate, please share the design and I can spin my own board up if need be, happy to assist with firmware if needed also.

bdring commented 5 years ago

Attached is the schematic.

I am away from my office until Aug 20th on vacation. I do not have time to create a proper repo for the controller.

I can send the gerbers via email or Slack, but I do not want to publicly post them without well documented repo.

The controller uses this branch of Grbl_ESP32. It requires this fork of the TMCStepper Library for daisy chain support.

Currently you can only write to the drivers. I have some code to read, but I have not tested it at all and I am sure it does not work yet.

To save pins, the SPI data lines are shared with the SD card. The SD has its own CS and the drivers share a CS. You must install 4 drivers or jumper the data through any missing drivers.

SPI_4Axis_V1_Schm.pdf

dvisser-zz commented 5 years ago

OK not a problem, I will take a look. Thank you for sharing.

dvisser-zz commented 5 years ago

What are the currently "working" methods of receiving grbl strings? wifi-serial, usb cable, read g-code file from sd card, etc? Cant really tell from what I have read (without reading through the source, which I will be doing) where you are roughly up to?

dvisser-zz commented 5 years ago

nevermind the above, I read the WiKi properly :-)

bdring commented 5 years ago

We treat the communications methods as separate channels. If Bluetooth asks for status, only Bluetooth gets the status.

The serial port is treated as the debugging port because it is always there and survives a reboot. You do not loose the connection during reboot. A lot of startup info and debugging is sent there.

See grbl_send(...) and grbl_sendf(...) for more info.

dvisser-zz commented 5 years ago

OK Thank you

bdring commented 5 years ago

@dvisser

Repo created

https://github.com/bdring/4_Axis_SPI_CNC

davidelang commented 5 years ago

I was hoping for something with 5 axis support (looking at the lowrider 2)

bdring commented 5 years ago

Maybe you should offer a bounty :-)

dvisser-zz commented 5 years ago

thank you bdring, I will check it all out now.

dvisser-zz commented 5 years ago

I have designed my own PCB to use your TMCStepper branch of firmware. It can be found here if you are interested, I have referenced you accordingly. https://github.com/dvisser/NodeMCU-ESP-32S-3-Axis-CNC-Controller

bdring commented 5 years ago

That looks good.

While the output of the spindle is PWM, it is on one of the ESP32s 8-bit DAC pins. With a small firmware change, you might be able to simplify your spindle circuit a little.

dvisser-zz commented 5 years ago

it is still a work in progress, but thank you. I am very much used to Altium Designer (over 10years experience) and still learning KiCAD.

I wasnt aware we had a DAC peripheral, that does sound interesting and is worth looking into, it would remove the need for the PWM to analog DAC IC I was planning to use.

dvisser-zz commented 5 years ago

please could you invite me to your slack workspace (never used slack, but keen to see what it is like)

email address removed after slack invite received.
bdring commented 5 years ago

The option to output DAC or PWM might make your board more flexible to use other speed controllers.

invite sent

dvisser-zz commented 5 years ago

indeed, so might be worth keeping the current hardware configuration I guess.

Just joined slack also. Thank you.

davidelang commented 5 years ago

what sort of bounty would get attention?

dvisser-zz commented 5 years ago

@davidelang ??

davidelang commented 5 years ago

you mentioned a few posts up that I should consider posting a bounty to get a 5+ axis controller that would support the smart stepper modules, how big a bounty are you talking about?

I understand the argument that homing isn't that important for a CNC, but for something like the MPCNC or the lowrider2, it's especially easy to get the two sides out of sync with each other and have the machine be racked. So having the ability to run against a stop as a squaring routine (potentially slightly different than a homing routine) is something I find attractive.

I have this problem on my dual Y 3d printer where the two sides can get out of sync when I transport it, and the lowrider seems especially sensitive to this problem.

so something that can drive 2130 or 2209 modules so that it can just run against limits rather than having to wire endstops is attractive. (long endstop wires close to electrically noisy routers isn't a great match)

David Lang

dvisser-zz commented 5 years ago

If using 2130's or 2209's you could use stallguard whilst homing/crashing to square up the axis. Pretty simple. I can assist with code if needed, but this is bdring's party :-)

davidelang commented 5 years ago

On Mon, 2 Sep 2019, Dan Visser wrote:

If using 2130's or 2209's you could use stallguard whilst homing/crashing to square up the axis. Pretty simple. I can assist with code if needed, but this is bdring's party :-)

That's exactly what I'm hoping for, but before you can write the code for it, you have to have a way to communicate ith the 2130/2209

David Lang

dvisser-zz commented 5 years ago

@bdring I got my PCB's today!

https://ibb.co/B3tv83b

dvisser-zz commented 5 years ago

@davidelang bdring's firmware (Tmcstepper branch I believe) already talks to daisy chained 2130's. Only one way at present. I have code that I have forked that does both ways. I bet he isn't far off either. I could always put a pull request in once my code is polished anyway. But you could use bdring's FW and simply set the diag1 pin to toggle on trigger of stallguard and connect that to your relevant axis limit SW input pin.

bdring commented 5 years ago

My branch only writes in daisy chain mode. I worked a little on daisy chain reading, but never finished it.

Setting up stallgaurd is tricky, especially if you have long stepper wires. You are trying to measure power losses in the motor coils. If you have wiring losses, that needs to be "tuned" out. Also, stallgaurd does not work well (false or missed stalls) at very low or very high speeds. You typically need to tell stallgaurd the speed window where it should trigger the stallgaurd signal.

You need to be able view a fast stream of stallgaurd data to do that tuning. Reading is needed for that. Also Grbl is not a very good platform for tuning. You want a program that runs the motor at a constant speed for long periods of time and has the ability to adjust the speed and tuning values on the fly.

I ended up making a separate program for tuning and just loading the tuning values into Grbl.

davidelang commented 5 years ago

On Mon, 2 Sep 2019, bdring wrote:

Setting up stallgaurd is tricky, especially if you have long stepper wires. You are trying to measure power losses in the motor coils. If you have wiring losses, that needs to be "tuned" out. Also, stallgaurd does not work well (false or missed stalls) at very low or very high speeds. You typically need to tell stallgaurd the speed window where it should trigger the stallgaurd signal.

stupid question, would it make sense to put the stepstick module closer to the motor and have the long wires on the CPU side of the module?

You need to be able view a fast stream of stallgaurd data to do that tuning. Reading is needed for that. Also Grbl is not a very good platform for tuning. You want a program that runs the motor at a constant speed for long periods of time and has the ability to adjust the speed and tuning values on the fly.

I ended up making a separate program for tuning and just loading the tuning values into Grbl.

interesting, thanks for the info

bdring commented 5 years ago

It is probably not better to extend the signals to the driver. Those are sensitive pins and are easily damaged with ESD that could be picked up over long wires.

My point was....You will need to tune any setup. so you need the read capability and firmware that helps you do it.

On my wall plotter project, one small NEMA17 was right next to the driver and one was about 8 feet away from the driver connected by a CAT5E (not ideal) cable. The two motors had significantly different tuning, but they both detected stalls fine.

One other thing to keep in mind. Stallgaurd will, at best, detect to about the nearest 1/2 step. On a machine like an MPCNC using 20 tooth GT2 pulleys (40mm/rev) , you are looking at +/0.1mm detection at best. A simple microswitch will do much better.

If you are just homing the machine before a job, that is fine, but if you are looking for reputability for fixtures, I would be concerned.

davidelang commented 5 years ago

thanks, I'm also thinking of the 8' range, so good to know that works.

I'm primarily thinking in terms of squaring up the machine. If it's parallel to 0.1mm, that's much more accurate than you will get from a tape measure. It's probably more accurate than your positioning of the microswitches as well :-)

I don't think that you are really going to be able to do any better than a 1/2 step in any case (yes, a microswitch can detect smaller distances, but you aren't going to be able to control the machine accurately to that precision)

David Lang