Microstepping and steppers in serie skipping (tmc2100)

Hi
In an attempt to get smoother surfaces and less noise from my Folgertech FT-5 (MKS Gen 1.4, Cartesian with two steppers for Y axis) I have replaced the stepper controllers with TMC2100.
This take the 1:16 from the MKS and make it 1:256 in "spread cycle" mode, which was was very nice for a few minutes, until the Y axis started skipping.
I have tried changing the voltage from 0.9v to 1.5v, with no result.

Is there anything fundamentally incompatible between twin motors and micro stepping? I have considered changing the machine to corexy, but that is a lot of work just to get microstepping working, beside I need it working and plan do make a custom corexy next year anyway. Should I abandon and do with the noise and rice paddy surface?

Thanks for reading!

Edited 5 time(s). Last edit at 10/11/2016 02:02PM by orea.

What is the specification of your motors? For most motors used in 3D printers, it is better to connect two motors in series instead of in parallel. Then turn down the motor current to be the same as for the other axes. However, Folger used high-inductance low-current motors in some of their kits, and if you have those then parallel connection is better.

Edited 1 time(s). Last edit at 10/11/2016 11:56AM by dc42.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

Thank you for the reply, the Y steppers are in fact in series (I got it wrong), the Z steppers that are in parallel (they work fine enough without micro stepping)
As for specs they are NEMA 17 1.8 degrees, 5Kg/cm, 1.5A. The original stepper drivers are Pololu A4988, I tried voltages +/-30% what worked originally (I don't know the calculation, but that worked for X and extruder). Hence the only thing that differentiates Y is that there are two motors. Will try reducing the micro stepping once I find how to combine the jumpers.

Reducing microstepping won't improve the torque.

Are the drivers getting hot? If so then you may need to cool them with a fan.

Don't assume that the voltage settings you used on the A4988 drivers give you the same current on your new drivers. It depends on the sense resistors on both types of driver. By my calculation, if your TMC2100 driver has 0.1 ohm sense resistors then 1V will give a current of 1.28A.

Edited 2 time(s). Last edit at 10/12/2016 03:59AM by dc42.

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

You could have a similar issue as the DRV8825 drivers. Due to the nature of their operation, and how 3D printers actually run, the microstepping is actually causing problems. How accurately do think it is really operating making those 1/256 steps? It is most likely not making some, and other times making a few too many.

Some of these stepper motors really need to be run at 24V and decent amperage to get the best microstepping control.

1:256 microstepping is just WAY too much for an 8 bit AVR to process.

I believe the best mix is to get 0.9* motors and stick with 1:16 or 1:32 microstepping max (well, maybe experiment at 1:64 and 1:128 with 32 bit electronics). Oh, and switch over to 32 bit, especially for delta printers.


When running the DRV8825 at 24V and lower amps (I get more than enough holding torque under the max rating), the mixed mode causes microstepping issues. I had to hack them to run in fast decay mode to rectify the problem.
My future printer upgrades include a 32 bit controller and 0.9* steppers.

Edited 2 time(s). Last edit at 10/11/2016 07:16PM by MrBaz.

@MrBaz


"1:256 microstepping is just WAY too much for an 8 bit AVR to process." agreed if it was really doing that

The TMC2100 input is still full step, 1/2 step, 1/4 step or 1/16 micro stepping externally, but internally it uses Interpolation and generates 1/256 microstepping

This is how is achieves its "quietest Pololu compatible stepper motor drivers" status

Unfortunately MKS missed one bridge on the TMC2100. So the driver are never in spread cycle and will miss steps very easily. It is a very small pad near the enable pin.

---

Triffid Hunter's Calibration Guide

--> X <-- Drill for new Monitor

Most important Gcode.

@dc42
They don't get hot, but then it didn't run particularly fast either, with 60mm/s print speed and 150mm/s motion. Thanks for the mention about the Apmerage, I will check.

@MrBaz
Yes 24V would have been nice, it is planned for my next years "great custom corexy". I'll try to avoir sinking too much money in the FT-5 in the meantime, so at worse it will do with the old A4988 drivers. For the 0.9 degree steppers yes absolutely, rather have it finer grained at that level, even have the pulleys add 1:2 reduction.

@Wurstnase
That could explain a lot, they are indeed _very_ silent now. "Stealth chop" mode seem to have too little torque according to various comments. I thought the TMC2100 MKS edition where made to run as "spread cycle" mode when all board jumpers where off, but will try to find some more info. It would be nice if that was the solution.

Edited 1 time(s). Last edit at 10/13/2016 04:57AM by orea.

In the top right red circle you find two pads. One is for the enable pin and the other is to set CFG1 to GND. In your current config the pin is open.

[github.com]

---

Triffid Hunter's Calibration Guide

--> X <-- Drill for new Monitor

Most important Gcode.

Hi Wurstnase , thank you for the info.

My TMC2100 MKS seem to have two points joined in the bottom left corner, where as the same on your photo is open. It seems it is just "En/CFG6" though, so CFG1 to GND is indeed open.. I will try joining CFG1 to GND using the "little wire between the pins" method Thomas Sanladerer here: [www.youtube.com]

I had hoped MKS version could be configured with MKS jumpers, but apparently no.

Thanks again for your help.



Edited 9 time(s). Last edit at 10/13/2016 06:19PM by orea.

On my picture it is not open. There is a 0 Ohm resistor in it. You need to close the bridge and the wire!

---

Triffid Hunter's Calibration Guide

--> X <-- Drill for new Monitor

Most important Gcode.

I have bridged CFG1 and GNG, as well as removed the configuration pins going to the motherboard.
It is now in theory in speadcycle mode, I can confirm the microstepping on the print surface, and it is a bit more noisy.

It misses a lot of steps though, in both X and Y, which is strange as the voltages where first the same as when X (single motor) didn't skip in stealth chop mode.
Higher voltages made the TMC2100 fairly hot, while steppers just lukewarm.

Will try around with voltages and check the different versions of CFG combinations..



Edited 2 time(s). Last edit at 10/14/2016 06:01AM by orea.

If you buying a TMC2100 from MKS you must solder the small pad.
When you don't solder that small pad

the pin has no connection to the chip. So your soldered bridge from GND to that pin does nothing.

---

Triffid Hunter's Calibration Guide

--> X <-- Drill for new Monitor

Most important Gcode.

Quote
Wurstnase
If you buying a TMC2100 from MKS you must solder the small pad.
When you don't solder that small pad
[attachment 85591 addon.jpg]
the pin has no connection to the chip. So your soldered bridge from GND to that pin does nothing.

so does bridging the missing connection complete the spreadCycle mode configuration, or does one still need to jump CFG1 to GND in conjunction with the missing bridge?

lets recap:

do both CFG1 the CFG6 pad pairings on the bottom of the board need to be bridged?
does the CFG1 to GND bridge on the top need to be done as well afterwards?

these are the ones i ordered (shown below).
they seem to have CFG1 pads bridged. so now I just connect GND to CFG1 on the top side?



Edited 2 time(s). Last edit at 03/22/2017 06:54AM by Careless.

According to the picture the CFG1 pad isn't bridged. Only CFG6/Enable is bridged.
If you have to solder the wire too, depends on the mainboard.
Ramps for example has a pull down resistor on MS1 and without the jumpers in place, CFG1 is automatically grounded. But it won't harm if you do the wire hack.
Except you have a Ramps board where the jumper pins are connected underneath!
Then you'd make a shortcut with the wire, so beep it through before firing it up.

I'm using an MKS Gen 1.4 which is basically RAMPS/Arduino together.

So I would remove all the jumpers underneath the stepper driver socket headers and then bridge CFG1 pads, and bridge CFG1 to GND just in case and it should be 16/256 in spreadCycle.... right?

I bought a 600x USB microscope, so when they arrive, I can take photos as well.

the MKS TMC2100 with the single bank of green header pins came in this morning. Can confirm that it is as shown on the detail sample photo, CFG1 bridge is missing.

will attempt to solder soon.

wellllll... i wouldn't bother buying the MKS version of these stepper drivers.

with the smallest tip available from Weller and a decent soldering iron, good solder, and a 600x microscope, modifying the CFG1 jumper on the bottom is a crap-shoot at best.

I jumped 2 of the 4 I ordered, so to me, these are too risky to be worth it.

Is there any other way I can jump these two pins? The solder pads are no longer on the board. Where else can I jump them?

I don't even know what to believe anymore about these drivers.Some users say they have the clones working with jumping the CFG-1 to GND, no mention of the solder pad on the bottom. others say remove the CFG2 pin from the header completely, in conjunction with removing the jumpers from the MKS Gen1.4 boards (RAMPS based). others say not to? How do you even know if they're in spreadCycle mode? How does one check?

even the reprap wiki is short on information regarding the CFG2 pin removal.

not sure if I even want to bother dicking around with these things because of those stupid solder pads. why anyone would expect a customer to successfully solder those with ease with pin headers in the way is beyond me, but whatever.

buy the watterott versions, they're documented better and don't require microscoping soldering

Hi,

A problem:
Few months ago I've replaced X and Y drivers with MKS TMC2100 (single green bank of pins), and initially I've been amazed how silently it can print. Indeed I've removed all jumpers from motherboard for these drivers putting it into StealthChop mode.
These few months I've used the printer rarely and printed models with simple geometry at slow speed (25-30mm/s) and everything was fine.
1 week ago I've tried to print dragonfly from Thingiverse and every time I've got skipped steps. So I'm here. I believe it is a result of much more complex geometry of curves on a wings, where printer needs to accelerate/decelerate hotend constantly.

Driver's fix and motherboard check
I have similar tmc2100 drivers as topicstarter has with missing bridge between CFG1 pin and microcontroller. I've connected these two pads with small blob of solder successfully - inexpensive soldering paste with flux has been used.

As a next step I was going to connect GND and CFG1 pins with a piece of wire, however I wanted to be sure first, that motherboard doesn't have pull-down resistors for MS1-MS3 pins. I've read, that some RAMPS boards do have them, but in my case (CTC i3 Pro B printer with clone of Geeetech GT2560 board) no pull-down or pull-ups resistors have been found, neither any resistance has been detected between motherboard GND and MS1 pin (tested it in both powered off and on modes).

I didn't want to connect CFG1 and GND pins with a wire, because in case of placing these TMC2100 drivers to another motherboard it may be harmful for that motherboard (short circuit in case of VCC level at CFG1 pin). So I've connected CFG1 and GND pins with leaded (through-hole) 100KOhm resistor. It ties CFG1 to GND level and is safe for putting VCC voltage to CFG1 pin in other configuration.


A result:
From the powering up it has become clear, that drivers have switched from StealthChop mode - to SpreadCycle obviously. Permanent "buzz" noise has appeared when motors are enabled, printing noises are noticeably quite, than with previously installed A4988 drivers, however significantly louder, than with StealthChop mode.
No skipped steps has been detected meanwhile.

Next step:
I have printer placed at living room, so less noise is always good for my family. Therefore I want to test another approach to have StealthChop mode working, as I have had few months of successful experience with that silent mode. To do that I'll need to detach that 100KOhm resistor to put the drivers back into StealthChop mode, but that's easy to do and return it back to SpreadCycle.

I've read, that skipped tests may gone if:

1. Decrease acceleration for X and Y. Now it's set to 9000, I'll try 5000 first as Teaching Tech recommends at YouTube
2. Improve cooling of the TMC drivers - I have copper heat sinks for small graphic card memory chips, that look for me better, that original MKS TMC2100 heatsinks (I believe they are Aluminium). Also I'm going to install additional fan to blow an air directly to these heat sinks - that approach worked for greengecko at ultimaker forum
3. Raise current at motors adjusting Vref in safe range. Now it's set to 0.77V, however I'll take a look if I can raise it a bit higher.

Hope it will make someone's journey to eliminate steps skipping shorter.

Cheers,
Konstantin