Choice of TMC2660
-
Sorry to sound rude its not my intention but if this board is overspec'd this isnt really a problem is it? If it were under-spec'd then it would be. The only issue is its a bit more expensive, but then this isnt aimed at the budget end of the market. If you chose to put tmc 2100's on your ramps board you do so knowing that the machine you intend to run from them is within their capabilities, but you are making that choice for just one specific setup. I think its commendable that the choice here was for something a bit better, given that some users probably will be running nema 23's or would be inclined to when the thermal testing is complete and max driver current increased.
-
Didn't mean to suggest anything…
Design choices are frequently highly multifactorial and a series of tradeoffs are made.
As well as running a Duet Wifi I'm generally interested in the whole stepper / driver thing, have played with the Watterott TMC2130 (spi configuration) and TMC2100 breakout boards and spent time browsing the Trinamic catalogue of chips and dreaming.
Just curious what my betters were thinking.
-
Sorry I was curt. I've spent quite a bit of time experimenting with various "alternative" stepper drivers lately and it is a fascinating subject about which I too know very little. The TMC 2660's on the duet wifi are for me the best stepper drivers I've used so far out of a4988's, tmc2100's, LV8729's and 8825's. I am inclined to have a play with external stepper drivers at some point and see how much improvement there is there, or whether its just higher current handling that is on offer.
This is also of great interest https://odrive.xyz/.
-
Interesting link.
Thanks
-
Worth noting that the TMC2130 drivers have higher output resistance (1.1 ohms max, measured at just 0.1A) than the A4982 (0.86 ohms max, measured at 1.5A). Which means that even though both drivers are rated at 2A peak, the TMC2130 drivers will generate more heat and need more cooling. It's also worrying that the TMC2130 output resistance is specified at only 100mA instead of at a sensible working current,
In summary, the TMC2130 drivers are worse than the popular A4982 and A4988 drivers from the point of view of the cooling they require and the maximum current they can take in practice.
-
That's an amazing difference.
In the TMC2660 datasheet they don't specify currents at all for the mosfet resistances cited, but give a range of temperatures instead: if I read it correctly the resistances are tiny by comparison (<20% TMC2130) with all of the aforementioned drivers. Great choice!
Having just milled a heatsink for the Watterott -TMC2130 board I'm playing with, was beginning to wonder where all the heat was coming from.
Feature-wise the TMC2130 does have the ultra-silent Stealthchop mode on top of the Spreadcycle it shares with the TMC2660. But nobody is complaining about the noise!
-
Another trick the TMC2130 has hidden away is that it can interpolate to 256 microsteps from as much as a full step. (TMC2660 is constrained to interpolating from 1/16 microstep input.) Which means if your processor is limiting your top speed (try not to think about that too much!) you could set the driver to accept 1/8 microsteps and yet retain all of the smoothness of 256 interpolation.
-
I've installed TMC 2100's on my microdelta with nema 14 motors, running vref 0.5v (0.5A Imax) the motors are warm and the drivers are warm with heat-sinks (properly mounted to the PCB) and a 40mm fan blowing from the side, but at 0.7v which I have my extruder motor set to the motor and driver is very hot.
Im also trying LV8729 drivers, which are very stable, run quite cool, (they have auto 50% current after 0.5sec without a step pulse) limited to 1.5A, but they must do some sort of interpolation as my flex3drive extruder is running 1/4 stepping and it quite quiet.
-
A version with the duet with the tmc2130 and the ability to do sensor-less homing would be a godsend IMO.
-
This is also of great interest https://odrive.xyz/.
Great!
For the next release of the Duet board, it would be nice to split the main board and the driver board, so we can choose what type of driver we want…
-
It would certainly offer great flexibility if the driver module was just that, a module rather than integrated. There could be a choice of tmc2600 module with 5 drivers and the ability to use more than 1 as required or a plug in driver module or a breakout module for external drivers. I believe there will be a modular architecture with a serial bus+power link cable between them. The versatility will be massive, and costs for those who need a simpler system might be lower as you would not need all of the modules.
-
On the other hand, developing such versatility will have a cost…
But we see more and more non-classic setup, where such modularity would be interesting. For example, a laser engraver does not need all FET/fans... They could be on a dedicated board too.
-
Well, it's not like the board is closed source… So if money isn't an issue, one could always do a custom run with the needed mods. Or just get an expansion board and remap the motors to whichever drivers necessary.
Even if you don't need the FET outputs, it's still much cheaper to get the stock board. AFAIK, cost of small components like transistors is negligible anyway.
-
A version with the duet with the tmc2130 and the ability to do sensor-less homing would be a godsend IMO.
The Duet already got sensor-less homing?
-
Yes the Duet already supports sensorless homing. See the wiki page about it.
-
The next generation Duets will be more modular. However, modularity comes at a cost because of the additional connectors and multiple circuit boards. So we still need to put enough on the main board to be sufficient for a large proportion of users. Currently we think the main board will have 4 stepper drivers and 4 heater/fan outputs, with expandability in units of 2 stepper drivers; but we haven't finalized it yet.
