DRV8825 vs TMC2100 vs THB6128

One would think different, but i definitaley see the difference between 32 and 16 microsteps in terms of print quality.
When looking at a flat wall my extruder (BulldogXL/E3D V6) has much less "ripples" which come from the pulsing of my extruder stepper.

the watterodt trinamic TMC2100 drivers have "256th" microstepping which is essentially 16th microstepping multiplied by factor of 16 to give 256th pseudo-microstepping.
In the usable mode (not the totally silent mode) every step is multiplied by 16 to give 256 microsteps with an accuracy of 16th microstepping but with accustic noise saving of real 256th microstepping.

In stark contrast to this the RAPS128 give real 128 microstepping with both the resolution of 128 microstepping plus the little noise one 128th microstep makes. I don´t think that I could differentiate 128th microstepping from 64 to 32 in terms of print quality, but as I already said 16 to 32 make a differnce IMHO.

So, the question is, what stepper drivers should I use?

Since I already use DRV8825s, and I do see a difference in print quality, going for the TMC2100 will result in less noise but will actually be a step back when we talk about microstepping... Or will the multiplied step ratio also result in finer stepper pulses and therefore a better print appearance?

Another thing will be upcoming, not yet supported features.
I think both (TMC2100 and THB6128) drivers have the feature to report missed/skipped steps to the firmware. This is something very cool an I think that future releases of smoothie/marlin will support this.


ok, so what to go for?
stay with the DRv8825s which are already a huge improvement to the 4988 allegros or invest some 60-80€ and try one of the new drivers? And If so, which one would you choose?

Keep in mind when you go to higher microstepping ratios torque goes down and your print speed may have to drop because the controller has limited steps per sec. Arduino/RAMPS can only spit out 40k pulses per sec max, combined with 256 ustepping and you find you can't turn the motor very fast. Smoothieboard can go as high as 100k pps which helps a little. Increasing microstepping ratio above 16 in a 200 step per rev motor does not increase resolution, only smoothness of rotation.

If I understood correctly, the TMC2100 is actually a hybrid between 16 and 256µSteps/Step. When even 8-Bit electronics give out 1 step as set with the M1,M2,M3 Jumpers to 1/16th microstepping the driver internally gives out 16 µSteps at 1/256th resolution at once.
This is the said mix between 1/16th and 1/256th, since the controller and firmware assumes that the driver can only do 1/16th microstepping. In reality however the motor gets 16 step pulses in 1/256th microstepping.
So theoretically every electronics that can handle 1/16th µstepping fast enough (all that I know of) can do the pseudo-1/256th as well.

Quote
the_digital_dentist
Keep in mind when you go to higher microstepping ratios torque goes down

I know.. there has been a graph somewhere here on the reprap site showing this fact (x-axis torque, y-axis microstepping setting) do you know where to find this graph?

Quote
the_digital_dentist
Increasing microstepping ratio above 16 in a 200 step per rev motor does not increase resolution, only smoothness of rotation.

Do you have proof for that? Not to piss you off, but I find this fact very interesting.. trying to understand how motors work in detail.

Both Oriental Motor web site and Gecko Drives web sites have good explanations of stepper motor basics, including explanations of drive techniques.

See here for an explanation of microstepping and resolution: [www.micromo.com]

On a different site I read that motor steps are not exactly 1.8 degrees- they vary typically 5%. When the microsteps get smaller than the 5% variation in step position, the resolution can't improve because you're "down in the noise". 1/16 is >5% but 1/32 <5%.

Edited 2 time(s). Last edit at 05/12/2015 10:04AM by the_digital_dentist.

Hi guys,
I've installed today on my printer drivers tmc2100
I have a corexy and as electronic I use an arduino mega 2560 with ramps 1.4
I installed the drivers only for x and y and I must say that the improvements in terms of smoothness and quietness of mechanism from the previous configuration (with driver drv8825) is outstanding !!
I also put on the drivers heat sinks and of course I installed a fan!
I immediately printed an empty cube 20x20x10 with excellent results and with good speed!
But in subsequent prints something really unexpected happens to me !!
layers are shifted as if there were losses of steps or even slacks belts and pulleys!!!!
But this is not my case, the mechanics are adjusted very well, indeed, for the avoidance of doubt, I reinstalled the drivers earlier (drv8825) and I did the same prints have more this problem!
So I'm wondering, what's the problem !!
drivers are soldered and installed like picture below
[cdn.instructables.com]

Edited 2 time(s). Last edit at 05/22/2015 12:51PM by mdcompositi.

I don't have any of these new drivers but I assume you can adjust the current to the steppers. If so, what is the current you have set for your steppers? What was it previously set to with the DRV8825's?

Quote
AndrewBCN
I don't have any of these new drivers but I assume you can adjust the current to the steppers. If so, what is the current you have set for your steppers? What was it previously set to with the DRV8825's?

You think that this is due to the current?
I have 0.75 Vref for x,y drivers when I use DRV8825(and all works well)
I have Set the same Vref for TMC2100

I believe the formula to calculate current is slightly different for the TMC2100 drivers, in this video they have Vref set @ 1.1V:
[www.youtube.com]

Also, did you configure them for spreadCycle mode or otherwise?

Quote
AndrewBCN
I believe the formula to calculate current is slightly different for the TMC2100 drivers, in this video they have Vref set @ 1.1V:
[www.youtube.com]

Also, did you configure them for spreadCycle mode or otherwise?

Poblem solved about shifted layers!!
I've set Vref @ 1.1 and now all work very well!!
I suggest you to test TMC2100!!

Quote
mdcompositi

Quote
AndrewBCN
I believe the formula to calculate current is slightly different for the TMC2100 drivers, in this video they have Vref set @ 1.1V:
[www.youtube.com]

Also, did you configure them for spreadCycle mode or otherwise?

Poblem solved about shifted layers!!
I've set Vref @ 1.1 and now all work very well!!
I suggest you to test TMC2100!!

Good! I definitely will test the TMC2100 drivers soon!

I read the page [www.micromo.com] but I am not entirely sure that I understand.

Is the reduced torque only a problem for holding the given position (microstep) and for reverse moves? I mean if the sum of the micro steps is still the same (and better) as a normal step?

For example;

With 16 microsteps, then each of my microsteps will have 9.8% of the normal full step incremental torque.

Let's say I apply a high load, say I apply 99.9% of a normal step - So without microstepping my stepper would barely have managed and it would have moved the entire 1 normal step (or full 16 microsteps)

So my motor starts microstepping, but it is not moving at first, it is building torque:

Step - Percent of full torque:
1st - 9.8 %
2nd - 19.6%
3rd - 29.4%
4th - 39.2%
....
10th - 98%
11th - 107.8%

And then it moves on the 11/16 step? But does it move all 11 steps ? If that is correct then I don't actually lose anything from microstepping. I would not have gotten myself a better performance had I used no microstepping.


But if it instead is that it only moves 1 of the 11 steps when it reaches 11, so that, once it can overcome the torque load, it moves just one microstep but overcomming that 1 microstep reduces the available torque by 9.8% and then we are below the required threshold and motor stops.


At the end of the day I am asking because I have a Z axis motor driving a trapezoidal screw and the motor is struggling at times. I am thinking if it would help to remove the jumpers to microstepping. But the thing is my "jumpers" are soldered connections so it is hard to test and also I would like to understand steppers better.

@LarsK
Instead of hijacking this thread I suggest you create a new thread, where you can ask questions about microstepping and torque.

Hello Andrew,

Read the original thread (it got side tracked a little down). This is very much a continuation of that. The answer to my question have a great deal of influence on how many microsteps it is attractive to use.

Simply, as I see it, if it is option nr 2 then you should not chose a microstep number which will give less torque then what you need to actually move the printer. At 128 microsteps you have 1.23% torque left. If your printer cannot move with that, is that microstep number still something you want?

Quote
mdcompositi
Hi guys,
I've installed today on my printer drivers tmc2100
I have a corexy and as electronic I use an arduino mega 2560 with ramps 1.4
I installed the drivers only for x and y and I must say that the improvements in terms of smoothness and quietness of mechanism from the previous configuration (with driver drv8825) is outstanding !!
I also put on the drivers heat sinks and of course I installed a fan!
I immediately printed an empty cube 20x20x10 with excellent results and with good speed!
But in subsequent prints something really unexpected happens to me !!
layers are shifted as if there were losses of steps or even slacks belts and pulleys!!!!
But this is not my case, the mechanics are adjusted very well, indeed, for the avoidance of doubt, I reinstalled the drivers earlier (drv8825) and I did the same prints have more this problem!
So I'm wondering, what's the problem !!
drivers are soldered and installed like picture below
[cdn.instructables.com]

Hello,

What tutorial did you follow to install them (I can see some unsoldered pins on one side) ? What are the settings of your three jumpers below one driver ?


Thanks

Edited 1 time(s). Last edit at 06/01/2015 11:29AM by -Nx-.

Did anybody ever thought of adding capacitors at the motor voltage inputs of the driver modules?

The capacitors that are already on the boards look very small to me, and I bet that they aren't low ESR types.
The more steps per revolution you make, the higher the pulse frequency gets, and the higher the losses in the capacitors will be.
Not only will the aluminium capacitors degrade faster because of that, they'll also not be able to deliver that much pulce current, resulting in reduced torque and step losses.
When switching from 16 microsteps to 128, you should use way bigger capacitors, with low ESR, and put big ceramic capacitors in parallel. You should even put additional capacitors on the stepper module itself, as the PCB traces and connectors have significant resistances and inductances.

Quote
uhrheber
Did anybody ever thought of adding capacitors at the motor voltage inputs of the driver modules?

The capacitors that are already on the boards look very small to me, and I bet that they aren't low ESR types.
The more steps per revolution you make, the higher the pulse frequency gets, and the higher the losses in the capacitors will be.
Not only will the aluminium capacitors degrade faster because of that, they'll also not be able to deliver that much pulce current, resulting in reduced torque and step losses.
When switching from 16 microsteps to 128, you should use way bigger capacitors, with low ESR, and put big ceramic capacitors in parallel. You should even put additional capacitors on the stepper module itself, as the PCB traces and connectors have significant resistances and inductances.

The microstepping makes no difference to the losses in the capacitors, because the drivers are chopping the motor current (normally at an ultrasonic frequency) all the time anyway. At high motor speeds, the effective microstepping rate will be much lower than programmed, because the theoretical microstepping rate will be higher than the chopping frequency.

---

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].

Thanks guys for the info. I'm about to try the TMC2130's for my BigBot printers and integrate positional error checking "Sounds" like I can expect better performance than the DRV8825's even, which I think perform very well, even from the arduino mega 8-bit controller.

To answer a previous question, the micro-stepping frequency doesn't have much to do with top speed of a motor. Voltage determines speed, and armature resistance and impedance thereby determine current current flow.
It's simpler to understand it in non-technical terms:
Stepper motors speeds are limited because the power supply voltage becomes not much more than the motor voltage (it's a generator when it's moving). When the voltage between motor and power supply is low, V=IR means your current from drops eventually to zero, and then your motor cannot spin any faster... more power cannot get into the motor because it holds a similar voltage to the power supply.

To help clear up some things also...more steps beyond 1/8 stepping or so does not mean more useable resolution, or a smoother drive. It's just that when the drive sends step pulses at a higher frequency the resulting pitch at that given rpm is above 20khz, and is now out of our audible range for speeds over 15mm/s or so, depending on your printer. All motors are also like speakers in that way, and tis why the pitch gets higher when the motor moves faster, and using non-tmc drivers that pitch is audible.

Actually the electronics freq for motor current control has nothing do with the micro stepping rate.

when the pulse to cause the next micro step at what every micro stepping resolution simply sets the current level dac to a
new value - - -

there is a comparitor that is used to determine if the desired current is above or below the desired current
there is an internal Oscillator that runs at a set freq - - - this is kind of like how a switching regulator works.
with voltage feedback - - - while here it is current level feedback - - -

next thing to remember is the motor is driven by magnetic fields - - - the changes in these fields (current levels)
cause the rotor to turn - - -

what does this mean - - - in effect there is a spring action - - - when out of desired position because of load or
other factors - - there is a spring effect trying to pull the rotor into position

with no micro stepping - - a single step means that it moves 1/4 of a full step cycle - - - at this out of position
there are forces that are repulsive and attractive - - - as you know North to South is what magnets want to be
being 90% of full step out means it has the attractive force to bring it in alignment and as it is getting near the next step - -
which would bring South to South or north to north there is a high repulsive force that is also wanting to bring the rotor into desired position

Now if you think about this - - - spring action - - - the rotor only moves when the magnetic forces cause the sprint action

Note it also takes time for the motor coil windings to change current because of the inductance of the coils.

the comment about micro stepping and less torque - - - has to be looked at is spring action and force applied - - -
and granted you can be a couple of micro steps behind on rotor position but torque builds up with out of position forces

sure you can over micro step a motor - - and the motor windings current will try to keep up - - - as long as it is not so much over stepped
that whole phase shift losses occur - - - there is not a problem - - - it will catch up.

I have a feeling that the TMC2100 does not sent out when doing the 256 micro steps at one time but spreads them out over the time period between
the driven micro steps - - - how is this possible - - - this would only happen at speed. at which point there is a clock rate - - - that smoothly increases and decreases
- - a phase locked loop running at 16 times the step rate would handle this - - - - if you look at all current processor chips - - - you add a crystal of any freq that you desire
and is low cost - - - the processor Phase locks the main clock to the desired operating freq. some extend this so far that you can get 100 Meg man processor clock
with a 32K watch crystal - - - that is right - - and have been doing this for years - - - -

Jim P

as stated above the micro stepping rate has nothing to do with the clock freq of the current control circuit.
the micro stepping simply changes the current set point - - -

the mechanical aspects have a big effect - - - - how large is each step current change - -
which causes a larger current level shift and the resulting magnetic fields to pull the rotor into position.
if you can smooth out those changes - - - less noise - - -

when you are applying 99% of rated motor torque - - you are asking for stalls - - - any little thing - - will cause a stall - -
that is why we run out steppers quite a bit away from this point

you are right that the more out of position the higher torque is present to cause the stepper rotor to move
it would be interesting to have someone actually measure the amount of movement with micro steps.

there are other factors in play - - a you comment friction, stickion (starting friction to be over come) and acceleration loads - - and required torque

my feeling is if you have so much friction that the rotor can not move you have a problem - - -

if the motor has no load - - free shaft - - how much will it move per micro step - - - the right amount or take a couple of micro steps before it moves.

I am willing to bet it will move the proper amount

the internal guts of the control chip switches at a fixed internal rate that has nothing to do with
micro stepping freq - - - it acts like a switching regulator by varying the fet on times - - -
in the stepper control chips - - there is a current sense resistor that provides the current feedback

thus the amount that the caps are worked at - - depends upon the phase angle - - - the decay mode
fast decay or slow decay - - - fast means more current change during the internal on and off cycle of the fets
meaning more ripply currents are present - - -

nothing to do with micro stepping rates - - -

I am playing around with some control electronics and looking at all ceramic caps - - - no AL caps - - why?
much better esr - - - as in close to '0' most designs have a small 1 uf cap close the control chip with 100 uf a short distance away
22uF ceramic caps at 16 volts are cheap now days - - - and are small 0805 case size - - - and feel that 4 of these will be more than
enough - - -

note the fet on and off times are in the 40 nSec range - - AL caps poor at those freq's.

So from reading this thread conclusion is microstepping does not increase resolution and reduce torque compare to lower microstepping.
So there is no point of higher microstepping driver beside reducing noise.
Maybe one slight advantage, it move smoothly causing less ripple and better curve movement.
Are these conclusions correct? If it is i will stick with a4988 which has higher compatibility with 8bit processor like my atmega 2560.
Wait, can someone please tell me why i should buy the higher microstepping driver? Is it much improvement?

Quote
clyevo
So from reading this thread conclusion is microstepping does not increase resolution...

Correct, once you go higher than about 16x. OTOH changing from 1.8deg to 0.9deg motors does increase resolution, which is helpful in delta printers and in ungeared extruders.

Quote
clyevo
and reduce torque compare to lower microstepping.

The torque per unit angular error - which is what matters - does not reduce. The incremental torque per microstep reduces, which is hardly surprising because each microstep is smaller.

Quote
clyevo
So there is no point of higher microstepping driver beside reducing noise.
Maybe one slight advantage, it move smoothly causing less ripple and better curve movement.

Correct. You won't necessarily notice the smoother movement, in fact switching to a good 32-bit firmware such as RepRapFirmware will probably give you a greater improvement in smoothness.

Quote
clyevo
Are these conclusions correct? If it is i will stick with a4988 which has higher compatibility with 8bit processor like my atmega 2560.
Wait, can someone please tell me why i should buy the higher microstepping driver? Is it much improvement?

High microstepping makes the printer much quieter when it is moving. If you don't care about the noise, you don't need high microstepping. If you do care about noise, do it properly and get a Duet WiFi or Duet Ethernet, which has TMC2660 drivers (which don't have the current limitations of the TMC2100) with up to 256x microstepping, and a fast 32-bit processor that can generate the step pulses fast enough.

---

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].