Best Stepper driver for Delta

I use dvr8825 on my delta and ran into problem " aka Tree ring or circle pattern

Here a picture of what I mean :


The bump you can feel them under your finger. Off course it also true for any line you print facing a tower and sometime the top infill too

I discover that its because of the carrier Z in this case, while doing a line in X axis the carrier move so slowly and at " low driver current zone" its create a circular pattern because the driver are not really good at that range and sometime there even step that don't register in that range. When I was doing some current testing while stepping the motor slowly its jump from 0 current to 0.6 or something, ignoring the step in between.


I discover 2 solution, one is to put the driver in fast decay, I try it and it reduce the problem but do not remove it, after fast decay the ring are still there but less apparent and you cant feel them with your finger.
The problem with fast decay is the driver now make lot of noise, even at 1/32

The other solution is to put diode in serie with one motor coil ,( one for each pair) to kinda remove that dead zone from 0 to .6 current, I did not try it yet because of the number of diode for 24V vs 12V and waste watt. Also I read this "hack" can cause more problem with stepper driver but its unclear and not well document.

So my question is simple, What are the best driver for delta if you put in consideration the low current zone and the slow and small move a delta have to do pretty often?

I was checking the Rasp128 (THB6128) since I have a radds its a combo you see often with it but Im wondering what are they performance on low current zone vs lets say tmc2100 ?

4988 are not in my choice of driver because of the 1/16 and I hear they doing the same circular pattern as dvr8825 anyway.

It will be great is some user of rasp128 or tmc2100 on delta can report they finding, or electronic engineer jump in from the datasheet side to give they opinion on those driver when you consider the problem of small and slow driver movement.

Im hoping this thread can help future delta owner to select the right stepper driver from the start and not waste money on those driver that dont do a great job with delta.

Thanks

I use a Duet with A4982 drivers, which on the Duet are configured in fast decay mode, and I don't get those rings. A4982 drivers (and A4988, which are simular) are fine as long as they are in fast decay mode.

1/32 microstepping is nice to reduce noise, but using 0.9deg motors (and keeping 1/16 microstepping) not only reduces noise but also improves positioning accuracy, because the torque per unit angular error is nearly doubled.

---

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].

There are many discussions about 8825 drivers on this forum and on google Deltabot forum.
The general agreement was these drivers shall not be used for printing (nor for CNC).
Unfortunately there are some boards based on it.
A nice paper about the problems is here :
[cabristor.blogspot.fr]

Quote
dc42
I use a Duet with A4982 drivers, which on the Duet are configured in fast decay mode, and I don't get those rings. A4982 drivers (and A4988, which are simular) are fine as long as they are in fast decay mode.

1/32 microstepping is nice to reduce noise, but using 0.9deg motors (and keeping 1/16 microstepping) not only reduces noise but also improves positioning accuracy, because the torque per unit angular error is nearly doubled.

I know about fast decay, but the problem is the noise you get from that on the dvr8825... Im just wondering there must be a better alternative. Changing my stepper to 0.9 one will cost way more than changing my driver and they not a lot of choice in 0.9 and are harder to find. The torque so far at 1/32 with 1.8A 1.8 degree ( set at 1.5A) stepper are more than fine for a delta, I dont know for a cartesian that have to move a bed but for delta its fine.

I will probably buy 5 rasp128 but before paying that premium price , I want to know if those will work great at slow speed, low current zone while in mixed mode or the TMC is maybe a better choice.


I know that PRZ, that where I read the diode trick, but I also read somewhere else some electronic engineer saying that not a good solution doing diode , plus if you going 24V that quite a number of them and watt...

Unless I forgot one , there only 3 driver with 1/32 + :

DVR8825: ( cheap, have to solder them in fast decay,have to mod the firmware for they different timing, are too loud in fast decay with my stepper (1.8A 17HS8401 , 3.2mH )

THB6128: (true 128 micro step, have a pot for decay so not solder , I have no idea what kind of specs they have on that low current zone)

TMC2100 For what I read they really 1/16 but use some trick to emulate higher micro stepping, again no idea they specs on low current zone)

Edited 1 time(s). Last edit at 04/18/2016 06:15PM by GroupB.

My point about 0.9deg motors was not about maximum available torque, it was that the torque PER UNIT ANGULAR ERROR is nearly double, which results in greater positioning accuracy. To put it another way, suppose you jog the head in the Z direction one microstep at a time. You will almost certainly find that it doesn't move on every microstep, but only every (say) 3rd or 4th microstep. If you now switch to 0.9deg motors with a similar specification and use the same microstepping, you will still find that it still only moves every 3 or 4 microsteps - but the microsteps are half the size now, so you have greater precision and resolution. This makes a small improvement to the print quality - see my blog for examples. My general advice is to use 0.9deg motors on any new printer build, but it is probably not worth upgrading the motors on an existing machine unless you have another reason to change the motors as well.

DRV8825 drivers are well known for being noisy in fast decay mode because of their constant of-time. A4982 drivers do not have constant off-time, so they don't suffer from the same problem.

I can't comment on the RAPS128 drivers, because I haven't used them yet. One advantage they have over almost all other plug-in driver modules is that the chips are designed to be cooled from the top, so heatsinks actually do work on them. But 128x microstepping is very demanding on the firmware unless you reduce the maximum travel speed, so you may find yourself using them with lower microstepping.

---

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].

Have you tried setting jumpers on your 8825's to be in 1/16 mode?

Im not so good with electronic datasheet , I know its your thing DC , if you look at they datasheet THB6128 you see anything special ? are they normal timing so I can return to normal firmware ? I hear they are everything a 8825 should have been with all the small thing fix like a better decay but all those graph in the datasheet is chinese to me. They are kinda expensive and im really thinking about switching to them but I want to be sure its the right choice.

I dont plan to run them in 1/128 but stick to 1/32 , or more in the future when the firmware get port to better hardware. My goal is just to get rid of those circular pattern and go back to a normal timing firmware ( I dont want to ask dan every new firmware for a special edition for me).

Actually the diodes are quite good solution. It is an easy way how to make the default slow decay of the chip somewhat faster so that the current does not build up because of the current sensor blanking time.
Of course, a better solution would be a driver which allows us to configure the portion of the PWM period which should be dedicated to fast decay (the rest would use slow decay). Separately for raising/falling portion of the stepper current.

Hmm, but DRV8825 data sheet claims:
Mixed decay mode begins as fast decay, but at a fixed period of time (75% of the PWM cycle) switches to slow decay mode for the remainder of the fixed PWM period.
Should not it be rather like this:
Mixed decay mode begins as slow decay, but at a fixed period of time (75% of the PWM cycle) switches to fast decay mode for the remainder of the fixed PWM period.

The DRV8825 was introduced in 2010 and the THB6128 in 2011. Are there any newer drivers available in the 2-3 amp range?

Quote
GroupB
Im not so good with electronic datasheet , I know its your thing DC , if you look at they datasheet THB6128 you see anything special ? are they normal timing so I can return to normal firmware ? I hear they are everything a 8825 should have been with all the small thing fix like a better decay but all those graph in the datasheet is chinese to me. They are kinda expensive and im really thinking about switching to them but I want to be sure its the right choice.

I have never been able to find a proper datasheet for the THB6128, only a "Development proposal" document. Unfortunately, that document doesn't give the timing requirements of the step pulses.

What electronics do you have? Bear in mind that if you are using 8-bit electronics, then even 32x microstepping makes it a struggle for the processor to achieve reasonable travel speeds. So with 8-bit electronics, I suggest 1/16 stepping on A4982 or A4988 drivers.

---

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].

In my case almost the same ringing behavior with A4982 drivers from Duet 0.6 board (replikeo) with Wantai 42BYGHW811 steppers (2.5A, 1.8mH) running at 1000mA config setting from 12V power supply.

Quote
roboprint
In my case almost the same ringing behavior with A4982 drivers from Duet 0.6 board (replikeo) with Wantai 42BYGHW811 steppers (2.5A, 1.8mH) running at 1000mA config setting from 12V power supply.

I guess it could mean that the A4982 drivers don't like those low-inductance motors. The motors and drivers are not a good match for each other - you should choose motors with a current rating below 2A for A4982s. You may need to increase the current. Running at less than 50% of the current is unusual, and means that the detent torque is comparable to the incremental torque per microstep.

---

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].

Quote
dc42

Quote
GroupB
Im not so good with electronic datasheet , I know its your thing DC , if you look at they datasheet THB6128 you see anything special ? are they normal timing so I can return to normal firmware ? I hear they are everything a 8825 should have been with all the small thing fix like a better decay but all those graph in the datasheet is chinese to me. They are kinda expensive and im really thinking about switching to them but I want to be sure its the right choice.

I have never been able to find a proper datasheet for the THB6128, only a "Development proposal" document. Unfortunately, that document doesn't give the timing requirements of the step pulses.

What electronics do you have? Bear in mind that if you are using 8-bit electronics, then even 32x microstepping makes it a struggle for the processor to achieve reasonable travel speeds. So with 8-bit electronics, I suggest 1/16 stepping on A4982 or A4988 drivers.

Like I said early in my first post Im using a radds so yah 32 bit, this is the same config you know I use since you answer my prev threads about firmware problem I had or my built thread. I dont want to be rude but you do that pretty often... maybe you read from a phone with a small screen and you miss things or read too fast... anyway no hard feeling, extra explication is always welcome to help new user that dont know about 8bit/32bit and micro steeping.

roboprint : I read somewhere its a question of driver/motor combo , like some motor (high mH) work flawless with 8825, some like mine (3.2 mH) not so much, I dont know for the A4982 but I know the 8825 like HIGH mH motor ( How high... I dont know more than 3.2mH I guess), too low and you run into ring pattern problem because of that low current zone, this must be the same for A49XX driver maybe not as bad but 1.8mH is pretty low.

This will be great if we can get all this info in one thread, like what to look for when selecting stepper for this or that driver.
Off course this is not really a problem for cartesian since they dont do slow movement like we do, they dont really require to match motor/driver as much as we do. Maybe that why there is not a lot of info about this already.

Yes I was using a smartphone when I wrote that reply. I did scroll back to see if you specified what electronics you are using, but I was scanning for words with uppercase letters in them so I didn't spot 'radds'.

---

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].

While you could find very valuable info in forum threads, their organisation is uncontrolled. The only way to have organised info is to report to the Wiki what is told in threads. But not many people are doing that, so information remains mostly scattered.

On this particular topic, I have created the page 'Choosing stepper for a delta' [reprap.org] which doesn't go in details, but try to define an approach.
It might be interesting to go a bit more in details about the internal operation of drivers, but that need a lot of time and particularly, that needs to create schematics. Also, I am of opinion that wiki page shall remains short and if going in details, that shall be done on a dedicated page, written by people having good knowledge of electronics, which is not my case.

I think you may find more technical info on steppers/drivers in the google deltabot forum. You may search for post of Ryan Carlyle.

Stepper motor and stepper driver matching:
If I did not make a mistake somewhere then the minimum current that a stepper driver (like DRV8825) can maintain is:
Imin = U/R * (1-exp(Ton/Tau)) / (1-exp(T/Tau))
Tau = L/R

You need to add diodes if:
Imin > I * sin(π/2/m)

If diodes needs to be added ten they need to have accumulated forward voltage at least:
Uf = U * (1-exp(Ton/Tau)) / (1-exp(T/Tau))

Typically Ton < T << Tau and we can write Ton = duty * T
Therefore we can approximate the above equations as:
Imin = U/R * duty
Uf = U * duty

Symbols:

• U - power source voltage
• R - stepper resistance in ohm
• L - stepper inductance in henry
• I - desired stepper (maximum) current
• m - microstepping used (e.g. 32)
• T - period of the driver PWM in seconds
• Ton - minimum on time (in seconds) of the stepper driver due to blanking (3.75e-6 for DVR8825)
• duty - minimum duty cycle of the stepper driver
• Uf - required forward volate of the diodes (if they are needed)

Edited 2 time(s). Last edit at 04/20/2016 06:22AM by hercek.

Hercek, very interesting, I just looked at the A4988/4982 datasheet and the blanking time (period where the sensing is neutralised to avoid false measurement) is 1 microsecond. Minimum on time may be a bit higher, but is not listed. That is a lower value compared to the values you list of 3.75 microsecond, meaning in similar conditions, A4988/4982 can operate a current much smaller than the DRV8825.

Yes, rather high Ton time of DVR8825 is probably the main reason people have problems with that driver. Anyway, it will show up only when the stepper is run very slowly, and during the period when the current is raising on a phase and the decay mode is slow or mixed. One can easily fix the problem by setting the decay to fast or by adding the diodes. I do not think diodes are that bad solution. They are cheap ... and they consume power but it is only in the range of few watts per phase.

Of course the fast decay has a bit higher micro-stepping error and the noise. Most people will not notice the error but they will notice the noise.

Also if DVR8825's mixed mode really does first fast and then slow decay inside one PWM period when current is dropping then the advantage of mixed decay compared to slow decay is not really there during the current dropping phase. You still get higher micro-stepping error there. And higher noise too. Or I do not understand something well. But I'm leaning in the direction that that sentence is wrong in the DVR8825 data sheet.

8825 has missing steps issue which will affect the print quality. check out my video explaining the issue [www.youtube.com]

I haven't tried Raps128 but I am very pleased with the TMC's (2600's I think) on my DuetWifi board. Smooth and silent and the print quality is very good. I have ordered some TMC2100 stepsticks for my corexy too. Yes they are 1/16th "real" micro-stepping, but they interpolate to 1/256th for noise reduction. I think we all get too hung up on what microstepping really does, its a noise reduction measure. I suspect if we set our machines to full steps we'd notice only marginal print quality reduction but they would sound like proper industrial machinery.

Edited 1 time(s). Last edit at 01/02/2017 06:38AM by DjDemonD.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

Quote
DjDemonD
I haven't tried Raps128 but I am very pleased with the TMC's (2600's I think) on my DuetWifi board. Smooth and silent and the print quality is very good.

The drivers on the Duet WiFi are TMC2660. The microstepping is configurable in firmware to 1, 2, 4, 8, 16, 32, 64, 128 or 256. At 16 and above you can also enable interpolation to 256.

---

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].

I'm using RAPS128 drivers on my delta with 0.9° steppers and on the XY axis of my Prusa with 1.8° steppers in 64/1 microstepping. The print quality didn't improve on my Prusa, but the steppers run way cooler now with the same current setting.
They are a bit pricy, so I'd recommend TMC2100 if you can live with the lower max current of 1.7A.