Duet 3 expansion board stepper driver thermal tests
-
I just tested the pre-production Duet 3 expansion board, using 3 x Nema 34 motors all running continuously @ 6.3A peak current/phase (4.45A RMS). This is the test setup:
This is the FLIR image after 10 minutes, with no fan cooling:
The reported MCU temperature was 62.4C. Adding a quiet 40x20mm Sunon fan brings the temperatures down nicely:
The reported MCU temperature came down to 43C. Ambient temperature during these tests was 27C.
-
Oh my, what big connectors you have!:)
Different drivers, or just different package on this versus the main board prototypes you showed off?
-
@bearer said in Duet 3 expansion board stepper driver thermal tests:
Oh my, what big connectors you have!:)
Yes, the Molex KK connector can't take the current.
Different drivers, or just different package on this versus the main board prototypes you showed off?
Different drivers. We switched from TMC5161 to TMC5160 + external mosfets. This has allowed us to increase the maximum motor current. Also, if a TMC5161 failed then it would have been almost impossible to replace; whereas if an external mosfet fails, with luck the driver will still be OK and just the mosfet can be replaced.
-
Great data!
What does the 51C package/board temp translate to on the die? Is that within the operating specs for the chips?
-
@mrehorstdmd the 51C is on one of the FETs rather than the TMC5160s which dissipate much less power within the package than a typical driver with built in FETs. Datasheet is here:
https://www.trinamic.com/fileadmin/assets/Products/ICs_Documents/TMC5160_Datasheet_Rev1.10.pdf
thermal characteristics indicate in the region of 21K/W of thermal resistance between the Junction and ambient or 8K/W between junction and the PCB ~ 1mm from the package leads. Using a FLIR camera is not exact but either way , even with the FETS heating the PCB and that heating the TMC5160s the internal temp will be well below the 125C max operating range.For the FETs (https://www.diodes.com/assets/Datasheets/DMT6018LDR.pdf) they will be dissipating max of ~0.34W per FET in a dual package = 0.7W which is also within spec.
-
@t3p3tony Great! It's going to be a very useful board!
-
I admit that the Duet3 is getting pretty exciting to me. I've been modding my Prusa MK3s and starting to hit the limits of what I can do with the EINSY. This guy, though...
-
@evan38109. I've been modding my Prusa MK3s and starting to hit the limits of what I can do with the EINSY. This guy, though...
How very dare you... don't you know that prusa printers are perfect and everything he produces is better than anything out and his own original idea.. well at least in his own mind anyway...
-
@t3p3tony said in Duet 3 expansion board stepper driver thermal tests:
dissipating max of ~0.34W per FET in a dual package = 0.7W
@t3p3tony, I am curious, when you computed 0.34W, did you need to factor in the chopper transition periods where the FETs have high resistance and significant current?
Also, I am counting 3 TMC's and 12 dual package FETs which results in 8 FETs per TMC instead of the expected 4. How come?
-
@calvinx said in Duet 3 expansion board stepper driver thermal tests:
How very dare you... don't you know that prusa printers are perfect and everything he produces is better than anything out and his own original idea.. well at least in his own mind anyway...
Heh, the Prusa is a great little machine, and if I just wanted to print, I'd leave it as is. Great for modding, too -- I got it printing 0.02mm layers and looking good.
I'd like WiFi, though, and more current to drive some different steppers without hitting thermal issues. (Though 6.3A is overkill, I'd be happy with a solid 2A.) Unfortunately, Duet 2 WiFi uses TMC2660 drivers, which don't support Trinamic's StealthChop, and I'm not willing to give that up. StealthChop is significantly quieter, prints smoother, and is easier to tune. The Maestro uses TMC2224's, which don't support StallGuard, so no sensorless homing. Well, no WiFi on the Maestro, either. If you can't tell, I've been keeping an eye out for a replacement that checks all the boxes.
Anyhow, apologies for drifting off topic. Looking good, dc42!
-
I think you are mistaken in thinking that stealthChop prints smoother. Also, printing in stealthChop mode makes it much more likely that you will get a skipped step if the nozzle hits a blob or obstruction. StealthChop doesn't work at all at high speeds, which is why the drivers switch over to spreadCycle at higher speeds. But stealthChop is effective at reducing noise at standstill and low speeds.
The TMC2660 drivers support coolStep which is the forerunner of stealthChop. It is configurable via M915.
-
@zapta said in Duet 3 expansion board stepper driver thermal tests:
Also, I am counting 3 TMC's and 12 dual package FETs which results in 8 FETs per TMC instead of the expected 4. How come?
Each H-bridge needs 4 mosfets. You need one bridge per motor phase.
Duet WiFi hardware designer and firmware engineer
Please do not ask me for Duet support via PM or email, use the forum
http://www.escher3d.com, https://miscsolutions.wordpress.com -
@dc42 said in Duet 3 expansion board stepper driver thermal tests:
You need one bridge per motor phase.
Bingo.
@dc42, one more question if you don't mind, do the mosfets get hotter with 12V than 24V since the current consumption from 12V source is higher for same stepper power?
-
@zapta not specifically because the stepper drivers are chopping to limit the current through the Motor could (and hence the FETS and sense resistors)
-
@t3p3tony, yes, you are right. I asked around the the mosfet's power depends on the coil current, not the supply current (which is roughly double for 12V compared to 24V).
I also run an experiment with a Pololu stepper driver and the temperature is about the same for 12V and 24V.
-
@evan38109 said in Duet 3 expansion board stepper driver thermal tests:
I admit that the Duet3 is getting pretty exciting to me. I've been modding my Prusa MK3s and starting to hit the limits of what I can do with the EINSY. This guy, though...
Do you have any particular reason to put Duet 3 on that though?
I feel like Duet Maestro is a better fit for the most part.
The regular Duet2 is more capable than most desktop users need. The biggest thing I want is direct support for 4 wire fans. Even though my bigger machine can use a little more motor current, it works pretty well at 2.4A.
