Duet maximum achievable step rates


  • administrators

    I've started a spreadysheet to record approximate acheivable step rates at https://docs.google.com/spreadsheets/d/1AWA1wLbOaYzxzdQa5LRZvn9rgEk2BuluHy6-_OnD6FY/edit?usp=sharing.

    EDIT: In case you want to test yourself.

    • Work out a long move that you can use on your machine to test the step rate. It needs to be long enough that most of the move will be at the requested speed, not doing acceleration or deceleration
    • Select x256 microstepping
    • Send M122 to clear the hiccup count
    • Execute that move
    • Send M122 to read the hiccup count
    • Repeat the last 3 steps at different speeds, to find the speed at which the hiccup count increases dramatically. For a 300mm move at x256 microstepping, I reckon that 50 to 100 hiccups is acceptable, more is not.
    • Use the steps/mm and the speed you measured to work out the step pulse rate.


  • Thank you for this!

    Its a nice reference point.



  • Might be a bit presumptive to ask since I do not know the heart of DSF very well, but would these rates conceivably be different with an SBC at a future date?

    I.e. is computational offload to SBC present/planned or will it remain as an alternative method of getting gcode to the board? Asking on behalf of my ~620step/mm extruder (duet 2).

    Also, good information though maybe on the high side? my duet2 cannot run a btech bmg at 417.29 steps/mm at 128 microstepping with 30mm/s retracts without accruing ~600 hiccups over a 2 hour print.

    Either way, this is some super good baseline information. Thanks.


  • Moderator

    @deadwood83 said in Duet maximum achievable step rates:

    128 microstepping

    Such high microstepping is rarely beneficial. x16 with interpolation to x256 is definitely the way to go, especially for a Duet 2.



  • @Phaedrux said in Duet maximum achievable step rates:

    @deadwood83 said in Duet maximum achievable step rates:

    128 microstepping

    Such high microstepping is rarely beneficial. x16 with interpolation to x256 is definitely the way to go, especially for a Duet 2.

    There is a usage case where higher micro stepping is desirable, as I have repeatedly pointed out. That is when using a mixing hot end whereby some of the filaments may be moving at single digit percentages of the total and therefore 16x does not give adequate resolution. There are other users who print highly detailed parts, using very small nozzles and layer heights. These users may need higher micro stepping and/or use highly geared extruders. Either way, a high step pulse frequency is required so it has been hugely disappointing to me that Duet3 has had a much lower step pulse frequency than Duet 2. I sincerely hope that the latest firmware will improve that situation but I'll wait until it moves out of the beta stage before I try it.


  • Moderator

    @deckingman yes I had your use case in mind when I said rarely.



  • @Phaedrux Understood. BTW, with firmware 3.1 and extruders connected to expansion boards, I get extremely high hiccup count using 32X micro stepping. That's with 1.8 degree motors (roughly 400 steps per mm) and in non mixing mode. If I used 0.9 degree motors (800 steps per mm) which is not uncommon, then I would be seeing that high hiccup count using 16X micro stepping. That's how bad the step pulse frequency is with firmware 3.1.



  • @deckingman
    What board do you use?
    Can you post more details e.g. the motor config section in config.g for others to better understand?



  • @LB I'm away from home at the moment but the low step pulse frequency is related to expansion boards and firmware version 3.1.1. The extruders are Bondtech BMGs with steps per mm of about 400 with 1.8 degree stepper motors. I can run these at only 16x microstepping and 30mm/sec retraction speed. Hopefully this situation will be improved with the 3.2 firmware but I'll wait for the beta testing phase to be completed before I attempt to install it.


Log in to reply