Any benefit in replacing my TinyG controller with a Duet or Smoothie?

[Jim Young]

Since I am in the process of completely redesigning my LitePlacer setup, I was wondering if there would be any benefit to replacing the TinyG with something more capable, like a Smoothie or Duet board. 

Would there be any benefit to the speed of motions or accuracy?

[bert shivaan]

Hi Jim, I can't say if you will see any benifit swapping on the machine side, I don't know much about TinyG.

Configing with Duet using their web browser is MUCH easier IMHO. So there is that. I think maybe it also does a better job with other things as Mark can point out.

On Sun, Jan 8, 2023 at 10:53 AM Jim Young <infinite...@gmail.com> wrote:

Since I am in the process of completely redesigning my LitePlacer setup, I was wondering if there would be any benefit to replacing the TinyG with something more capable, like a Smoothie or Duet board. 

Would there be any benefit to the speed of motions or accuracy?

--
You received this message because you are subscribed to the Google Groups "OpenPnP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openpnp+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/6594e4f4-24ac-4b8e-b580-c4e958230979n%40googlegroups.com.

[mark maker]

Hi Jim,

That's a tough call. I hope you read the following carefully, I spent a long time writing it 😁

The TinyG has true jerk/S-curve motion control (a.k.a. 3rd Order Motion Control) which is really a very good thing to have on a Liteplacer, because this machine design is really "challenged" in terms of stiffness and balancing and vibrates/swings like hell if driven with aggressive constant acceleration.

Constant acceleration motion control is the method of practically all the other Open Source controllers I know, like Duet and Smoothie. Marlin has simplified jerk control, but I'm not sure this one is any good, judging from the source code it plans with constant acceleration, then carves the "S" shape out of the ramps afterwards, which I think means that short moves need to be very slow in order to not stall the motors on long moves (I never actually tested Marlin). Unfortunately, we have many short moves in PnP vision. I'm not actually 100% sure the same does not apply to TinyG as well, I never spent much time analysing TinyG's source there.

Now, I have added so-called Simulated3rdOrderControl to OpenPnP, so you can get a practical "jerk control" effect on constant acceleration controller, like Duet or Smoothie:

https://www.youtube.com/watch?v=cH0SF2D6FhM

But this interpolation will never be as smooth as TinyG's. I don't actually know how it would compare in speed to an equally well-tuned TinyG in real life pick and place. TinyG can be fast if well-tuned as seen here:

https://youtu.be/ol07YgZOKgs?t=45

The effectiveness of controlling jerk on TinyG:

https://www.youtube.com/watch?v=6SBDApObbz0

Also watch the following video, which is actually about camera settling, but thereby shows why this jerk control is so important for OpenPnP, especially on a Liteplacer (my own):

https://youtu.be/Pxg6g3KI5_E

Always keep in mind, that while you can diagnose the camera images shaking like this, the same thing happens to the nozzle tip too, so picking and placing will be similarly bad / imprecise if vibration is not addressed.

Other important things to know:

The stock steppers of the Liteplacer are 0.9° per full step, which is more precise, but limits higher speeds. With the TinyG's built-in stepper driver and a maximum of only 8 microsteps, that's probably the right choice.

But stepper drivers with 16 microsteps (more I don't recommend) are OK with the more common 1.8° steppers, which generally go faster.

Pick and Place wants speed rather than force, so high voltages are much more important than current. Double the voltage roughly means double the max. speed! To understand why, search for "Higher Voltage, please!" here:

https://makr.zone/choosing-a-motion-controller-the-panucatt-azteeg-x5-gt-32bit/455/

TinyG is limited to 24V. Duet3 6HC is now rated for 48V. However, you can also drive external stepper drivers with TinyG.

So to sum it up:

• If you want to completely revise the machine, including upping the voltage, go with the Duet3D. This is also a more versatile controller which can drive many other sensors and high current actuators etc. (TinyG does not). And it has a state of the art MCU with constantly evolving firmware, which might one day include 3rd order control or other means of vibration control (it already has a solution to suppress ringing, but currently I doubt it will be enough for a Liteplacer, as it only supports one global ringing frequency, and probably only high frequencies, where the heavy Liteplacer head swings slowly).
  
• For a more moderate revision, you can keep the TinyG, but spend more time tuning it.
  
• As an interesting middle-ground, you could keep the TinyG, but wire X / Y with external step signal driven closed-loop motors like these:
  https://www.sorotec.de/shop/CNC-Steuerungstechnik/Closed-loop-Systeme/

_Mark

On 1/8/23 16:53, Jim Young wrote:

Since I am in the process of completely redesigning my LitePlacer setup, I was wondering if there would be any benefit to replacing the TinyG with something more capable, like a Smoothie or Duet board. 

Would there be any benefit to the speed of motions or accuracy?

[Jim Young]

Mark, thank you so much for such a comprehensive response! This give me a lot of useful information.

[Jim Young]

The video of Дмитрий Голик's machine is pretty impressive. I wish I knew what feed, acceleration and jerk values he is using. I can't say that I've really tuned the motion performance at all. When I moved to OpenPnP I translated the motion settings over from the LitePlacer software. My motion is definitely slower than what is shown in the video. I feel like I'm flying blind when it comes these settings. 

[Shai]

Check out our smoothie board spinoff called Rapid Board, specifically designed for OpenPNP pick and place machines :) https://www.deltaprintr.com/product/rapid-star-board/

[Jim Young]

• As an interesting middle-ground, you could keep the TinyG, but wire X / Y with external step signal driven closed-loop motors like these:
  https://www.sorotec.de/shop/CNC-Steuerungstechnik/Closed-loop-Systeme/

Would one need new firmware to use close-loop motors?

[mark maker]

> Would one need new firmware to use close-loop motors?

No, I see no reason why. I guess you just have to follow this:

https://github.com/synthetos/TinyG/wiki/TinyG-Using-External-Drivers

_Mark

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/d924504f-8c44-43b1-85c8-ff3c1b60f509n%40googlegroups.com.

[dc42]

Hi Jim,

It's worth remembering that although S-curve acceleration is effective at reducing some resonances - especially high-frequency ones - it's much less effective low-frequency resonances. The Klipper firmware guys found it necessary to implement input shaping even though Klipper already supported S-curve acceleration. Input shaping (which is implemented by Duet/RRF since version 3.4) can be targeted at the specific bands of low-frequency resonances typically found in belt-driven motion systems.