Re: [OpenPnP] MARLIN VS SMOOTHIEware FOR OPENPNP ??

[Arthur Wolf]

Marlin/Smoothieware only have a speed difference in pretty rare/extreme cases.

How fast do you expect/need your machine to move at, both in terms of millimeters per second, and in terms of microsteps per second?

On Mon, Mar 22, 2021 at 3:16 PM Siddharth Valambhia <svala...@gmail.com> wrote:

I am building pnp & I am having mechanical background so I am bit confused which hardware and which firmware I should use. 

Openpnp github suggest smoothieware while in this forum people says marlin has more speed (& also much better tutorials & possibly future updates). I am concern about speed and firmware should support adding camera,feeder etc without much coding.

I have btt skr v1.4 laying & due + shield with me which is best hardware  ? and firmware ?

--
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/6d510612-41e2-41d6-a1f4-fb5c4f3424a0n%40googlegroups.com.

--

勇気とユーモア

[Marek T.]

Hi Arthur,

In my case I would expect at least 200kHz on the Step output.

However my machine is not some extremely fast (IMO) and 400kHz would be more appreciated.

Second advantage of Marlin over Smoothie is Dual motors' control (no need to wire the motors).

Third advantage of SKR over Smoothieboard are dirvers sitting in the sockets and Trinamic drivers handling over individual serial ports (routed on the pcb).

In all the rest the Smoothie is better :-). I personally especially appreciate the switch modules not existing in Marlin.

[Arthur Wolf]

On Mon, Mar 22, 2021 at 5:46 PM Marek T. <marek.tw...@gmail.com> wrote:

Hi Arthur,

In my case I would expect at least 200kHz on the Step output.

However my machine is not some extremely fast (IMO) and 400kHz would be more appreciated.

Mhm, I know *you* have those crazy servo drivers that you can't run below that for reasons I don't remember, but I was curious about Siddharth, as in my experience, needing more than 100khz is *extremely* rare, and people who think they are going to need Marlin for this specific reason, 99% of the time are just misinformed/mistaken (and so I'm trying to figure out if that's what's happening here).

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/02b61cc1-4af2-4488-91ce-fb9394c7093an%40googlegroups.com.

--

勇気とユーモア

[Marek T.]

I thought you ask to start some general discussion about it...

You are probably right about these 99%!

My unique problem comes from the external drivers which have dividers on the step inputs, and need 2x or 4x (programmable) more step pulses than normally. Without it 50kHz would be absolutely enough!

[Jarosław Karwik]

I do have my CNC controller on LPC4330 ( equivalent of Smoothie V2) configured to 150kHz ( hardware max is around 195kHz with 4 real concurrent axis).  

These days it drives mostly Delta servos - it is enough for them as the machines do not go over 2 meters/second.

But with steppers/pseudo steppers it was never above 100kHz.

BTW - Stm32H7 which you plan for LPC4330 replacement should easilly get 400kHz :-) - even if not needed. I will be able to test my new controller soon to check how much 480MHz Cortex M7 with timers connected to step pins can really do.

[ma...@makr.zone]

Marek, Arthur

I doubt 100kHz is enough for fast machines. The blend between accuracy and speed in Pick & Place is quite unique. Neither 3D printing, milling nor laser cutting have the same blend (maybe the latter for positioning moves).

It is already nearing the limit for my simple Liteplacer, if I push it. And I have my microstepping tuned to be quite optimal at 80 microsteps/mm.

I get up to 42m/min = 700mm/s, with at 80 steps/mm translates into 56kHz already. Remember, this is a simple MakerSlide, delrin sandwiched wheels (ouch!), belt & stepper machine, and I believe all that is missing to go significantly faster is more voltage to the drivers.

https://makr.zone/pick-place-machine-first-simulated-small-test-run/66/

Surely a servo machine can do many times that top speed!

_Mark

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/0276a9a0-d64e-4c2b-bf54-a723d9f29f22n%40googlegroups.com.

[Marek T.]

Agree with you Mark. In my mavhine 50kHz would be enough in the drivers setup of your meaning. 

But my external drivers has two encoder inputs (A/B) and they require 2 times or 4 times more Step pulses than "normally".

Two - if driver is wanted to react only on rising edges, four - if we want react on every rising and falling edges. The same it's decision if we need 0.01 or 0.02 resolution. So to get full speed at 0.01 I need 4x50kHz, 200kHz so. If have only 100kHz (2x50kHz) I can have full speed with 0.02 resolution or half speed with 0.01.

I've chosen option of full speed and limited resolution - and frankly speaking 0.02 is rather ok. But the fact is that I can't use full parameters of my drivers/machine having 100kHz.

But, it's not the thread about my machine... :-)

[Arthur Wolf]

On Mon, Mar 22, 2021 at 7:31 PM ma...@makr.zone <ma...@makr.zone> wrote:

Marek, Arthur

I doubt 100kHz is enough for fast machines. The blend between accuracy and speed in Pick & Place is quite unique. Neither 3D printing, milling nor laser cutting have the same blend (maybe the latter for positioning moves).

It is already nearing the limit for my simple Liteplacer, if I push it. And I have my microstepping tuned to be quite optimal at 80 microsteps/mm.

I get up to 42m/min = 700mm/s, with at 80 steps/mm translates into 56kHz already. Remember, this is a simple MakerSlide, delrin sandwiched wheels (ouch!), belt & stepper machine, and I believe all that is missing to go significantly faster is more voltage to the drivers.

The thing is: increased microstepping isn't increased accuracy. People typically get to microstepping in the 32-256 range, but 4-8 is about where you stop getting accuracy benefits.

100 khz at 8 microstepping is much more speed than the laws of physics will allow you to play with anyway, on most machines.

And as you don't get accuracy benefits above 8 microstepping, interpolation becomes acceptable above that.

And that's why it's extremely rare for us to see a user who actually needs more than 100khz. It happens. It's super rare.

To view this discussion on the web visit https://groups.google.com/d/msgid/openpnp/a2c2a81a-4504-c315-6520-cc1cc7c09297%40makr.zone.

--

勇気とユーモア

[Jarosław Karwik]

I have checked my machine.

I do have there 20mm lead screw ( 20mm travel for one rotation), coupled with 1:1 belt from servo with 1000 ppr encoder ( it gives 4000 edges for one rotation).

This gives  5um resolution - so for 1m/s I would need 200kHz.

The most straightforward way would be to just use it like that - without setting multipliers in servo - although I am likely to tune it down as such resolution is quite theoretical.

Anyway - what I wanted to say is that once you go beyond steppers, you will get easily potential for higher control frequencies . Especially that it is common to put 2:1 or 3:1 on belt to get extra machine accuracy.

PS. My machine ( still in conversion) is 20 years old Philips CSM ( rebranded Yamaha). Mechanically perfect :-) , just the electronics was junk ( well, it even actually worked, just  PC AT is ancient myth these days). 

[Siddharth Valambhia]

https://embeddedtronicsblog.wordpress.com/2019/08/03/smoothieware-step-pulse-test/ 

jim did lot of test with different hardware...  smoothie can generate about 3000 rpm.