Advanced Search

Those drives have more-or-less infinitely adjustable microstepping.

HIGHLY recommend adjusting the following alarm setpoint.  I have similar drives (Nema34's) from Stepperonline, and the following alarm was set to 4000 counts from the factory.  They've got 1000-line (4000ppr) encoders, which means from the factory they won't alarm for a full turn.

Once set to a reasonable level, you can see if they are actually losing a few steps.  I'm also wondering if your g-code is output with lots of line segments instead of arcs and you need to adjust your G64 P-value (or turn it on).

BTW, the "M61..." is from one of the GUI screen buttons (UNLOAD SPINDLE BUTTON)

Looking back at the old thread I linked, it's pretty clear I - and the PB Dev who was responding - thought #5210 was for setting/clearing G52/92 offsets.

I don't know why the author would want to clear the offsets during a spindle unload.  The original subroutines had a bunch of numbered parameters added by the authors - this may have been a mistake and they overlooked the fact that #5210 was already in use.

And yes, I, too, am convinced this is a mistake.  I'll raise an issue on PB-GH tonight.

Thanks for kicking this around with me.

Building from source should still work, and it is quite easy, and you can build the debs so if you need to install on another pc or reinstal, just use the debs.

Forgot, these closed loop steppers, i have no idea how the position geedback is setup, so if you can:
-set the drive to x1 or x2
-lock the shaft so it can not turn
-in linuxcnc jog in the smallest possible increments
How many steps before a fault?
-
Geckodrive servos had a 128 step fault tolerance, and that is ok for high ppr encoders, but not ok for say 100ppr encoder as that is more than 90 degree of deviation before fault.

Impressive Andre! I have a the bigger brother of your MV-Junior; a MV 35/40 school machine and a SL25-B that i am planning to retrofit. Do you have even more information/pictures of your build?
I see you are Norwegian as well, please hit me up on FB or IG if interested in sharing experiences
 

As a general rule with microstepping, use only powers of 2, and as low as possible with some acceptable vibrations.
So x4 or x8 (800 or 1600 steps), rarely 3200 (x16). X32 should be the absolute max.
Microstepping is never precision, it is usefull only to lower vibrations, as steppers can be easily positioned between 2 poles but not easily betwen 16 divisions of those 2 poles, and any microstepping lowers the torque.

@DPFlex ... from 2.8 I'm unable to use halcompile .... AndyPugh suggest to use a rip install ..... you download on Lcnc_292_wharehouse ..... search on these new folder scarakins.c and change it with modify version .... than run process for make .... when obtain compilation file on rtai folder inside Lcnc_292_wharehouse folder of rip install, use sudo command for root permission, search for original one scarakins.c of your installation (if not use a rip install) and cancel these, copy with root permission your scarakins.c compiled from Lcnc_292_wharehouse  folder to system folder. Reboot ... all done.

Use these tryk for all modifications. Or learn you about customkins.c (never looses these time really) .....

If you have a rip istall .... use the same trick, but change modify scarakins.c from your "compiled Lcnc_292_wharehouse  folder" to your preferred rip install .... reboot and voilà.

regards

To be absolutely honest I don't really understand why they are setting #5210 = 0 in the first place.

What does unloading the spindle have to do with g92/g52?

Also here:
M61 Q#3991 G43 H#3991 #5210 = 0

Why the need to set #5210 = 0?

[edit]
The more I look at it the more this looks like a mistake to me. It just simply does not seem to make any sense at all.

and in witch SO?

tommylight wrote:

No easy way to install LinuxCNC on Linux Mint 22, yet.

no easy way like first installation on mint19.2? Because was quite easy in my opinion .... easy not equal "write 2 row of command" .... in my case use 2 row of command give me more and more problem than use a long list of command where I can see what happens ....

Cant do this anymore bye all wrote:

If you are having trouble with the library you require why are you considering Lmde6 ?
 

.... I'm not a masochist .... but if mint22 doesn't have performance comparable to LMDE6 maybe it's better to try to make LMDE6 work, right?

Of course, it would be better to be less rigid and facilitate installations on Ubuntu.... but here I make the impression of Don Chishotte.

... please tommylight ... cancel these post row....

straight cutting in the X or Y direction seem fine, It is when i have both axis moving together doing a cut it shows up. direction seems to make no difference. i have not seen any kind of gouge marks. it like a very slight chatter. i have tried micro stepping from 4,000 to 12,000 and that made no noticeable changes to rule out resolution being an issue. it seems like osculation either mechanical from a belt flexing against the pinion i guess tuning on the feedback could look that way as well, DOC and speed seem to make no difference.it only shows up on radial cuts i have not seen any surface finish issues facing or on the floors of any cuts. i can go 1" deep and cut 1/16 with the side of a 5/16 3 flute at 30 to 40 IPM on my Y axis and things look great? Like i said it's just really odd and hard to pin point a cause.

the belts are a 3mm pitch arc tooth and i think it has a .907 pitch diameter and the OD is .872 (24 tooth)

same drives and i have not done anything with the settings internally. i do have the cables for that just never got the far into the tuning of the drive.

things are set in inches i LCNC

i'm not losing steps the closed loop feedback to the drives seems to be doing what it should be doing for that part of it. i have the drives dealing with everything too on that end.

Aciera wrote:

I certainly hope there is no such wording in any of the documentation.

I think it's not painfully obvious - at least to me.  The G-code Overview section:

I think it's possible to misinterpret that statement.

Not until you dig deeper in other sections of the manual is the behavior a bit more clear.  From G-code/Coordinates/G92 Persistence Caution (5.3):

I think the solution is to remove the #5210 parameter, and add G92.2 to the very beginning of the toolchange sequence, and G92.3 to the very end.  That way any offsets used in a program are restored and not just erased after a tool change.

Freak wrote:

Who knows if is possible to use these boards in a industry plant? Is possible to certify a machine using these boards? (CE for example here in Europe)
 

Most likely no, usually components that can also be used individually, need to be certified, especially electrical ones. So to certify a machine with drives an motors, they should also bear their own CE.
To certify the drives, you have pretty much 2 options.

1. You do it on your own.
This is quite a big task, the only "difficult" one will be the EMC stuff, you need a radio room, you can find example in some universities etc. and then you need to fix your mistakes. Don't worry there will be mistakes and shit to take care of.
The other stuff is more documentation stuff like, manuals (EU is a bitch as technically the CE requires manuals to be in the target country's language, so in most cases english is not enough), risk assessments etc. This takes a lot of time and a lot of effort.

2. You give the task to a company like Intertek, TUV...
I have always gone this route, expensive but then you get all CE, UL, CSA done at the same time and you are somewhat covered as someone else signs your work. Even in this case the EMC is always done before in another place. Some schools example rent those radio rooms out 10x cheaper what the time at intertek costs. So you try to do everything that you pass all the radiated and conducted emissions at the final test site. This will save you thousands.

3. Mix of the 2, you do it on your own and get someone to consult and help.
Probably cheaper than example the intertek route but a lot of work.

Some of the directives related to servos would be:
Machinery Directive
2006/42/EC
EMC Directive
2014/30/EU
Low Voltage Directive
2014/35/EU
Then most likely ROHS and REACH (but I think its hard to buy components now days that dont comply to both).

Then to find the subsections from those, I would ask someone that actually knows about this. Or see from any "major" brand like yaskawa, panasonic, mitsubishi, siemens... and just go with what they certify with (they will all have them on the manuals).