Custom "Transformed Axes"

[Slavko Kocjancic]

Hello...


I build simple machine. It has two heads like "Peter's Head", but motion
is different.
I wish to utilize whole stroke of linear rail (so wan't unused head to
rest on top instead in middle). The kinematic is similar to
"ReferenceCamCounterClockwiseAxis", but not same.
Is it possible to somehow insert custom formula or at least table with
around 16 points for forward transformations?
My motion is similar as sin(-45deg to 135deg). So slowly start then go
fast and again before end goes slow.


Thanks...

[Slavko Kocjancic]

Really nobody know?

[Jan]

Hi Slavko!
Peters Head is a dual nozzle head operated using a single motor for
both Z axes. So I'm guessing that you'd like to use a single head with
two nozzles and one motor but different translation between motor
rotation and nozzle z motion. If that's the case, there is no option for
a custom transformation. IIRC the difficult part is, that the inverse
transformation is also needed, which makes a generic solution almost
impossible. However, you might try to combine the cam transformation
with other transformations. If that does not work you might implement
your own, which shall be not that difficult if you start from the
existing cam transformation.

Jan

[Slavko Kocjancic]

So custom formula is not possible even if I provide inverse transformation too. 

[Jan]

AFAIK no.
Other question: do you really need a new transformation or coundn't you just go with an existing one excepting that it's not perfect? Cam does the rotation to linear (sin/cosin) but does not apply limits like when the nozzles are stopped in the balance point. So if you send one nozzle up the other will silently go down. As long as you can configure PCB and feeder z and guarantee that nozzle tips and parts carried on nozzles are in safe-z when commanded openpnp should be fine.

Jan

[Slavko Kocjancic]

Indeed I need to check how much error will be using existing transformation.

The very close aproximation of my kinematic is drawn below.

The Arms angle is 0 degree, so only one arm is present. (red vertical line)

From top of arm there are two ropes (green) connected to motor slider.

So when arm goes clockwise the right motor will go down by spring/weight as rope can only release motion.

The left motor will indeed go little up as rope will seat on pivot but the efficient cam radius for left motor will be maybe 1 or 2 mm but for the right one will be 20mm.

So I expect that at very top the position will be inaccurate as machine will probably thint that botor is way higher. In midle point it should be pretty accurate and then in lower point again some inaccuracy.

The question is is that matter at all. The tool will go down by spring or weight. If hit surface no harm is done. So if I have well defined clearances it should be ok.

Does Z accuracy have some other important role?

--
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 visit https://groups.google.com/d/msgid/openpnp/D222EA88-9A1C-463A-8906-53889A22A6A9%40googlemail.com.

[Jan]

Interesting design! I guess you will be fine with an arm angle of 0° and a cam radius that reflect something between the join between rope and nozzle and arm length.
Z accuracy is IMHO relaxed because nozzles or tips are spring loaded. So +-0.1mm shall be fine.
Feeder and PCB z are teached in so there is no absolute position requirement just enough resolution.
With static Safe Z nozzles are fully retracted -> no requirement as well.
When parts are placed on the PCB they are lowered down to PCB Z plus part height. This requires either some precision or individual tuning.

Jan