LitePlacer.com forums

So I tried OpenPNP this weekend, and it was a lot less problematic to get working than I was expecting based on previous comments on the forum. Tested a board placing 0402's.

I was using OpenPNP latest Stable and a newer standard LitePlacer build + the recent upgrades (new cameras) and Samsung nozzles. TinyG firmware version 0.97, build 440.20

To facilitate switching between LitePlacer and OpenPNP, when I built my Liteplacer, I used a DPDT switch on the Z Min/Max limit switches to switch them around as LitePlacer and OpenPNP have them switched the other way around.
(there's no need to make any Z Motor stepper changes)

Then switching between LitePlacer and OpenPNP is as simple as:

OpenPNP: Save controller settings in Liteplacer, Flip the DPDT switch and reset the TinyG, start OpenPNP

Liteplacer: Flip the DPDT switch and reset the TinyG. Start Liteplacer and load back in the controller settings you saved above.

I copied a lot of my settings over from Liteplacer and TinyG, for example, use "$$" on the TinyG from the Liteplacer console to get all the settings BEFORE switching to OpenPNP, then copy those into the gcode driver CONNECT section as appropriate (i.e. acceleration, jerk, feed speeds etc.)

Also a lot of your current settings in Liteplacer (nozzle offsets etc.), can be copied in.

A did have to make settings changes to previously suggested openpnp machine configurations, some of the changes were:

1. Adding the Samsung Nozzles
2. Increasing machine timeout
3. Enable / Disable to switch on/off motors
4. Camera resolution increase

The following are working for me:

• Z Probing (I just set the height of my board/components to -40mm) and it works
• The same visual homing dot (i.e. black dot on white background), but I had to change the vision pipeline to invert

Pros:
The main advantages are the Up camera, improved vision and better choice of feeders.

Cons:
The only irritations I've found are openpnp is not aware of it's axis limits, and also there are no "No Go Zones", like the Nozzle area and no rotation backlash and no "Force" nozzle status. (I'll probably get around to fixing those). So caution is needed when moving around with Jog and when setting up the nozzles co-ordinates.

I used the instructions and the machine config file here as a starting point:

https://wiki.apertus.org/index.php/Lite ... nPnP_Setup

And here's my final machine configuration (including nozzles, 0402 strip feeders and visual homing (black dot on white background)):

Out of curiosity, Is it possible to "personalize" the LitePlacer code? Quick question, what language is it written in and what development system is needed? It would seem that the creator is open to contributions, so how would one go about contributing?

robert@serve101.org wrote:Out of curiosity, Is it possible to "personalize" the LitePlacer code? Quick question, what language is it written in and what development system is needed? It would seem that the creator is open to contributions, so how would one go about contributing?

It is. Written in Java (works on Windows/Mac or Linux). You should be able to develop on any of those platforms, you just need to install JDK and I believe Maven (I haven't
yet, but have installed those for other open source projects with no issues). The creator is open to contributions, provides guidelines on how to do it, maintains standard and a lot of
code is contributed. Also the community is large as it covers a lot of machines. It's all on GitHub.

Results so far (I'm not finished tuning) have been surprising. Vision pipeline works well with the OSHPark Purple and Up Camera produces a perfect placement
of 0402's 0 or 180 rotated (frankly I'm stunned with the accuracy).

I just have one issue left with 0402's accuracy when rotated 90 or 270, which I'll track down then I'll post a picture of the accuracy.

It's interesting to see the difference in accuracy between use without Up Camera Vision and with. OpenPNP has Up Camera Vision. With it switched off, component dance on the nozzle and alignment of the component in the tape pocket and rotation all conspire to produce component alignment issues.

Alignment/tuning complete.

Amazing how accurate the machine can be (I have the fixed belt attachment points).

Here’s a board placed with bottom vision in OpenPNP 0402’s, a SOT23, 0.5mm pitch FPC connector and 0.4mm pitch molex mezzazine.

OSHPark board (the weird board reflection is due to post it note type glue being sprayed on the board) for adhesion.

Click on picture to fully view the board, SD Card for size reference.

And for comparison purposes, this is the same machine and settings in OpenPNP but with the Up Camera not used for placement.

This is a really good case for why an Up Camera is required in the process, hence why I switched to OpenPNP.
it’s a combination of errors, alignment of the component in the tape hole, nozzle dance when the object is picked up, error magnification due to small offset when rotated.

I really need to work on the up camera usage. (Actually, I am, but my progress is slower than I'd like).

> And for comparison purposes, this is the same machine and settings in OpenPNP but with the Up Camera not used for placement.

What is your pickup vacuum time? I got such imperfect placement when i had 50 ms, and it turned (relatively) perfect when i set 100 ms (and further increments had no effect).

wormball wrote:> And for comparison purposes, this is the same machine and settings in OpenPNP but with the Up Camera not used for placement.

What is your pickup vacuum time? I got such imperfect placement when i had 50 ms, and it turned (relatively) perfect when i set 100 ms (and further increments had no effect).

2.5 seconds (still tuning). We’re you using 0402s?

Why so long?? It's 250 ms by default.

Jet wrote:We’re you using 0402s?

I am using different parts and observed this effect on different parts (but mostly on heavy parts as i remember, but i had not tried 50 ms with 0402).