Current status of the manual assisted placement.

[lilltroll]

Hi

I'm interested in your PnP machine.
My need is very similar to your original needs, assembling prototype PCBs.

I need to be able to place some 0.4 mm pitch components.
I did that manually under microscope on the current PCB revision, but shaky hands made that a pretty annoying work. (The 10X stereo microscope made the visual part problem free).

I am wondering about the status of the manually assisted placement* and also what is the smallest delta step possible in X,Y + component rotation of the steppers in manual mode?
What is the alignment precision of the top camera vs the placer needle vs the bottom camera ?
Would I need a small third macro camera close to the vacuum needle for human visual feedback to achieve the precision for 0.4 mm pitch IC to ensure that the IC is in the correct position relative the pads?

* 3: Manually assisted placement: Machine takes a picture of the part, picks it up and places the image as close as the real position that it can; the user fine-tunes the placement manually. I have done some work towards this. Obviously the snapshot image quality needs enhancement, but this image tells the idea:

[JuKu]

> I am wondering about the status of the manually assisted placement

The basic building blocks for that do exist. Right now, the effort goes to getting the first batch out, but I'll implement that as the next development step. That will be out hopefully in April and certainly in May.

> what is the smallest delta step possible in X,Y + component rotation

Theoretically (using 8x microstepping) X, Y step is 0,0125mm, rotation step is 0.05 degrees. However, the microstepping is not a fully accurate method for resolution enhancement, it only gets the motor somewhere between adjacent steps. In other words, if the microsteps would be 1, 2, 3, ... and you ask the machine to go to 2, it only gets you somewhere between 1 and 3. Still, I would say there is enough resolution. In practice, if I move the cantry in 0.01mm steps manually, I might not see a difference on one step, but doing two or three steps definitely shows movement.

> What is the alignment precision of the top camera vs the placer needle vs the bottom camera ?

About 0.1mm - 0.05mm

> Would I need a small third macro camera close to the vacuum needle for human visual feedback to achieve the precision for 0.4 mm pitch IC to ensure that the IC is in the correct position relative the pads?

We'll see for sure when the feature exist, but I don't think so.

[lilltroll]

Any news here ?

[JuKu]

Not yet. This is next on my to do list, once I get my machine back from Fedex (I used my own machine at Makers Faire UK). I might have something before I take off for Makers Faire Bay area, but don't hold your breath; most likely, I'll have something to announce late this month / early next month.

[cheeef]

Any news on this?

[JuKu]

What I got is not good enough. Still searching for good solutions.

[cheeef]

Ok, do you have a working solution to place fine pitch integrated circuits like a QFN32 with 0.5mm pitch?

[Spikee]

I do not see why you couldn't manually assist / Jog the parts into place in the current state of software.
The finest resolution that the jog allows is 0.01mm which should be plenty.

Somewhere at the end of this week / weekend I will test this with a few 0.5mm qfn chips on a design I have to assemble. Will report the findings.
But I do not foresee any issues regarding this as I have jogged other parts into place before.

[mrandt]

You may want to try with the latest rmod / version 2 software.

Reza got quite good at edge detection which worked reasonably well for QFP (no leads).

I think generally speaking there are two approaches:

a) Detect part center (based on edges) and measure exact position and possibly rotation using down cam before pickup, correct needle position during pickup and rotation before placement

b) Pickup part, move to upward cam, detect edges and / or leads, measure offset from needle and rotation, correct both position and rotation before placement

Using high resolution cameras I can't see a reason why this should not work - though I am not able to program that myself

Cheers
Malte

[Spikee]

The reza software that I downloaded from the "official" post works correct except the tape settings (as i posted in the topic).
It may possibly be an old version of the software ... not sure. I need to find someone to send me the latest version as a release build so I can just install / run it.

Because spending hours figuring out why visual studio is accepting my references is not what i'd like to do