Electronics > Manufacturing & Assembly

Pick and Place Machine TVM802A / TVM802B

<< < (205/206) > >>

FYI I just received software V3.31 for my TVM802BX from QiHe... I did not see major functionality changes, a changelog was also missing. They did add a user interface language switch button to the user interface and also started to implement what seems to be some kind of user management, you can switch between "Operator" and "Manager" but funny enough you can not configure these users - tse...

Will test this further later.

Some weeks ago I did a job with the machine and had quite some issues with plastic tapes in the feeders, these black ones, made from thin plastic. The pressure of the spring metal things in the feeder base were way too strong and held the tape so tight that the head was not able to pull / feed them. So I took the spring metal out and bent it to loosen up the tension, after that it worked better, but still not ideal. The thin plastic tape often tilt and twisted a little which made it getting stuck. Need to test this more when I get more of this type of tape.

BTW did anyone try the TVM802xx with other software than the one from QiHe? Like OpenPNP or such? Last time I checked OpenPNP support was discussed but not implemented.



I want to place the ICs precisely as in the video. but I couldn't make it. How can I do this fine tuning? can you help me. What adjustment could have been made in the video?

The ic is constantly inserted incorrectly.
skewed or not fully on target.
The chips are checked with the down camera but the problem persists.

So, first then let me ask a couple of questions...

How do you try this? Do you use a bare PCB and "dry" place the chip? If you do then this will not work, I tried that too and failed. You need something on the PCB that the chip will stick to or the nozzle will shift it in a random direction while retracting. I am using a pretty transparent double sided sticky tape on a test PCB, that works well for placement testing.

If you have something sticky on the PCB and still have false placement results, are the errors you see consistent? I.e. is it always the same error for the same component on the same place?

If yes, then most likely your nozzle and/or camera calibration is off - and this is the really painful part. Quite naturally the machine needs to know where the camera positions are in relation to the nozzles and eventually also a rotation angle correction. The calibration for this can take a while and you need some tools for that.

First of I would recommend either finding some PCB or printing something with a printer that looks like a bullseye. Then attach the two smallest nozzles you have, put the bullseye on a fixed spot on the machine (so it can not move) and then aim the center with the down camera. Now enter machine settings and there nozzle settings, select "measure" (IIRC, I am writing this from memory). Aim the bullseye with the camera, set this as camera reference and then aim the center with the first nozzle. This is now very painful since you can not move the head while the nozzle is down so you have to aim, lower nozzle, up nozzle, correct, lower nozzle etc. until the nozzle lands perfectly in the middle. Now do that for both nozzles and save the results.

Now your machine knows where the down camera and nozzles are relative to each other. For parts placements without using the up facing camera this should now work in regards to position. You can try it out with a known layout (e.g. the test PCB that came with the machine), select a component and right click "aim with nozzle" or "aim with camera". Nozzle and down camera should land in the center of the component.

Now for the up facing camera. There is also a calibration menu for that. What you basically have to tell the machine is where the center of each nozzle is in relation to the center of the up camera. In my machine the nozzles are black so you actually do not see much of them in the camera view. But what I found to be useful are the values in the camera view, i.e. the offsets for X and Y while lowering the nozzle all the way down to the camera until the nozzle touches the glass. Now the camera logic will try to detect a "component" and draws a rectangle around the nozzle. The offset values now tell you how far the detected center is away from the camera center. Bingo! Exactly what you want. Now up the nozzle again, move the head and down the nozzle until these values are as small as possible.

Save and you are done!

Next you need to look at the IC tray and feeder settings as well as the component settings. You can have rotation offsets in there as well as vision thresholds. If the rotation is wrong your part will be assumed rotated and lands wrong. If the vision thresholds are wrong the margin of error for positioning the component can be either too high (aka "anything goes") or too low and vision will never complete correcting the position (and eventually abort and discard).

The calibration could IMHO be done a lot smarter and much more automated by the machine. But well... it can be made working with some patience - take your favorite hot or cold beverage, sit down, relax and be patient while teaching your poor little machine, it will be very grateful by placing your parts correctly next time.

Hope this helps?


There is an easier way to get the offset cameras, I have used this on some other pick and place, the principle works.

I put a little bit of blutak or some other soft material on the board, plasticine for example ... anything soft where the nozzle will make a mark.

Lower the nozzle until it touches the board and makes a mark in the blutak ... raise the nozzle (you might have to clean it!) and now you can switch to the camera and just centre it on the mark.  Very quick, very easy.

Oh, indeed, that is actually pretty clever and avoids parallax effect errors. Just the question which kind of material to make the mark in? I am also a bit worried about the residue on the nozzle.
Hmm... maybe create $something with a hole in it which pretty much exactly fits the nozzle, move down the nozzle into that hole, fix the material in its location and then move the camera to the hole's position? But anyway, yes, doing it the way around seems pretty clever, indeed :)



[0] Message Index

[#] Next page

[*] Previous page

There was an error while thanking
Go to full version