Electronics > Manufacturing & Assembly

Designer/assembler looking for reasonable pnp machine

(1/5) > >>

liudr:
First post on this forum! I have been designing and hand assembling circuit boards with SMD components for almost a decade now. Building batches of boards has become boring and tiring and I am willing to invest some time and money for a pick-n-place machine. I'm looking for some advice to help me make decisions soon. I know some principles of how it works from online reading and watching videos. If it comes to price vs. work it takes to build and maintain, I would probably pay higher price to reduce my time to build the machine and maintain it.

I know amazon is probably not the best place to buy such thing but here is something I found as a reference or starting point:

https://www.amazon.com/shipping-machine-CHMT36VA-feeders-mounter/dp/B08S3CW7ZB/

I don't know the company or what software it's using, probably not open source. It looks pretty much pre-assembled. So this is my reference for almost ready-to-go systems requiring least time to get it going and more expensive.

I also looked at openpnp, which is like you build your own machine:

https://github.com/openpnp/openpnp-openbuilds/wiki/Build-Instructions

This seems to require the most time and the least money to build and is customizable but I currently don't have a lot of free time.

So my current need is simple:

My current bom has about 25 chip components, mostly 0805 because I could easily place those. I have some 0603 for tight spaces. Resistors, inductors, capacitors, diodes, fuse, ferrite beads. I also have a small number of ICs, small qfn, 24-32 pads, and also I need ESP32 module (maybe different nozzles?). I think it would be great if I could expand capabilities with say swappable reel carriers (is that a thing?) or additional reels say on opposite sides of the machine etc.

I would like to have enough space for an IC tray. I could probably saw one tray in half so one half has some different size ICs than the other half, say some TQFP on one half and QFN on the other half.

Board size I want a minimal size of maybe 4in by 4in (smaller would be better) and max size hopefully a panel of 9in by 9in or at least 4.5in by 9in with spare space for IC tray. I don't design large boards. I'm currently limited by my reflow oven as well but that's a cheaper device to upgrade to larger size I think.

So if you have some directions for me to look at, regardless of the specs of the machine you recommend, please feel free to share. With the uncertainty of covid and lockdowns, and the price hikes of overseas fabs, I think it's time for me to invest in-house small batch assembly :D

Thanks for reading!

lutkeveld:
I have the CHMT36 (without vision) and it's great for small runs.
If you do small in-house runs, also be sure to have a good stencil and reflow step.
Those quickly become the bottle-neck, both in quality and through-put, once you have a pick and place.

Good paste application is especially important if you have smaller devices like QFN's.
Rework, or even worse: rejection, takes a lot of money and time if you are running batches.

And once you hit around 30-50pcs, you probably still want to switch to an outsourced assembler.
Even though a pnp greatly reduces time spent per board, it still takes up more time than you think:
Stencelling, machine setup, reel swaps/jams, placement, post-placement inspection, reflow, post-reflow inspection, THT insertion, soldering etc.

liudr:
Thank you for sharing @lutkeveld!

I hope with vision included, the placement will be better because if all components are placed, then checking qfn parts will be challenging. I usually hand place them first so I can look all around them to make sure I have good alignments.

For stenciling, I order stainless steel stencils with boards from JLCPCB, which is good enough for small batches of small boards. Solder paste still find its way under the stencil eventually. I reflow up to 16 boards at a time (small). So far after say 16 uses on the stencil in one run the bleeding and smudging isn't too bad. I have so far not rejected a qfn since switching to qfn on some ICs due to TQFP out of stock (1 use for each board since I'm not panelizing boards yet). I may panelize in the near future so I can apply paste less times per the same number of boards. If I have a panel with 4 boards, I only need to stencil 4 times to get 16 boards done so less bleeding of paste to the bottom. I'm currently using a crude way of setting up old PCBs as guides around a new PCB and taping the stencil over the PCB to apply paste. I may have to get a good printing machine and framed stencil for more serious batches :)

So far qfns are good with hand placement. If I discover that they are off, after reflowing, I can use a hot air rework station to fix them. You have to push "hard" enough to click them one pad to the right or left once solder melts. I did some of that during the early days of using qfn.

About the CHMT36 you're using, does it run the openpnp software or some older proprietary software? I'm reading on their website that they have openpnp but not sure if they produce parts for people to build their own openpnp or their pnp are running openpnp.

Time-wise, yes the time saved from hand placing parts isn't ALL the manual labor as you pointed out already, inspection, repairs, flashing firmware and tests, then thruhole component soldering and packaging. I think this can all be done by an assembler although it will be more costly and takes time to properly communicate. If I have a product that I don't need to update, say due to chip shortage  :palm: then I can try that route. Right now I'm thinking about hiring an intern, college student, so it will be their time mostly, if I find a good intern. This way I can save my own time to do designs, updates, and a lot of firmware development.

jmelson:
I started in 2007 with a Philips CSM84, very old school machine with no vision, it used mechanical centering jaws to center parts on the nozzles.  No auto nozzle changer.  I replaced it with a Quad QSA30 (made by Samsung) with flying vision and an auto nozzle changer.  The programming is more complex, but it has a LOT of self-diagnostic capability.  It was apparently not stored in a conditioned space and had a lot of things break down shortly after I got it, so that was a downside.  But, I'm happy with my choice of getting a used, high-end commercial machine instead of a Chinese low-end machine.  I have had no problem getting parts for the Quad as it is based on a fairly popular Samsung base machine that was used widely in the East.

So, if you have the room, a used production machine might be a good choice.  Both of these machines I had could handle a WIDE range of part sizes, from 0603 up to 21mm FPGAs.

Jon

liudr:
Thanks for sharing @jmelson. I'm not ready to purchase retired production quality machines. There's too much I don't know about pick and place machines. This first machine will hopefully be a positive enough experience so I can forge ahead with learning more and maybe eventually go for production quality stuff, used or new. I don't know if "Juki" means anything in the world of pnp machines. This low-end table-top system claims to use these nozzles.

Navigation

[0] Message Index

[#] Next page