Getting started with the eC-placer

[lilstevie]

if you are well organised, have a good placement assistance software package ( theres a great free one ).  800-1000 parts per hour  for a well practiced operator is well acheivable.

Any suggestions on this? I've just been playing around with MeatbagPnP and it works, but there's definitely room for improvement

[mrpackethead]

I'm sorry but anyone claiming an actualized & averaged place rate anything approaching that with a manual system is talking complete tosh.

Look. Let's say I'm assembling 3 boards. It takes really long time. Why? Because what I need to do is this:

1. Look at the BOM. Select the next component (say it's three 0.1uF cap). Look at the board map, see where I need to place them. 30 seconds.

2. Go find the reel or bag where the component is stored, unpack, count 9 pieces, cut off, pack back, put back into place (so I can find it next time). Bring the strip back to the table. 1 minute.

3. Stick the strip to the table, make sure it's firm, peel of the cover. 30 seconds. If I screw up something, it may take much longer, and may require going back to step 2.

4. Get an air pickup and place 9 capacitors at the desired places. 30 seconds. BTW it's 1080 parts/hour rate.

If you sum this up, it's 2.5 minutes. If I could get a magic tool which would do step 4 momentarily, it would be 2 minutes. This would save me whooping 20% of my time, and instead of spending 5 hours on my 3 boards, I would do it in 4. Not much, huh. Why even bother?

But, if I had my components in feeders with automatic PnP machine, all I need to do is load the boards and press a button, and at the same 1080 parts/hour rate it does the whole job in 30 minutes while I'm away drinking coffee.

Same placement rate. Very different results.

what you describe is the reason why i have both a PNP line with 250 reels of parts pretty much peromantly on it, and a manual pnp.   Its why civil contractors have 30T diggers and a shovel.  its why  Logistics companys, have motorbike couriers and container ships.

[spongle]

It doesn't help with part organization, but Kicad IBOM (https://openscopeproject.org/InteractiveHtmlBomDemo/html/OSPx201.html) is pretty fantastic as a board guide. I have an old ipad dedicated to it at my assembly station.

[SMTech]

It doesn't help with part organization, but Kicad IBOM (https://openscopeproject.org/InteractiveHtmlBomDemo/html/OSPx201.html) is pretty fantastic as a board guide. I have an old ipad dedicated to it at my assembly station.

oo I like that, could even be a handy inspection aid

[spongle]

It doesn't help with part organization, but Kicad IBOM (https://openscopeproject.org/InteractiveHtmlBomDemo/html/OSPx201.html) is pretty fantastic as a board guide. I have an old ipad dedicated to it at my assembly station.

oo I like that, could even be a handy inspection aid

Yep, in the latest github versions they even show the traces and you can click on a pad to see all connections. My dream is something like this combined with computer vision for an augmented reality overlay on the physical board.

These guys are working on it, but I think the display tech isn't really there yet: https://www.inspectar.com/

[Jeroen3]

I'm sorry but anyone claiming an actualized & averaged place rate anything approaching that with a manual system is talking complete tosh.

Look. Let's say I'm assembling 3 boards. It takes really long time. Why? Because what I need to do is this:

1. Look at the BOM. Select the next component (say it's three 0.1uF cap). Look at the board map, see where I need to place them. 30 seconds.

2. Go find the reel or bag where the component is stored, unpack, count 9 pieces, cut off, pack back, put back into place (so I can find it next time). Bring the strip back to the table. 1 minute.

3. Stick the strip to the table, make sure it's firm, peel of the cover. 30 seconds. If I screw up something, it may take much longer, and may require going back to step 2.

4. Get an air pickup and place 9 capacitors at the desired places. 30 seconds. BTW it's 1080 parts/hour rate.

If you sum this up, it's 2.5 minutes. If I could get a magic tool which would do step 4 momentarily, it would be 2 minutes. This would save me whooping 20% of my time, and instead of spending 5 hours on my 3 boards, I would do it in 4. Not much, huh. Why even bother?

But, if I had my components in feeders with automatic PnP machine, all I need to do is load the boards and press a button, and at the same 1080 parts/hour rate it does the whole job in 30 minutes while I'm away drinking coffee.

Same placement rate. Very different results.

This is why we just order all parts counted and bagged from Mouser.
And assemble with VisualPlace.
There is work that entails sorting the bags to bom order, to prevent playing memory on one big table of mouser bags, but otherwise it’s all sitting: open bag, look at board, place parts, F7 (i think).

[Styno]

It looks like I have the same experience as mrpackethead. We have a vintage PnP machine but still use our manual Essemtec Expert quite a lot as we have for the past 20 years. It was quite an investment back in the days, more than 4000 EUR if I recall correctly, but we built many thousands boards with it. I cannot imagine that we would/could have done that with tweezers (the horror!).

It's mainly used for prototyping and weird components nowadays.

For very small runs the manual easily outruns the automatic PnP machine but when you start thinking about manual runs that last more than a day that starts to depend a lot on how smooth you can setup your workflow on both types of machines. One-off and small strips of cut-tape on a real feeder is always a pain to set-up (leader, trailer, feeder adjustments and fine tuning), the initial setup component loss from the automatic PnP is also quite high.

So yeah, there is definitely a use case for these manual machines. But -as with automatic ones- it might be difficult to find the right person willing and capable to operate it in a small company, especially if you cannot hire someone specifically for this as a full time job. We are primarily a software company and very few colleagues are willing to spend more than a day placing components (one of our receptionists likes it and is meticulous as well so she's a great asset!).

[liteyear]

Yes, we're big on only buying equipment we're willing to become adept at and have good workflows around. Too many shops with all the gear and no idea.

Shipping Update

Finally, a week after placing the order, it was marked as complete and a tracking number issued. It has not shipped yet.

[SMTech]

Here's one of the manual machines that pops up on eBay now and then https://www.ebay.co.uk/itm/OK-Industries-SMT-8001-SMT-Placement-Machine/362401276626?

For very small runs the manual easily outruns the automatic PnP machine but when you start thinking about manual runs that last more than a day that starts to depend a lot on how smooth you can setup your workflow on both types of machines. One-off and small strips of cut-tape on a real feeder is always a pain to set-up (leader, trailer, feeder adjustments and fine tuning), the initial setup component loss from the automatic PnP is also quite high.

That has be be highly dependent on the machine, I don't really have any of those issues bar the one-off parts and very small strips (5 is too short, 10 wouldn't mind a trailer on it), also not a fan of SOICs in a tiny cut off piece of tube.

[Jeroen3]

Here's one of the manual machines that pops up on eBay now and then https://www.ebay.co.uk/itm/OK-Industries-SMT-8001-SMT-Placement-Machine/362401276626?

We just discarded one of those because of missing parts and no recognizable branding or model whatsoever...

[kylehunter]

Yes, we're big on only buying equipment we're willing to become adept at and have good workflows around. Too many shops with all the gear and no idea.

Shipping Update

Finally, a week after placing the order, it was marked as complete and a tracking number issued. It has not shipped yet.

Awesome! Please let us know your thoughts. I will be buying it depending on your thoughts.

[kylehunter]

Yes, we're big on only buying equipment we're willing to become adept at and have good workflows around. Too many shops with all the gear and no idea.

Shipping Update

Finally, a week after placing the order, it was marked as complete and a tracking number issued. It has not shipped yet.

Any update? You received it yet?

[liteyear]

Arrival!

The unit actually arrived on 11/11/19, but a few days prior my wife gave birth 4.5 weeks ahead of schedule so the great unboxing had to wait a couple of weeks!

Anyway, here's my immediate reactions:

• Well packaged, with custom foam protection everywhere.
• Not much assembly required - basically just bolting the arm tower to the base.
• Instruction booklet surprisingly poor for an American company - unclear photos and missing steps.
• The lazy susan rotating base is a solid bit of kit - serious ball bearing design and high quality materials. Bit of a pity, because it seems to be the least useful part of the machine. Our intention is to ditch it for a custom non-movable base.
• Pump, pedal, switch, etc. all are intuitive and feel fit for purpose. The power pack came with an American plug, but a few seconds with the pliers and we have a perfectly fine Australian version.
• The pump hose fits loosely into the top of the pick-up tool, and each of the pickup tips fit snuggly on to the end of the pick-up tool. So easy, toolless set up.
• Something, probably the aluminium component bin case, has sharp edges. Pretty easy to end up with painless bloody cuts after handling.
• The hand rest truck and the slides it sits on are all very smooth and comfortable - but only if you're positioned higher than I was expecting. That is, your forearm really needs to be horizontal so it can move in a x-y plane, which means your shoulder ends up an upper-arm's length above the hand rest truck. That puts your head at least 50cm from the pick-up tool tip, which for me is uncomfortably far from the circuit board. But I might get used to it over time.
• Up/down motion, and x-y motion is well weighted, but both suffer from non-consistent stiction. In the up-down direction the collar often catches in a stick-slip fashion, and in the x-y motion the movement is smooth in the arm but then high-friction once the arm mounts have to slide on their shaft. This results in some awkward movements, that I'm not sure will become more natural with time. Given that it's this x-y-z movement that the unit fundamentally provides, I was expecting it to be more consistent.
• As well as 3 different sized pickup tips, 3 different suction cups are provided, but there is no instructions on how to use them. I assume they become female parts to the pickup tips, but they don't fit well so I'm not sure how to get the best out of them.

All I've done so far is pick up a few grains of rice and drop them on a postcard. It works! This week we'll be putting the unit to task on an actual PCB, so stay tuned for the usage report.

[tech_builder]

Hi,

I am curious about your thoughts after having had the machine for a little while now. Did the stiction issues clear up? Did any other issues arise that weren't obvious at first?

Thanks

[VladaAca]

...can You align and place accurately QFP's without camera?

[liteyear]

Yeah sorry for the radio silence. As I mentioned, we had a baby and 3 years later I still haven't recovered.

We did end up putting the ezPick to work. After a bit of experimentation we ended up creating a dedicated "Population Station" for it which housed the ezPick itself, pump, foot pedal, lights, reflow oven, IPA, brushes, squeegee, etc. And we created a few accessories for it, like:

• a long sheet of ~6mm timber to use as a work surface, instead of the included turntable. It can slide left and right to facilitate loading and to maximise the reach of the ezPick. I even added a solder inlay of our logo. Alas, no photos.
• 3D printed runners for aforementioned timber sheet.
• Various tubs for component and accessory storage.
• 3D printed c-section to clamp the ezPick to the bench, so it doesn't tip over when you lean on it.

And we iterated on a scheme to convert you BOM from boxes and boxes of bags and bags of components, into a sheet where all components are laid out and labelled for speedy picking and easy storage.

Unfortunately I have few non-baby photos from that time and it was all a blur. But there were some really cool ideas that came out of it.

Now, usability:

• yeah we smoothed the stiction issues sufficiently with a bit of lubricant (mineral oil I think).
• there's no better way to pick and place than with a vacuum pen - picking from cut tape is a nightmare without it, and rotating prior to placement by rolling your fingers on the pen is a dream.
• coupled with iBOM, it's a great to have your components laid out before you start a population, so when you inevitably get interrupted, everything is where you left it and iBOM tells you where you were up to. We ended up adding a touchscreen on a floating arm so iBOM was always within reach.
• because there's a setup time, it's important to have a dedicated station with power, the pump, the screen and all the nozzles and accessories ready to use, otherwise it's too tempting to just grab some tweezers and get going - that's always a more frustrating option but just so convenient. So using the ezPick has to be convenient too if it's going to become part of your normal process.

Conclusion: this is a way more comfortable way to manually place, with far fewer risks (dropping parts, losing your place). But the workflow has to be a bit different to not require excessive setup/pack-up time. So if you're used to dumping the contents of your cut tapes onto the all-purpose soldering mat and going to town with tweezers, then you have some new habits to practice. It's worth it, but the gratification is delayed.

Final thoughts: I just bought a "hobbyist" automatic pick and place. Regardless of how convenient the ezPick is, manual PnP is mind-numbing, requires a lot of concentration, and absorbs the time of a human whose mind could be applied elsewhere. Beyond a couple hundred components (eg. a couple of complex boards or 5-10 simple ones), it's robotic work for a human. Automatic PnP will never fully replace manual PnP, but as I get older I get more interested in robots doing robotic things and humans doing human things. My hope with the "hobbyist" PnP is that I can massage it into being my co-bot - working alongside me on the mundane while I focus on the bits that require squishy fingers and grey matter.


In the last image (IMG_4117 2.jpg) take note of the two sheets of card towards the centre of the photo with the cut tape stuck on them. The sheet material is the stuff our stencils come packaged in, the cut tape is stuck down with two-sided tape, and the BOM printout is autogenerated from a KiCad script. This is the process we were iterating on to go from DigiKey bag to ready for the ezPick. You can just drop these sheets on to the ezPick table and pick from there.

[liteyear]

can You align and place accurately QFP's without camera?

Well, in my mind you can do that with tweezers, provided you have a steady hand and good visual guides on the board. And the ezPick is quite a bit more steady (no tweezer shake), so it only gets easier. If the paste is bad or the visual guides are poor, then all bets are off.

A camera probably makes it quicker because you can be more confident in your alignment as you make the final z-axis movements. Without the camera you need to pause just above the surface and confirm with your eyes. But in our experience the camera looked like a cool novelty that we never justified doing. We had a trinocular microscope for big screen visuals.