Open Placer A New Affordable PCB Assembly Machine

[NorthGuy]

No matter who you are or how much experience you have - it takes time to fiddle with all the little details to get a new design loaded, programmed, proofed, and placed.

This is a very inefficient approach. You spend time once - when you set up your machine. You can fiddle with the software, you can write your own pieces, and then everything gets automated and loading a new design takes one click. It only takes few days, perhaps couple weeks to set up automation. Manually fiddling with every design would take lots of time. Probably doesn't matter if you do big production runs, but for prototyping it's probably easier to place manually.

Even loading feeders takes considerable dexterity.

Probably. But say, I have a BOM which contains 70 lines. But 3 most populated lines may encounter for more than half of parts. This means 3 feeders is enough to place most of the parts. 30 permanently-fixed feeders will cover most of the parts. Loading 30 feeders once in 3 years doesn't scare me much. The rest would be ICs from trays, and, if something is left, I can place it manually. This is really nice way of doing prototyping.

This is not my idea. Someone on this forum suggested this. But I think it's brilliant. I would be doing this already, but $10K for a machine with real feeders seems too steep for a prototyping tool

[ar__systems]

This is a very inefficient approach. You spend time once - when you set up your machine. You can fiddle with the software, you can write your own pieces, and then everything gets automated and loading a new design takes one click. It only takes few days, perhaps couple weeks to set up automation. Manually fiddling with every design would take lots of time. Probably doesn't matter if you do big production runs, but for prototyping it's probably easier to place manually.

I take you are not talking from your own experience? Maybe if you have a top of the line SMT equipment that is super reliable, it could be a reality, but for cheap 10K Chineese machines it's just not going to happen like that. I'm now 2.5 years working with Chineese machine, and I made significant progress, but I'm still not where I'd like to be (knowing EXACTLY how much time an assembly will take).

Probably. But say, I have a BOM which contains 70 lines. But 3 most populated lines may encounter for more than half of parts. This means 3 feeders is enough to place most of the parts. 30 permanently-fixed feeders will cover most of the parts. Loading 30 feeders once in 3 years doesn't scare me much. The rest would be ICs from trays, and, if something is left, I can place it manually. This is really nice way of doing prototyping.

This is not my idea. Someone on this forum suggested this. But I think it's brilliant. I would be doing this already, but $10K for a machine with real feeders seems too steep for a prototyping tool

That's just common sense, nothing brilliant there

[ar__systems]

the trick to working with cut tapes is not to use feeders.
The trick is to use a flat platen covered in double sided adhesive.

IMHO it is even worse than strip feeders. Works ok for paper tapes, but completely useless for plastic tapes.

[rx8pilot]

the trick to working with cut tapes is not to use feeders.
The trick is to use a flat platen covered in double sided adhesive.

IMHO it is even worse than strip feeders. Works ok for paper tapes, but completely useless for plastic tapes.

I tried the double stick tape method and it was frustrating for sure. As stated - the plastic tapes are even more challenging.

These strip holders are adjustable in 4mm increments to handle anything from 8mm and up. 6in or 12in strips.

[NorthGuy]

I take you are not talking from your own experience? Maybe if you have a top of the line SMT equipment that is super reliable, it could be a reality, but for cheap 10K Chineese machines it's just not going to happen like that. I'm now 2.5 years working with Chineese machine, and I made significant progress, but I'm still not where I'd like to be (knowing EXACTLY how much time an assembly will take).

No, not from experience If even $10K Chinese machines are so unreliable even for prototyping, then Open Placer doesn't have much chance.

I have read some reviews here and from what I read I concluded that $2-3K machines are complete junk, but $10K machines are Ok. What machine do you have? Can you explain what are the typical problems you're experiencing, and what exactly is unreliable?

[ar__systems]

No, not from experience If even $10K Chinese machines are so unreliable even for prototyping, then Open Placer doesn't have much chance.

That's wrong way to put Making machine efficient for prototypes is much harder than making it efficient for small series runs

For example, if I have a board of 160 parts total, and 50 unique parts, a minimally trained tech can do 1 board in a couple of hours easily. If I have to setup the machine for that assembly, I'll easily spend 2 hours or more just setting up the parts. If I need to do 10 prototypes, then I'll be better off with the machine.

I have a TVM 920, which is ok mechanically, but the s/w is crap. Going from CAD to their s/w requires manually checking orientation of the part in library vs in the tape or tray, assigning the height  to the parts, assigning nozzle to the parts. So if it a brand new design with a bunch of parts I never used previously, all that adds time. In addition you need to decide which nozzles to use and configure the s/w appropriately. Then you need to adjust each feeder. Machine feeders all vary slightly, so replacing one with another moves pick up location slightly. For strip feeders adjustment is also required. 4mm pitch is nominal and depends on the tape. Over length of 30 parts a deviation of 0.5mm can easily accumulate. So it's not just "take part A from strip 1", fire-and-forget.

I have developed very elaborate s/w to translate Eagle files into QIHE s/w which makes it almost "1-click". In another 6 month I hope to be able to outperform manual assembly of 2 boards

[mikeselectricstuff]

Going from CAD to their s/w requires manually checking orientation of the part in library vs in the tape or tray
I have developed very elaborate s/w to translate Eagle files into QIHE s/w which makes it almost "1-click". In another 6 month I hope to be able to outperform manual assembly of 2 boards

+1 - Having a smooth path from CAD, libraries with consistent rotations etc. makes a huge difference to setup time

This is a key difference between having your own machine for your own projects, and a subcontractor situation where you have to take designs in from other people. The requirements from the P&P software are somewhat different between these 2 use cases

[Smallsmt]

Over length of 30 parts a deviation of 0.5mm can easily accumulate

@ar_systems
Do you use uncalibrated feeders on your TVM920??

[ar__systems]

Over length of 30 parts a deviation of 0.5mm can easily accumulate

@ar_systems
Do you use uncalibrated feeders on your TVM920??

That was in relation to STRIP feeders, not reel feeders. If you lay 2 tapes side by side, and match their first holes, the 30th holes won't necessarily match.

On another note, I would love to know how to calibrate the Yamaha CL feeders.

[ar__systems]

libraries with consistent rotations etc. makes a huge difference to setup time

In my process I'm kind of independent of libraries. Since I read the Eagle files, (as opposed to XY file most assembly places would want) I know exactly how the part is oriented in the library. For many packages there is a standard way to orient them in the tape, so I hardcode that info into the s/w. If the part is off-centered in the library, that is another source of issues, and again, by reading original data I can correct it automatically. In the beginning correcting angles was a huge pain, since it was done in the QIHE s/w, which is horrible and does not allow easy fixes for many common scenarios. I.e. if you move part from front to back feeder, it does not fix the rotation automatically Right now my s/w is done to the point there I don't need to edit anything in QIHE, all of the data preparation is done in my s/w, in a way that makes sense.

I'm actually aiming to be able to assemble other people's boards efficiently, despite knowing that it is probably a bad idea

[NorthGuy]

I have a TVM 920, which is ok mechanically, but the s/w is crap. Going from CAD to their s/w requires manually checking orientation of the part in library vs in the tape or tray, assigning the height  to the parts, assigning nozzle to the parts. So if it a brand new design with a bunch of parts I never used previously, all that adds time.

I have seen OpenPnP Web site, where they reverse-engineered TVM920 interface:

https://github.com/openpnp/openpnp/wiki/TVM920

Then you can throw away crappy manufacturer's software and go with your own. This seems easy enough to me. Since it's only for yourself, you don't need much of a user interface. Have you looked at this option?

Then you need to adjust each feeder. Machine feeders all vary slightly, so replacing one with another moves pick up location slightly. For strip feeders adjustment is also required. 4mm pitch is nominal and depends on the tape. Over length of 30 parts a deviation of 0.5mm can easily accumulate. So it's not just "take part A from strip 1", fire-and-forget.

That's a pity. I though the holes in the tapes are done for automatic positioning, so when you position the hole, the part is automatically positioned over the pick place. If it doesn't use holes as a reference, then precision will not be good - if anything, the paper expands and contracts when its moisture content changes, plastic expands and contracts with temperature. It's surprising it works at all.

If it doesn't use holes for positioning, may be it is possible to use vision to make sure it picks correctly? Does your machine have an overhead camera which you can use when picking from the feeders?

[ar__systems]

That's a pity. I though the holes in the tapes are done for automatic positioning,

They are, and it it works for reel feeders. Again, I was referring to strip feeders. Hoping to just stick the tapes into slots and peeling the covers is not enough.

OpenPNP... I know people tried it and but apparently it is not very convenient as far UI goes. I loaded it once, and I just could not understand it. I'm doing my own.

[ar__systems]

, may be it is possible to use vision to make sure it picks correctly? Does your machine have an overhead camera which you can use when picking from the feeders?

That would be a nice feature. They all have overhead camera but I don't think any of Chinese machines do that.

[NorthGuy]

OpenPNP... I know people tried it and but apparently it is not very convenient as far UI goes. I loaded it once, and I just could not understand it. I'm doing my own.

You don't necessarily need to use their GUI. The important thing is that they found a way to disconnect the original software and drive the machine directly, which simplifies writing your own software to control it.

If you write it for yourself, you don't need much of a GUI - just a command line executable which reads the control files and execute, spewing out the log. It is much easier to calculate coordinate transform in your own program than to figure out how to explain it to the manufacturer's software.

[NorthGuy]

They are, and it it works for reel feeders. Again, I was referring to strip feeders. Hoping to just stick the tapes into slots and peeling the covers is not enough.

That's what everyone using the machines seems to agree upon - only real reel feeders are reliable - everything else is PITA. But, between the cost of the machine and the cost of the feeder, buying the whole reel (instead of cut tape) is not such a huge expense for most components. Even for more expensive parts, many suppliers offer custom reels by cutting tape and then putting it on an empty reel, or you can do it yourself. Have you tried these?

[ar__systems]

You don't necessarily need to use their GUI. The important thing is that they found a way to disconnect the original software and drive the machine directly, which simplifies writing your own software to control it.

If you write it for yourself, you don't need much of a GUI - just a command line executable which reads the control files and execute, spewing out the log. It is much easier to calculate coordinate transform in your own program than to figure out how to explain it to the manufacturer's software.

Yeah, that part of work was almost done, but it was not complete, which probably explains lack of wide adoption of openpnp for tvm920.

It won't work without UI.... It is not a fire-and-forget system, not for a while at least. Feeders jam, parts fall of the needle, vision system needs adjustment etc etc.

[NorthGuy]

Yeah, that part of work was almost done, but it was not complete, which probably explains lack of wide adoption of openpnp for tvm920.

It wouldn't be hard to complete. Then you make the software as good as you want.

Feeders jam, parts fall of the needle, vision system needs adjustment etc etc.

This is serious stuff. You certainly cannot improve this by changing software (perhaps may be by decreasing speed and such). I guess one part falling on an almost-placed board would mess up things quite a bit. Not something you want to deal with.

So, your advise would be to avoid TVM920 (TVM925 now) machines for prototyping purposes, right?

[rx8pilot]

It is not a fire-and-forget system, not for a while at least. Feeders jam, parts fall of the needle, vision system needs adjustment etc etc.

Yes - there is always something.....even with more expensive systems. My setup is a full-on commercial system from the 90's. It is physically very robust, precise, accurate, repeatable.....yet still mis-picks and messes up. The feeders may not advance, jam, not be repeatable. Each job is loaded with little tweaks and fixes. Some jobs, I use the same feeders as the last job......some jobs I have to load 20 feeders and 10 cut tapes and 3 tubes. Each one requires settings, pickup test, placement test and a final verification. It is very easy to cross similar parts and your 20 placements of 1k resistors just became 100k. Or all the diode are backward.

I feel like for prototyping - a system that uses vision to image each pickup would be good. Vision alignment is very very cheap and can overcome a myriad of mechanical deficiencies. In a production world, it may be too slow but for prototyping, setup effort dominates the total time. If I could have marginally accurate feeders and marginally places cut tapes where the machine can find the edges of the part - it would save a lot of time and expense. In a budget-constrained world of makers and DIY - leaning on cheap vision instead of very accurate mechanics seem like it would yield a better solution overall.

Going from CAD to their s/w requires manually checking orientation of the part in library vs in the tape or tray
I have developed very elaborate s/w to translate Eagle files into QIHE s/w which makes it almost "1-click". In another 6 month I hope to be able to outperform manual assembly of 2 boards

+1 - Having a smooth path from CAD, libraries with consistent rotations etc. makes a huge difference to setup time

This is a key difference between having your own machine for your own projects, and a subcontractor situation where you have to take designs in from other people. The requirements from the P&P software are somewhat different between these 2 use cases

This is CRUCIAL to making it all work. I too put in considerable effort to get to a '1-click' from CAD to P&P. That effort was very well spent and drastically reduces the effort for new designs. I also scanned eBay like a hawk and have been continuously purchasing various feeders. I now have 200+ feeders which is enough to keep most parts loaded all the time, still, have leftovers for new parts and spare parts when they break. Early on when I was very feeder limited, I spent enormous amounts of time swapping parts in the feeders.

Two years after the initial acquisition.....I finally have the process worked out. A lot of hours and a lot of additional money to make it a practical process.

[jmelson]

I guess one part falling on an almost-placed board would mess up things quite a bit. Not something you want to deal with.

Rarely a big problem.  I occasionally have this happen on my Philips machine.  A part (usually a small passive) landing randomly will have a huge probability of landing on an area with no pads and paste (at least on my lower-density boards).  So, it just falls off in handling.  I guess it could land in a spot where a part will later be placed.

Anyway, this is a pretty rare occurrence.  Usually, if parts are going to be dropped, they fall when the nozzle raises over the feeder and the centering jaws touch it.  If bad vacuum is detected, then the head moves over the dump bucket and cycles the centering jaws a few times.
Either the vacuum improves and the part is placed, or it falls off there.  If vacuum is still below limit, then the part is dumped in the dump bucket and a new part is picked.  So, dropping a part on the board happens a couple times a year, maybe.

On a machine without centering jaws, I'd expect it to be even more rare.  Either the nozzle picks up the part, or it doesn't.

Jon

[rx8pilot]

Shipping has started! Fingers crossed.

Ooooooohhhhh....An unboxing!!!

[TJ232]

Shipping has started! Fingers crossed.

IGG definitely needs to improve their recommendation system.
I'm NOT interested in that!

You are, you are, you don't know yet but you are!
Inside are some very good vibration motors perfect for your vibration feeders 

[Richard Crowley]

The IGG page seems to be closed, and their website is just a nearly-empty boilerplate.
Will this be an ongoing product or was this a one-time event?

[rx8pilot]

When I released my first product, I was so overwhelmed with all the un-anticipated details that a wedsite was low on the priority list.

Perhaps this is similar.

Short and misplld from my mobile......

[RickJames]

As I understand it, this is supposed to be a product line that will be for sale at some point. The point of the IGG fundraiser was to get the word out. I've had trouble with the website too. Tried it again today and finally got through. You can get to it by following the "ALL PRODUCTS" link on the Store Page or you can use the link below. Their main site lists them as a "Kale CNC brand" and they're sold through that site.

http://www.kalecnc.com/index.php/shop-with-sidebar-2/

The devices are currently listed as sold out (as of this post), but parts for them are available. I pre-ordered a mini so I'll give my 2 cents on it when I get mine (which should be soon).

[RickJames]

Just got my Open Placer Mini and got super excited when I saw the package. Sorry guys, not gonna do an unboxing, but I'll give my 2 cents on it.

As expected for a project with this kind of budget, the Open Placer looks fairly plain. That said, overall it doesn't look or feel cheap. There are a couple of spots where epoxy is visible, but nothing major to be concerned about. Cables are routed well, the assembly is clean, and the important parts are made of metal.

It came almost fully assembled which is nice. However, it was missing the instructions for the parts that had to be assembled so now I'm waiting for a response from the team. There's a section of the website dedicated to setup, but it's missing the instructions I need.

I'll put up some more pictures and give a full review once I get it fully set up. So far it looks as advertised.