Eclipze's SMD Pick'n'Place Build....

I hadn't seen this video before, so I thought I'd share. It covers from solder paste application onto a stencil, pick and placing, to skillet plate reflowing.

There are actually several videos of this machine if you search youtube. In the video link I posted, he has to inspect the build and reposition a few parts at the end, before soldering. Not particularly accurate, but might give someone some ideas for a low-cost machine.

The machine itself isn't very sophisticated. It's made of wood, and a luer-lock needle serves as the pick-up nozzle. He uses EMC and G-code for the controller.

While it seems like it is certainly useful for making multiple boards with larger parts and non-fine pitch components, the reason I want a pick and place machine is so I can use components that CAN'T be placed by hand.

[ame=http://www.youtube.com/watch?v=CVlad7l9HvI&feature=related]redFrog Pick and Place machine - example of the pick and place process - YouTube[/ame]

Quote:

---

Originally Posted by Nails

the reason I want a pick and place machine is so I can use components that CAN'T be placed by hand.[/url]

---

Just because I can place a part by hand doesn't make me happy placing a 1000 of them...

In my experience the syringe needles work very well..

Thanks for sharing the vid

Don't get me wrong. I think that the machine is a good example of a limited, yet still useful DIY machine built on the cheap. It's just not for me.

There were many things I noticed that seemed to work well that I thought might be of interest to others.

1) Wooden construction - Significant cost savings.

2) Using V-grooved bearings and thin metal plates on the sides instead of more expensive linear rails.

3) Using a single Leur lock nozzle with slide-on rubber tips - pausing the program for the tool change, instead of needing to make several nozzles and a tool-changing mechanism.

4) Using the nozzle to advance the tape. It was interesting to see how well it worked.

5) Not having to drill a hole in the shaft of the stepper motor.

6) Securing of the board.

7) The use of EMC and G-code helper routines to make the main program more simplified and readable.

8) Seeing the skilllet reflow method. I've heard of it, but never saw it in action.

9) Seeing how the tape strips go underneath of the base, instead of curved down to the floor.

Quote:

---

Originally Posted by Nails

...There were many things I noticed that seemed to work well that I thought might be of interest to others.

4) Using the nozzle to advance the tape. It was interesting to see how well it worked.

---

Both Eclipzes and my machine use a tape feed pin. That's what we were trying to tell you ages ago in this thread that it works fine, when you were designed the mega-complex auto feeder reels... ;)

Quote:

---

Originally Posted by Nails

...
5) Not having to drill a hole in the shaft of the stepper motor.

---

I mentioned that earlier in this thread, he has used my design for the 180' rotation side hose almost verbatim. There's no need for 360' rotation or continuous rotating vacuum flanges.

Quote:

---

Originally Posted by Nails

...
9) Seeing how the tape strips go underneath of the base, instead of curved down to the floor.

---

Again, Eclipze's and my machine use the same system, mine uses double sided feeds and an elevated machine base so the tapes exit under the machine on the table.

One interesting thing he does in the video is to feed the tapes with the nozzle in the empty part pocket. That is a neat idea as it still allows feeding the tape for larger parts even with the rubber tip still on the nozzle. Innovative idea (although it does look lacking in precision).

If you want to go at any significant speed or slight production volume, you'll need a feeder.
You can buy used feeders from the market, which go for a very low price compared to new feeders. Siemens dual electric feeders go for ~$50 and the newer style dual 8mm Siplace feeders go for ~$200. The Siplace feeders advance in 40 ms, the tape is disposed at the front and the cover tape peeled and shuffled into a bin. The feeder contains 4 Swiss gear motors, lasting easily several years in 24-7 production environment.

Although initially I didn't want to get involved in SMD anymore after all jobs went to China, I started this open source feeder project to contribute with my knowledge and experience.

The open source feeder contains two motors, one for peeling which also advances the tape and one for the component cover, which is required if advancing at high speed.

Both motors are wired into an IC socket together with a gate sensor for the sprocket wheel. Else, I do not tend to supply any control electronic, because the control logic differs from machine to machine.
Feeder width is 10.5 mm which could be plugged into a control board at any number and spacing. In order to permit accurate placement of the feeders on to the control board or feeder table, the feeder contains 3 mm bushings.
The gear motor is promised to last 300 hours, which comes to several hundred million cycles at 40 ms. The pager motor for the cover/shutter does not have to be operated for 1206 or 0805 in paper tape and supposed to last as long as the gear motor.

I've almost completed the drawing phase, but will take your feedback for changes and optimize the product for your requirements.
Open source also means, I take your critique and concerns.

Regards,
Stefan

My new PnP head. Stepper is a hollow shaft stepper. The yellow pickup nozzle is an Fuji AMPH nozzle, I got 3 off ebay. The nozzle just slides in easily. I could easily build a nozzle changing station to work with it.

The two piece blue and aluminum form like a spider (think motor shaft coupler), they are held apart by 3 light springs which allows some compliance if the placement height isnt quite correct. The air easily passes through the middle yet doesnt leak much at all (the blue extends into the hollow shaft).

The spike is offset so it falls between the 8mm tapes without interference.

Wow, my new PnP head unit left everyone speechless?!?! ;)

(this thread has been quite for over a week...)


Designing my pnp head was for nothing afterall, I have a refurbished Samsung CP20V coming in this Wednesday, sweet! Full vision, laser, dual head, and fast! Yeah, I am throwing in the towel to DIY PnP. I got my DIY PNP working but I need something better and I need it now! :) I'll still be watching this thread though.

Lucky! I wish I could afford a commercial machine, otherwise I would probably have gone that route after the time and effort i've put in so far.

I'm getting really close, I have mine feeding and picking up parts, just need to fine tune it.

That is looking good! Put a lot of thought into the cover tape pullback. I really struggled with that. I assume you will be using lead fishing weights to pull the tape back? The part tapes have different thicknesses and that makes it hard to tension the part tape right so that the lead weight doesnt move the part tape while the machine is moving (vibrating). I solved it by putting different thickness of blank part tape under the thin part tapes to take up the slack.

Lucky isnt the word..."brave" or "foolish" maybe, ask me a year from now! lol. I maxed my credit, so we'll see if I made the right decision or not.

That's the problem with DIY pnp...so much tweaking. I have a commercial product that is taking off and I cant be spending all my time trying to get a PnP solution working. It just got to the point I had to take the dive and go for it. There are some good second hand equipment out there for 34k plus a bunch of feeders...with full vision, sensors galore...I'm talking real production stuff.

Quote:

---

Originally Posted by jcdillin

Lucky! I wish I could afford a commercial machine, otherwise I would probably have gone that route after the time and effort i've put in so far.

I'm getting really close, I have mine feeding and picking up parts, just need to fine tune it.

---

Oh I couldn't agree more, I run a small webshop selling built PCB's and I refuse to outsource the building to china and can't really afford to have them built in the US. So for now I hand place all boards but if I were ever to get really busy I wouldn't be able to keep up.

Good luck with your machine, post up some pictures once you get it!

You're right to not trust outsourcing out of the country. I had an India company eager to do my assembly. They talked about how safe and confidential they are, blah blah blah, I reluctantly gave them pics of my product, manuals and BOM. 2 weeks later, they wouldnt talk to me because they are making their own competing product. "oh, but we are making our own design and it wont be like yours." BS. They did their own pcb routing, but it looked like a clone of mine. surprise, surprise.

I will post some pics or youtube vids when I get it running.

Quote:

---

Originally Posted by jcdillin

Oh I couldn't agree more, I run a small webshop selling built PCB's and I refuse to outsource the building to china and can't really afford to have them built in the US. So for now I hand place all boards but if I were ever to get really busy I wouldn't be able to keep up.

Good luck with your machine, post up some pictures once you get it!

---

You can probably find someone local enough to you in the USA who will do pick and place cheaply enough. I've done a number of commercial products and farmed out the PNP placement and soldering, once the quantity of PCBs comes up it it gets cheap, about a couple of cents per component. It's hardly worth doing them yourself if you have a popular product!

Of course making DIY PNP machines is very cool too! ;)

Guru_florida- nice piece of work on that placement head. I guess you will sell your DIY PNP now you have a pro machine arriving?

I really hope my product is that popular one day! :)

It doesn't need to be "popular" as such! You can get companies to make your product (ie solder through hole and/or SMD) in quantities down to a few boards, although it gets a lot more cost effective if you do batches 100+, or better 500-1000 because they generally charge a setup fee or minimum batch charge. Even if you only sell 1 PCB a week a batch of 100 is pretty reasonable. Just look in the yellow pages and make a few calls. :)

I've never gotten assembly done before, I always did it by hand. I priced it up many times but I was always astounded by the expense. Best quote I was given was about 25'000USD for 500 units. The ROI on that was not great considering the up front cost and without knowing the (new) market. (Would it sell?)

So I scraped up 35k for a machine. The product may still not work out, but at least I am left with an asset instead of useless product if it fails. Depreciation on the unit is about 5-10k a year from now, so my risk exposure is only 10k worse case. So buying the unit is the safest option.

Now the flip-side...I'm not arguing with your points, what you say is true too - it depends. I probably didnt get the cheapest price, but the few I got were way out of the ballpark. I did my first 200 units by hand to see to the market response. If you can go "turn-key" and still make a good profit, that's a better way to go. Believe me, I've torn my hair out with all sorts of manufacturer issues that you wouldnt be of any concern with turn-key. Turn-key would leave you more time to concentrate on your business matters or the next product.

I think there might be a market for "incubator services". Where a company does much more than just turnkey assembly, it takes on some of the risk for a greater reward. I guess this has been around for years, but going with an inventor company you usually lose your whole pie. Anyone have experience with those inventor companies? Perhaps a middle road?

Quote:

---

Originally Posted by RomanLini

It doesn't need to be "popular" as such! You can get companies to make your product (ie solder through hole and/or SMD) in quantities down to a few boards, although it gets a lot more cost effective if you do batches 100+, or better 500-1000 because they generally charge a setup fee or minimum batch charge. Even if you only sell 1 PCB a week a batch of 100 is pretty reasonable. Just look in the yellow pages and make a few calls. :)

---

Quote:

---

Originally Posted by guru_florida

I've never gotten assembly done before, I always did it by hand. I priced it up many times but I was always astounded by the expense. Best quote I was given was about 25'000USD for 500 units.

---

$50 per board?

I just thought I would put a plug for the PCB Assembly House we use in Chatsworth CA. I can't say enough good things about them. When we were starting they helped us out with small batches of like 30 boards. A 25 sq inch board might cost $30~40 each for assembly. These are RoHS boards. We provide a kit of parts and one hand built sample and a BOM. A stencil for ~$200 and there is a ~$200 setup charge. If you need more assembled there is no stencil or setup charges. What I really like is on larger batches they run several boards, stop, let us test them, then proceed with the rest. (They might only do this because we are local). Turn around times are as short as 1~2 weeks. Our board yield is about 98%. There is no way I would out source to China as long as they are around. Please use them!

http://www.hartelectronic.com

Regards

Quote:

---

Originally Posted by guru_florida

I've never gotten assembly done before, I always did it by hand. I priced it up many times but I was always astounded by the expense. Best quote I was given was about 25'000USD for 500 units.
...

---

Like TomKerekes I'm also stunned by that quote! :eek:

I've had plenty of 500 Q batches done, either for my products or for clients. I would expect $150 to $300 one-off cost for solder stencil depending on size and brass or stainless. Then maybe a $200 setup fee to load the machine then on 500 Q PCBs about 5 cents to 10 cents per component for SMD placement and solder and inspect. That depends on type, larger simple components (0805, SOIC) are cheap, things like tiny parts or fussy fine pitch chips (SSOP) will slow the machine down and bump the price.

So for 500 large PCBs with say 100 parts per PCB I would expect about $200 + $200 + (500 * 100 * $0.05) total about $2900, and possibly less because generally once the price gets over $2k it gets a bit negotiable. If they were small PCBs with 25 parts that total would be maybe $1200?

Most machines will do an easy 2000 parts/hour and even at 5 cents a part they make real good money per hour. A place with a fast machine and higher quantities can get down to 2 cents a part. Those are local prices, Asia will do much cheaper but quality may suffer and the wait time and hassles grow.

Please don't think I'm criticising your purchase of a machine, it's wonderful to have machines in house. :) But it's also wonderful to mail a carton of reels and empty PCBs out, and two weeks later receive back a carton of finished inspected PCBs and your leftover parts. :)

That's awesome. I wouldnt have bat an eye at that price. Say at $3000 for 500 qty, 6$ board would have been an easy decision. I spent countless hours placing these things by hand to sell the first 200. It sucks. I had about 28 unique components per board, 3 QFNs, 1 0.5mm pitch SOIC, 1 atmel 48pin QFP. Should be pretty easy except maybe the QFNs.

I had so many manufacture problems and missed so many sales because I was out of stock the majority of the time and DIY PnP was taking too long to get working. I did learn a lot though and tweaked my PCB to get better results, so the education was maybe worth it.

To top it all off the India company I got one quote from decided to duplicate my design after promising they were so confidential and trusting. They did their own pcb placement but stole all that was unique about my product (visuals and functionality.) This was a big reason I went in-house beside cost.

I still dont regret the purchase regardless. I have a goal where I want my new company to be and to get there I definitely need in-house assembly capability. It'll pay dividends.

TomKerekes et al: plug away, I am sure many would appreciate the recommendations. I know I would have a few months ago.

C

Quote:

---

Originally Posted by RomanLini

Like TomKerekes I'm also stunned by that quote! :eek:

I've had plenty of 500 Q batches done, either for my products or for clients. I would expect $150 to $300 one-off cost for solder stencil depending on size and brass or stainless. Then maybe a $200 setup fee to load the machine then on 500 Q PCBs about 5 cents to 10 cents per component for SMD placement and solder and inspect. That depends on type, larger simple components (0805, SOIC) are cheap, things like tiny parts or fussy fine pitch chips (SSOP) will slow the machine down and bump the price.

So for 500 large PCBs with say 100 parts per PCB I would expect about $200 + $200 + (500 * 100 * $0.05) total about $2900, and possibly less because generally once the price gets over $2k it gets a bit negotiable. If they were small PCBs with 25 parts that total would be maybe $1200?

Most machines will do an easy 2000 parts/hour and even at 5 cents a part they make real good money per hour. A place with a fast machine and higher quantities can get down to 2 cents a part. Those are local prices, Asia will do much cheaper but quality may suffer and the wait time and hassles grow.

Please don't think I'm criticising your purchase of a machine, it's wonderful to have machines in house. :) But it's also wonderful to mail a carton of reels and empty PCBs out, and two weeks later receive back a carton of finished inspected PCBs and your leftover parts. :)

---

Guru_Florida, I'm sorry to hear you got ripped in India.
Two honest shops in US to add to TomKerekes recommendation:
LED Lighting, Engineering & Design, Corp.
GM Technical Services Inc::An Electronics Manufacturing Service Provider that specializes in Sub-Contract Electronic Assembly, Printed Circuit Board Assembly (PCBA), Prototypes, Quick-Turnaround (24 hours), Rework and Repair, Low Volume Production Ru

Yeah bummer about the Indian company ripping you off.

Have you got the new machine? It would be nice to hear your impressions of it and experience getting it set up etc.

That's assuming Eclipze doesn't mind of course! This is supposed to be his personal build thread! :)