Advice on setting up a low-mid volume SMT line for <$130k?

[jmelson]

In addition to the air filtration I'd like to get a nitrogen bottle to see if nitrogen soldering is really so great - manual soldering, not reflow.

Hmmm, never heard of nitrogen manual soldering!
 

I have to say I'm just a little bit underwhelmed at the JBC. Even the large tips on the T245 are about hopeless for 8-layer ground planes but maybe I had too lofty of expectations. The tips also don't wet nearly as well as my trusty Weller.

My favorite right now is the Weller WSL station and WMP pencil.  I use a chisel about 4mm wide, I think as much as I can, this gives GREAT thermal conduction.  I do have some finer tips for high density chip soldering and rework.

I've been also thinking about PCB cleaning. Right now I do it in a very tedious manual way, with a lab hotplate, flux remover, and then an IPA rinse. Takes about forever. After finding out how cheap DI water is I'm planning to run a water line to the lab to then run through a DI water generator/filter system so I can have it on tap when needed for cleaning. No drain in the room but there is a shop sink in the hall.

If you use water soluble flux, you can just rinse the boards off with warm water.  I use no clean for anything that doesn't have high impedance circuitry or high voltages, but the WS stuff really works well.

Jon

[coppice]

In addition to the air filtration I'd like to get a nitrogen bottle to see if nitrogen soldering is really so great - manual soldering, not reflow.

Hmmm, never heard of nitrogen manual soldering!

Soldering in an inert gas (generally nitrogen) is extensively used for both manual and automated soldering. Look at the web sites of Hakko, JBC and probably several other soldering product makers and you'll find compact nitrogen generators and tubing systems to supply the nitrogen to a soldering iron's tip.

[SVFeingold]

ERSA has got one as well, don't think Metcal does which is a shame since after using the JBC I'm really tempted to demo a Metcal unit...I think part of the JBC problem is the compact station, I don't believe it can deliver the whole 140W. Even so, I never see the power meter going above 20-30% even with a monster tip on a through-hole pin connected to 8 ground planes. On the other hand I've seen the Metcals do some very impressive stuff, so even if they are "only" 80W that might be plenty if they can actually deliver the 80W.

Given how cheap nitrogen is and the potential benefits for tip wetting, tip oxidation reduction, general solder wetting, cleaner joints, less flux needed, lower temperatures, etc... I really want to give it a shot.

Some neat videos on the topic here:





The only time I can drag solder like in all of those soldering porn videos is by using a lot of paste flux, EDSYN FL22 is my go-to. But then you have to clean the boards cause they look like shit...otherwise I'm constantly dealing with getting the tip to wet fully and easily.

[mrpackethead]

I've got two Yamaha YV100-ii machines.

You should pick some of these up for your FV feeders http://feederfinger.com/product/yamaha-fv-feeder/ would make your setup time much lower. We wanted to buy these for our company, but they don't make one for our assembleon feeders. 

At $200 each, i belive they are massively overpriced for what they are. the idea is good, but an entire CL feeder is only USD$60. $50   Will wait till some chinese clones appear.

[Kjelt]

Thanks, I didn't know that. I have Argon for my wine preservation.
In contrast with nitrogen it is heavier than air and will stay on the bottom.
Filling a low box once i could do multiple pcbs without special tools.

[mrpackethead]

We use Henkel/Loctite GC10 paste..   I've noticed a difference in the soldering joints between boards that are done in the convection reflow vs boards in the vapour phase.  The boards that are done in the Vapour phase have joints that are visibly more 'shiny', than those done in the convection oven.    Two differences, one is temp.. the max temp in the VP is 230..   Theres no need to go higher..    In the convection its more like 260C.      The other difference is that the boards are in an inert enviroment. The vapour displaces the air..    I wonder if i'm getting an improved joint quality..

thoughts?

NB. for the low volume stuff we do, i'm sold on the vp

[Kjelt]

http://www.indium.com/blog/dispelling-10-myths-about-nitrogen-reflow-part-i.php

[forrestc]

Curious what others do for PCB cleaning of larger batches. PCB washing systems are loony tunes expensive.

In 10+ years of doing this I haven't washed a single production PCB.   For most electronics, a no-clean flux chemistry is perfectly acceptable.  Occasionally a re-worked pcb will get some manual cleaning, but that isn't really washing, more spot-cleaning.

One thing to note:  If you use no-clean flux you generally don't need or want to clean it.   You *can* wash it, but generally the residues are hard to remove because they are designed to turn into something plastic-like which is inert.

On the other hand if you use water soluble flux, you generally *MUST* clean, as the flux will continue 'cleaning' your board after you ship.   Especially with the slightest part of water.

[SVFeingold]

I understand it's not necessary but it's a capability I'd like to have. A lot of the boards I've done (almost all of them) have some parts that require manual work or at the very least adding additional flux at some point. Depending on the rework I might use rosin flux to make life easier. It's nicer from my POV to use the best tools/materials I have available and then just clean the board. And while I haven't done any boards that MUST be cleaned of even no-clean flux, I'd like to have the option available if anyone does have such a board.

That said I see no reason to spend a ton of money on a cleaning system at this point when flux remover, DI water, and a vapor degreaser will probably do everything I need for a while.

I've had a hard time removing no-clean flux before, using MG Chemicals 413B, but I recently got a 5 gallon can of Techspray E-line cleaner and it works phenomenally well on hard (reflowed) no-clean flux. Like...5 minutes in a heated bath and it's gone.

My more pressing concern for the immediate future is the floor. It's concrete but has a few rough spots, I'd like to even it out with a floor grinder, etch, and coat with urethane, on top of which I'll try the Statguard ESD coating mentioned above. It's as much for vanity as it is for functionality, I'll admit, I like clean and sealed floors.

Anyone have a recommendation for sealing brick?

[jmelson]

We have some boards that are run with 400 V across some capacitors, and leakage causes noise.  So, we clean them with an ultrasonic cleaner filled with ethyl alcohol.  sometimes it takes 3 cycles to get them clean enough.  But, if you rinse off with a little spritz of alcohol after coming out of the cleaner, it works pretty well.

Most of the digital and ordinary analog stuff is not cleaned, just left with the no-clean flux in place.

Jon

[jmelson]

Anyone have a recommendation for sealing brick?

Hmm, I use clear masonry sealer to seal the brick on our fireplace.  otherwise, it leaks into the house.  You can put it on with a roller, brush or spray painter.  It will be pretty smelly for a day or so.
Indoors, it should last forever.

Jon

[SeanB]

Concrete floor just go out to the industrial cleaning suppliers and buy a few 20l drums of liquid floor polish and apply with a mop. Seals the floor quite well.

Memories of basic training, and having the tent floor, brown sand, and having to use that polish to make that sand floor shiny. Only plus was we had liquid polish, some others only had paste wax, to do the same. Still better than the Zulu method, which involves a complex bit of work with fresh cow droppings, fresh blood and milk, and a lot of polishing. Plus it stank less.

[SVFeingold]

Hey everyone,

Figured y'all deserve an update after giving me so much good advice!

Short version is I'm slammed with work from every direction. We're still getting the financing and an LLC set up. Sadly our seller fell through which is a real pain as they had what I now consider the perfect equipment for a small shop. All those feeders... Granted they did still offer some feeders but the gist is I have to find new PnP machines and Assembleon MG-8s are hard to come by at the price we had agreed on.

As for the space itself we're neck-deep in preparation work. I've got another pallet of concrete patching material coming this week, and over this weekend we're using a concrete planer to grind down the high spots and clean the surface. We've already pressure washed the brick walls which was a right pain as we had to run the gas generator outside and knock out a window to do so. The concrete planer is also gas powered but it turns out we don't die after running it indoors for a couple hours at a stretch so hey, onwards!

After the concrete planing (which thankfully reveals more weak spots in the concrete) we'll edge out the weak areas with a diamond cutting blade and patch them. Once THAT is done, we're renting an electric scissor lift to install a drop ceiling. Really wanted to keep the old industrial look but there is too much dust from the ceiling and I refuse to take another week to clean and seal it. I'm looking at these clear drop ceiling tiles which will at least keep it from looking like a boring office building: https://www.ceilume.com/panels/Polyline/Clear/

Before the ceiling is installed I need to patch the floor and seal the bricks. Once that is done and the ceiling is in, we'll then get a concrete grinder to smooth the floor. After that we can FINALLY put down the epoxy coat. Somewhere in there the double doors will be installed and we'll pull the trigger on the 200 amp subpanel that needs to be installed by the building management.

Honestly had I known the amount of work the floor would require I might have just hired a contractor to do it. We're going to be about $5-8k into this space by the time all is said and done. All for a freaking 700 square foot space. I think I deserve at least a month of free rent for all this trouble but we will see.

And once all of THAT is done, I can finally FINALLY start moving my electronics lab there and be free to lock down the PCBA equipment and arrange shipping. And of course catch up on all of the work I haven't been able to do.

We're getting there, very slowly, but I am waiting for the day when I can freaking relax and have a seat and a stiff drink in my shiny new lab. Hopefully it will all be worth it! I'm considering building a mezzanine as well to get more usable space, which shouldn't be hard with the 16 foot ceilings. On the other hand I don't know shit about building mezzanines...

Anyway here are some pictures of the progress. Imagine this with a glossy white floor and clean as a whistle...

https://imgur.com/a/zkGdC

I've definitely felt the urge to say "fuck it" and drop the whole idea but I can't let myself do that. Spent 4 hours today driving around to find helicoil inserts to fix the damn concrete planer rental and as soon as I got back I rolled my ankle again in the exact same way as I did last year. So I've been grinding the floors very angrily. But I trust it will all be worth it in the end.

I'll update as things keep happening!

[Mr. Scram]

Thanks for the update! I love the insight into a process normally very hidden.

I hate to ask, but that cut floor looks awfully thin. Is the floor you put in up to the job?

[SVFeingold]

It is rather thin. I think it's thick enough to take the weight of some mostly static machines which in the grand scheme are rather lightweight. I sure hope it is because I'm not going to pay (nor would I be allowed) to get it torn up and re-poured.

The stuff that looks like dirt in the hole is actually concrete, it's just not particularly hard. We have rather soft concrete in the south. God knows how old it is...this building was built in the 1880s! There are a few weak spots like that which the planer is revealing. You can clearly hear when you get to a "hollow" spot which is no longer bonded to the underlying concrete. The thickness of the top layer varies quite a bit, from 1/2" to 4" depending on where you are in the room. Those weak/hollow areas I'll cut out with a diamond saw and patch.
 
Unfortunately building management nixed the blue floors which I was quite looking forward to. Instead I'm going to go for a high-build white epoxy to liven things up.

[Kjelt]

That is a nice old building with lots of charm, needs a lot of work unfortunately too.
Thanks for sharing. 

[SVFeingold]

Hey everyone, back from the dead!

As it turns out our 3 month trajectory didn't pan out so easily.

Financing for startups is - as always - about 10x more time consuming than anyone expects. But...they're getting closer! I've been too busy with other aspects to make much tangible progress on the space, like scrambling to get PCBs assembled manually as needed, getting real tired of that!

In any case I'll have some more time this month to continue working on the space, getting it ready for the new shop. We've settled on a "stopgap" PCBA line until the full financing is available. Similar to what I planned before. This time with a big fancy PnP and then me on either end manually pasting stencils and running my Controleo3 oven for all it's worth! Quite an odd arrangement but still more than adequate to do 100 boards per day.

The stopgap budget is in the ballpark of $40-$50k, and a big chunk of that is facilities. I have a great lead on an Opal Xii + 145 feeders for $30k. Trying to jump on that ASAP.

I briefly considered going with the Neoden4 but ran into a couple issues. The first being that damn height restriction, the second being the limited feeder capacity. To do the upcoming boards in one pass would require TWO Neoden4 machines, and at ~8k/ea + $3k for additional feeders we're just about at $20k. I'd much MUCH rather push a little to get a truly solid PnP - even if I have to subsidize out of pocket.

On the plus side the shop will be much more spacious until we eventually get a stencil printer, reflow, and 2nd PnP. I spent much of today fantasizing about a Mycronic paste jetting system instead of doing real work. 

The shop is also being expanded to 1000 sq ft. in the next month or two - meaning I get to do all that floor prep work all over again on the new side, PLUS tear down an interior wall.

In other news, I just switched to Loctite GC10 paste and all I can say is WOW! I manually pasted and assembled 75 boards - almost 8,000 parts - and had ONE single short I had to clean up on a 0.5mm pitch QSON package. I stencil by hand and it was a breeze, pretty much every board printed and reflowed perfectly. A couple boards I printed poorly with the paste shifted over half a TSSOP pad-width across the whole board. Feeling lazy I said screw it, I'll place it anyway...turned out perfect! Even boards I pasted but didn't have time to actually place for 2 days came out perfect. Unlike the Kester and Chipquik paste I've used it was still tacky enough to make placement easy, after TWO DAYS!

I'm beyond sold on GC10, it's my hands-down recommendation for anyone on the fence out there. Between that and a new roll of Kester KL100D solder, it feels just like using leaded solder.

In the absence of more exciting news I'll also give Controleo3 a plug as I just made the upgrade. Love it, custom programming is pretty simple and I was able to make a profile that matches the GC10 specs almost exactly.

Sorry I don't have more exciting news. Things should finally start moving in the next month!

[mairo]

I also dream quite a lot recently on the Mycronic paste jetting systems. I have been doing some research to see how can one build something "similar" (bench top machine?) by using existing (or build yourself) XY gantry robots and integrating a jetting valve and controller from companies such as Nordson, Vermes & Musashi. I think the Mycronic system uses a Nordson valve..

I am not looking to achieve the speeds of the Mycronic system (I think theirs high end machine can do 1.5 million dots/h?), but something in the range of 10k-20k-50k would be great. Essemtec has a system that is capable of doing 100k+ dots/h (not sure on the exact numbers) and Martin (Dave (the boss here) has a video on one of theirs older paste dispensers at Altium) has now a desktop version of theirs paste dispenser, but not sure on theirs speed (may be they can be used as a rough figure of how many one can achieve on a desktop version).

I understand that for a bench top version the speed would be greatly reduced as the system will not be as stable. The jetting valves + controllers seems to be quite expensive ($5k-10k?) and also the valves have a life, but have not been able to find how often one would need to replace them.

Sorry for side tracking from the main topic of the post.

[mikeselectricstuff]

I also dream quite a lot recently on the Mycronic paste jetting systems. I have been doing some research to see how can one build something "similar" (bench top machine?) by using existing (or build yourself) XY gantry robots and integrating a jetting valve and controller from companies such as Nordson, Vermes & Musashi. I think the Mycronic system uses a Nordson valve..

I am not looking to achieve the speeds of the Mycronic system (I think theirs high end machine can do 1.5 million dots/h?), but something in the range of 10k-20k-50k would be great. Essemtec has a system that is capable of doing 100k+ dots/h (not sure on the exact numbers) and Martin (Dave (the boss here) has a video on one of theirs older paste dispensers at Altium) has now a desktop version of theirs paste dispenser, but not sure on theirs speed (may be they can be used as a rough figure of how many one can achieve on a desktop version).

I understand that for a bench top version the speed would be greatly reduced as the system will not be as stable. The jetting valves + controllers seems to be quite expensive ($5k-10k?) and also the valves have a life, but have not been able to find how often one would need to replace them.

Sorry for side tracking from the main topic of the post.

With stencils available so cheaply it's hard to see a good case for paste jetting outside of exotic builds that would need stepped stencils. You need to wait for the PCB to arrive so the leadime on stencils isn't really a problem

[Kjelt]

With stencils available so cheaply it's hard to see a good case for paste jetting outside of exotic builds that would need stepped stencils. You need to wait for the PCB to arrive so the leadime on stencils isn't really a problem

I agree , the only real benefit for that machine is that it also can simultaneously dose thermal glue dots for two sided smd mounting which with the ongoing miniaturization and pcb costs is becoming more and more standard.

[mikeselectricstuff]

But glue dots are much less demanding than paste - a syringe dispenser will do this just fine

[Kjelt]

True, stil a lot of work to do manually.

[SVFeingold]

I can see the value in a fast paste dispenser particularly for a low/mid volume or prototyping shop, or just to add flexible prototyping capability to a high volume shop. Seriously, look at this thing run:



And 2 simpler boards:



Sure a stencil printer could do that cycle in <10 seconds, which will make a difference when you're doing larger runs. For smaller runs though this would be a godsend. If you find you need to change solder paste shape/volume on any particular pad or type of pad, you can do it in 5 minutes instead of having to re-order a stencil.

You could also vary paste volume per component type, or fine tune pin-in-paste components. It becomes downright enjoyable and fast, no worrying that you didn't get your stencil juuuust right. And if you need to run a mix of say, 5 boards, you don't need to replace/clean stencils, deal with solder paste, squeegees, storage, solder paste waste ($$$), etc. It just loads the program for each board, uninterrupted. For those who need it I imagine you could also switch between lead/lead-free with ease if the paste is available.

Some of the guys on SMTNet who own one seem to agree, it's an enormous advantage. One guy said that the Mycronic will keep up with something like a 35k CPH PnP. Not Foxconn speeds but pretty damn quick I think. Now, a company making 100k runs...yeah, they won't want this.

Downside is that last I heard it's a ~$100-$150k machine. Damn impressive though. A desktop version would be cool but far as I know the primary advantage of the piezo is speed. For more mundane dispensing you could find a regular dispenser robot for way less. IMHO the absolute biggest differentiator will be the software quality.

[SVFeingold]

On topic, I just found an Emerald Xii with an LCS tray feeder that the seller is willing to let go for $30k. No feeders. Now I'm at a total loss deciding between this and the Opal. The Opal has more feeder slots (100 vs. 84) although the Emerald supports twin-tape feeders AND has the LCS which holds 40 trays without sacrificing feeder slots. If I could set it up to use JEDEC-sized cut-strip trays (I assume such a thing exists) then it could hold quite a few BOM items indeed. Although I believe (not sure) that the LCS feeder has a little conveyor that goes into the machine and actually picks the parts internally, which would certainly limit the speed and size of parts if that's the case.

What would you guys do? I guess the only downside to the Emerald is the speed and lack of feeders, but I may be able to purchase the 145 feeders from the original seller separately. Series datasheet is attached!

EDIT: Think I found my answer:

[jmelson]

On topic, I just found an Emerald Xii with an LCS tray feeder that the seller is willing to let go for $30k. No feeders. Now I'm at a total loss deciding between this and the Opal.

Good to see you back, I wondered what happened.

Hmm, a set of feeders will cost a BUNDLE, so a machine WITH feeders at the same price sounds like a good deal.
Otherwise, check the age of the machines.  How many nozzles does the Opal have?  Looks like maybe 8?
Can it handle large chips?

Some of these machines are far more flexible than other models.  You want one that can go from small passives up to big FPGA, all in one machine.

As for the LCS, I doubt you really need that.  I made up a little platform that holds a single waffle tray.  If I have TWO larger chips, I just repack them onto one waffle tray (so far, the only board that needs that has two chips of the same leadframe).  I have a hose with a suction cup for ICs on the end.  I just stick it in my mouth and move the chips quite fast.
So, unless you have a lot of different larger chips, the LCS is just another complex gadget to program and keep working.

Jon