At what point is vision required on a stencil printer

[Mangozac]

We are planning to install a new line and my intention was to go for a standard Chinese semi-automatic style printer like this: https://kayosmt.eu/kayo-5088

These printers are affordable, relatively simple and seem to be well regarded for volumes like ours. Our requirements are fairly simple: up to 20k boards per year and very rarely anything below 0.5mm pitch. Only large pitch (1mm) BGAs if they can't be avoided.

At what point is a camera alignment system (whether manual or automatic) actually necessary? The only machines with vision in our budget are used machines and I feel that there's too high a gamble with them due to their complexity and cost to repair in the event of a failure. A semi-automatic machine without vision is a fairly simple piece of equipment, not relying on ancient copies of windows on obsolete 386 hardware.

[SMTech]

The reality of a semi-automatic printer like that with or without vision is that they are reliant on every single PCB aligning up with the stencil perfectly for each print based on how the fixture held the first PCB. The reality is that isn't always something you can rely on. Hold the board by the edges and the PCBs can vary just a little, align with holes and they need to locate in such a way that the pins don't come thru the board and are enough to hold the PCB firmly, thin boards or ones with no holes then become tricky. Some semi-auto printers with cameras let you check alignment for every print e.g a Miniko, TWS SR2700 or PBTs offerings which helps. Camera alignment on a semi-auto doesn't in itself make it more accurate, its still you twiddling dials and looking through apertures to the board beneath (unless you have a mylar screen type) that still has its very obvious weaknesses. What the camera certainly does help with is getting a view exactly perpendicular without leaning over the board in a fashion that may not be comfortable or possible.

However they are still much slower to setup than an automatic printer, it all looks very slick in the video but that's a lot of bending down and reaching under and how often does a stack of panels just drop into a fixture perfectly secure just like that? I also really have no idea how anyone can actually sell that design in Europe with all those moving parts your actually supposed to put your hands under. 

However alignment for every board is only half of what an automatic printer offers, they also make up part of an automated line, they (can) inspect the print for solder deposition consistency, they typically take 29" frame as standard so print quite a big board (much harder with manual alignment) they can store the locations for support pins and then laser designate them or even place them for you, options like RED-E-SET/Gridlok actually exist for them (the semi-auto versions are no longer). They are quicker to setup/changeover. Stencil underwipe/cleaning is basically standard.

However to your question "at what point?", well at the point you can identify not having one is causing problems on the boards you manufacture. I probably build similar volume, we still do it with manual/semi-auto printers without vision. Like you we avoid BGA, 0.5mm in opinion is right at the limit of manual aperture alignment, we have a few 0.4mm pitch ICs and they are not fun, just about doable, but not fun. We're at the point we can say we do need an automated printer so I have done some of the legwork on the options, current circumstances however aren't exactly conducive to investment are they . I don't think there are any really bad choices but there are some that are clearly not as good as others.

Desktop vision semi-auto NEW : ~10-15k
2nd User DEK: ~ 10K-45K. Really depends on the model, features and age but built like tanks and in the UK at least, really not an issue to get serviced even on contract.
Standalone vision  semi-auto NEW (PBT FA23/Reprint R23): ~ 20-25k there is also an auto alignment version of the FA23 which makes it a unique hybrid.
Chinese Auto NEW: ~25-35k depending on brand and where you are buying it (note they all have the exact same weird weakness, accept a 29" stencil but only print a board you would expect to be able to print in a 23" printer)
ESE: 35K+ these printers are pretty affordable and I've seen lots of praise too (https://www.elite7group.com/ has a price right on the site)
DEK NEW: Never had a formal quote but it has been suggested a deal can be done at ~40k for an intro model, I'm assuming that's a Chinese built E-series.
Speedprint: 50-60K start. Seen a few meh comments but Europlacer/Speedprint are very nice people to deal with, even according to those who say meh about the printer.
MPM: 60-80K
SJ Innotech: 80k+ , they make no bones about being a high end large format specialist.
Mycronic MY700 Jet printer: Somewhere north of 100k, 70-80k might get you an official factory refurb. These have interesting economics, some swear by them others don't think it adds up.

[Reckless]

After using every method: credit card, manual, semi auto, fully automatic I recommend buying a used MPM/DEK.  If you buy the right model not expensive to work on.

I'm selling an extra DEK horizon for $8500 (windows xp) and an older mpm up2000 (DOS) for $5000.  Personally, I like the MPM but one of my operators is used to DEK Horizon after using it 5 years at his old job with zero issues.  If I thought semi automatic was worth it I'd recommend it. I had ekra E1 which is best semi imo and the mpm/dek are much nicer/easier to use.  20k/year boards is automatic level, less than 3k/yr pcs is semi.  Even the Ekra had more points of failure than the MPMs and the DEKs.  Ive had 5 mpms and 4 deks (3 manual essemtecs and 1 ekra as well) and I am surprised how well the automatics last.  They are constructed much better than the semi automatic, industrial grade.  My MPMs are 25 years old and run great.

[Mangozac]

Wow, thanks for the extremely comprehensive reply, SMTech! You have confirmed my understanding of how the various options work and as you suggest, the frequency with which the positioning needs adjustment depends heavily on the quality of the PCB manufacturing. Unfortunately that's something that can't really be easily planned for. Can I ask what semi-automatic unit you've been using?

However they are still much slower to setup than an automatic printer, it all looks very slick in the video but that's a lot of bending down and reaching under and how often does a stack of panels just drop into a fixture perfectly secure just like that?

Indeed that's the big unknown to me - how often does a stack of panels drop in like that without any issues? People I've spoken to have indicated that unless going for very fine pitch it's generally not a big deal.

I also really have no idea how anyone can actually sell that design in Europe with all those moving parts your actually supposed to put your hands under.

I actually played with one of these when I was last in Shenzhen and made the same comment. I couldn't tell just how dangerous it would be if your hand got stuck in there but I wasn't game to find out 
In reality though with perhaps an additional couple of e-stops I see no issue with a trained operator using it.

After using every method: credit card, manual, semi auto, fully automatic I recommend buying a used MPM/DEK.  If you buy the right model not expensive to work on.

Reckless you always have a sales pitch. I don't think there's any piece of equipment you haven't had for sale 

The MPM 2000 machines seem to be available for very low prices - apparently nobody wants them because they're too afraid if something goes wrong. Have you had to repair one? They seem to be very common machines and I can get a 20 year old one here for the price of one of those semi-automatic Chinese units. It looks like MPM still sell spare parts for them but those boards look veeeery expensive!

Edit: I just looked up some videos of the UP2000 and holy crap that is a complex piece of equipment, on par with the pick and place machine! I feel like it would be a nightmare to service or repair so it's no wonder nobody wants them once they're old...

[Reckless]

Like you, I'm learning from experience.  I usually get lots and sell off extras or units we have upgraded.  TBH, I'm not necessarily trying to sell it to you... Just putting it out there.  For us I got 4 of them currently running fine, sold 1 a few weeks ago.  One of mine needs a new fiber optic cable as it gives intermittent problems which we have located the part but waiting for covid to end before installing it.  Personally the MPMs and DEKs are easier to work on since they dont jam everything supertight like asian machines.  To my knowledge MPM doesn't sell parts for them but they are pretty easy to come by.  Its important to buy a model that sold a lot of units in the states to find spares cheap (better if is tier 1 company).  Just like its easy to find spares for universal gsms, the mpm up2000/2500 line was a big thing years ago, bigger than DEK.  They sold a lot of them for a long period allowing them to perfect any bugs/issues.  Plus being a pioneer and top ranked supplier they had to make quality industrial last forever machines.  The older ones are even better, no junky plastic covers.  Everything is super solid and stable. My gut tells me these older machines will outlast newer machines even though newer machines have canbus system (makes them much easier to work on but more expensive). 

My experience is I like super simple/easy machines.  The more complex the bigger the pita especially used (i hate training new operators).  Dos based mpms are my kind of machines especially the HIE models.  Simple, no nonsense, easy to program.  Windows adds unnecessary complexity imho.  Same reason I like Essemtec no PC, same reason I like Fuji CPs no windows (using unix).  Making a manufacturing line with every component having windows esp NT/2000 is a recipe for disaster especially if you reboot daily.  Wherever windows is you have crashing, bsod, virus, trojans, ram issues, problems. 

I believe pricing for these machines new are much higher than what SMTech is thinking.  I heard older DEKs were 6 figures reaching quarter mil optioned up. Im usually intimidated to go to their booths at SMT shows because they are trying their best to sell me new ones and I'm trying the opposite how to make old ones last forever, save my money and learn the history of old machines.  Ive heard the Fuji reps tell me they think Fuji messed up by making their older machines too good, no one wants to upgrade them.  Sure the new models are way slimmer and sexier but these older machines were built in an era where things were over engineered and built to last.  As an old school computer engineer I like those die hard models that never quit and dont give unnecesary headaches.  I'd be more scared of any chinese machine over one of these.  I tried for years to find a chinese automatic screen printer for under $5k.  Nothing looked promising but I havent been to shenzhen for 3-4 years.

[Mangozac]

My experience is I like super simple/easy machines.  The more complex the bigger the pita especially used (i hate training new operators). 

From what I've seen though, "simple" is not the adjective I would use to describe an automatic inline screen printer!

These guys seem to be selling spare parts for MPM machines and have a full catalogue, but like I said those are not affordable looking boards: http://www.goppm.com/mpm-up-2000.php

[SMTech]

I believe pricing for these machines new are much higher than what SMTech is thinking.  I heard older DEKs were 6 figures reaching quarter mil optioned up.

I have quotes <12 months old on all of them, (and a few others) except officially from DEK . On DEK I have had one seller of other printers who knew those were way outside any budget suggest we might have that ballpark range price made them his suggestion based on what he'd heard other customers of his get quoted and a few years back I've had another seller say "I can get you a Reprint R23 for £17k or I can get you a DEK and it would only be about £35k", I have no idea what model or spec that would have been (it was pre - E-series) but it does suggest DEK will do a deal to get a sale (not uncommon with new equipment, rarer on the Asian kit). These were UK prices on fairly entry level models but with the basics you would want, some SPI, stencil underwipe, 29", no emphasis on superfast cycle times or SPI as good as a standalone unit or labels or glue/paste dispensing or dual lane or traceability options that you can add to ramp up the price.

However yes both DEK and MPM sell high end models than can cost way more and if you get a 2nd user machine out of an automotive EMS, there's a good chance that's what they will have had and they can be £45k @ 10years old. I've been offered ex-demo MPMs for more than a new one because the features on-board are so expensive. As I understand the other printers that have quite high initial prices are the P&P own brand ones from Panasonic/Fuji/Yamaha, however these days you buy those so that your whole line can easily self correct issues as every system feeds back from the others.

I'd be reasonably happy buying a 2nd user popular MPM/DEK but I'd be even happier buying it from or with help from someone who services them. In the UK that could be https://www.adamtech.co.uk/ perhaps, MPM slightly more complicated due the number of brands and smaller companies that were manufacturing some models if you go back 20 years. However I think its the optimum route to go down and one worth researching for your region.

[Reckless]

I agree.  Not sure about the entry models but to me its very ridiculous pricing.  It doesnt make sense to buy new unless you have 8 figures worth of business and I dont know if you get much more for your money.  The way I see it better to buy more used machines than 1 new.  If machine is in good condition these things last years with almost no repair costs.  Depends how beaten they are and if you are not careful with electricity (have blown a couple amps with accidental shorts installing other equipment, but bought replacement amps for $45 off ebay). 

I dont see a need to have someone service them, unless you are crossing a few hundred panels per day (even then these machines will take a beating before giving an issue).  This is why I recommend them.  Almost a no hassle machine especially if pampered.  I also thought they looked overly complex but once you use them you get used to it.  I got lucky and bought some equipment from a prototype lab which baeely ran it.  I love it when I buy like new used equipment.

I am surprised no one has found a cheaper way to get this done.  It looks like a very simple machine to build.  My cnc style solder robot cost $750, this isnt that much different, similar gantry system with a different head. 

[Reckless]

Even though I love the MPM UP2000.  Its possible it might not be best suited in your application.  I have a hard time imagining your workshop.  It almost sounds like a garage/basement/tiny shop if you are running a chinese pnp.  I was searching something else and I came across this.  Size wise it might be better for your application even though hard for me to imagine UP2000 not being best solution. 


https://www.dimadispensing.eu/dispensing/products/jet-dispensing/dispense-master-jet-dispensing
I think chinese make a variation of this as well that might be cheaper AND BETTER.  I have not heard good things about jetting in general.  I think its very messy and requires a lot of maintenance and cleanup (things I dont like).  I am even thinking about getting one of these to save on stencils for prototype units.  Although im hesitant as programming time might make stencils cheaper.  Ive been very curious about jetting technology in general. 

[Mangozac]

Jet paste dispensing would be cool but our volume would unfortunately be too high.

[SMTech]

Only Essemtec & Mycronic have paste jetting solutions good enough to replace stencil printing. Time/Pressure dispensing just isn't accurate enough, good for glue or topping up paste quantities for large devices, maybe printing the odd missing not too fine footprint between stencil revisions. 

[robint91]

@SMTech, do you think that jetting is good enough for 0.35mm pitch BGA (Thank you NORDIC for creating such parts), I can do that manually within a few tries, and inspecting the paste under a microscope. But eventually I want something more consistent.

[SMTech]

I would direct that question right at Essemtec maybe with some example footprints and then tell us what they tell you. Anecdotally I'm told Essemtec were getting some stick from clients/re-sellers as the jetting option was limited to type 3 paste. I was told that they were aiming to fix that and make HyQ feeders cheaper (or make a cheaper version of them) last year, possibly they announced it @ Munich? I would  presume jetting was more sensitive to particle size than printing. For what you do I would think you could probably make a good case for retrofitting it to your machine if its specs are good enough.

[fcb]

@SMTech, do you think that jetting is good enough for 0.35mm pitch BGA (Thank you NORDIC for creating such parts), I can do that manually within a few tries, and inspecting the paste under a microscope. But eventually I want something more consistent.

Consistent 0.35 BGA is perfectly possible with semi-automatic and automatic machines. The quality of the stencil (angle of sides, polish etc..), stencil mounting frame system and exact paste type (at least type 4 or 5) as important as X/Y/Theta alignment.

Never been a fan of jetting paste.

My experience with 2nd user DEK's has been playing whack-a-mole with the air-lines and fittings, there are also a bunch of chinese made DEK's out there..  Blakell's EP7xx machine is a good mix of performance, price and CUSTOMER SUPPORT if you're buying new.