"So i want to have my own SMT assembly line"

[mrpackethead]

I don't know how many posts we see " i want to have my own SMT assembly line ".   Do you think For every 50 or so threads someone actually does one?

I thought i'd pose some questions to start a discussion.    Some of us have some experience.

(1)  What is the motivation for having the capablity.
(2)  What do you want do ( technically )
(3)  How much work do you want to do.
(4)  What expereince do you have.

[jmelson]

I don't know how many posts we see " i want to have my own SMT assembly line ".   Do you think For every 50 or so threads someone actually does one?

I thought i'd pose some questions to start a discussion.    Some of us have some experience.

(1)  What is the motivation for having the capablity.

The main reason I do this is because I do very small runs.  Anywhere from 20 to 60 boards of one type.  That makes a contract assembler very expensive.  Of course, if I had not gotten a REAL DEAL on a used machine, this would have never made sense.  And, of course, the reason to do such small runs is so as not to have thousands of $ of stuffed boards sitting around for a long time.

Jon

[boB]

Nothing wrong with wanting to populate your own SMT boards.  We started our production using re-furbed machines.  The original cost was
still thousands of dollars and part of that was having someone to come out and calibrate the machines and show us how to use them.
Then there is the cost of the parts feeders. We found even more machines and even a couple of free machines.  Then a broken
machine for cheap but with feeders for cheap as well.

We are now running full time with 10+ QUAD machines or so with a couple extra for parts.  What I wanted to get at was that
one of these extra machines are slated for engineering.  This is for prototypes too but mainly so that the PCB layout guys can make sure
that their PCBs  are layed out and files output so that production can work with them easier and find the issues earlier.  Great learning experience.
 
But be prepared to spend lots of time and money away from actual circuit design while getting things running.  A lot more money spent gets you
a new machine with a warranty and maybe some training.

boB

[mrpackethead]

I don't know how many posts we see " i want to have my own SMT assembly line ".   Do you think For every 50 or so threads someone actually does one?

I thought i'd pose some questions to start a discussion.    Some of us have some experience.

(1)  What is the motivation for having the capablity.

The main reason I do this is because I do very small runs.  Anywhere from 20 to 60 boards of one type.  That makes a contract assembler very expensive.  Of course, if I had not gotten a REAL DEAL on a used machine, this would have never made sense.  And, of course, the reason to do such small runs is so as not to have thousands of $ of stuffed boards sitting around for a long time.

Jon

This is 80% of my story too.

[mikeselectricstuff]

Me too. At the time it was a bit  of a "new toy" and I didn't think it would pay off for the forseeable futire,  but I was so wrong. The time saved documenting things for subcontractors alone is a big saving, not to mention their setup costs and the cost of the big stencils they insist on using.
I think I paid about £5K, though I subsequently that was over that was over the odds for that (Versatronics Rv4s) machine - they typically go for £1-3K on ebay with feeders.

IMO the availability of sub - $10K solutions, either old used or new Chinese machines makes in-house a much more viable option  for many low-medium volume users. 
Yes, it does take a lot of work to get things up & running , but once running the benefits can be huge.

[ubbut]

I do small production runs of about ~100 boards. In the last years I have done this via various assembly houses in Germany, but always ran into issues. Main problem was the long wait time - sometimes 8 weeks.
So I thought I'd give one of the cheaper Chinese machines a try. Did not expect too much and was very aware this may just be something for prototypes, or pre-series runs. Took me about 1 week to get the machine running and already assembled 2 products, 100 boards each. Pleasantly surprised so far, let's see how this pans out..
Of course I don't have a wave soldering machine, so THT is a bit more complicated. Some connectors however can be soldered 'pin-in-paste', so that's less of a problem. Also there is this drag soldering tht cartridge: https://www.jbctools.com/c245669-cartridge-drag-soldering-125-s2-l-product-916-category-4-menu-4.html

[coppice]

I don't know how many posts we see " i want to have my own SMT assembly line ".   Do you think For every 50 or so threads someone actually does one?

I thought i'd pose some questions to start a discussion.    Some of us have some experience.

(1)  What is the motivation for having the capablity.

The main reason I do this is because I do very small runs.  Anywhere from 20 to 60 boards of one type.  That makes a contract assembler very expensive.  Of course, if I had not gotten a REAL DEAL on a used machine, this would have never made sense.  And, of course, the reason to do such small runs is so as not to have thousands of $ of stuffed boards sitting around for a long time.

Jon

Small runs can be expensive when you are far from the madding crowd. Everything from gathering the parts to getting them assembled is a huge drag on your time and finances. If you are somewhere like Shenzhen, you are so close to many people who can supply parts from stock, and who can do various kinds of small volume manufacturing, that your ways of working can be completely transformed. This is the reason so much electronics work gathers around Shenzhen. Many people who try to set up shop in the cheaper parts of China end up moving back to Shenzhen. China has a really excellent freight train network, which can reliably take something you have finished at the end of the day in Shenzhen, and have it working on a production line a 1000km away the next morning. Still manufacturers find logistics issues hamper them, and they move everything back to the Shenzhen area.

[jmelson]

Me too. At the time it was a bit  of a "new toy" and I didn't think it would pay off for the forseeable futire,  but I was so wrong.

OH, MAN!  I remember with HORROR the days of building boards by hand!  ARRghhh, I NEVER want to do that again!  I still do the first article of an all-new or major redesign of a board by hand, to make sure the board doesn't have some totally catastrophic error, like using the wrong FPGA pinout, that saves me making a set of solder stencils for something that won't work.  But, I've built some thousands of boards on my P&P, and can't imagine ever having to go back to manual.

 The time saved documenting things for subcontractors alone is a big saving, not to mention their setup costs and the cost of the big stencils they insist on using.

I have never used an assembly house, but I have recently seen some prices that look less HORRIBLE than they were a decade ago, but still the setup charge/job would be pretty high with my volume.

Jon

[rx8pilot]

I have posted numerous times  - many of those may sound like doom and gloom but in the end....I am VERY VERY happy I have a P&P line.

To qualify that a bit better, I am only happy because it works and places the vast majority of parts that I need to place. It sits turned off most of the time. When it is needed - it is generally an 'emergency' and there is literally no off-site assembly option at any price that could deal with my needs. For example - a new set of PCB's arrives on a Friday and I need working PCB's by Monday. Perhaps 2-3 different designs, double-sided, 5 pcs of each. 100 unique parts, 750 total parts, some parts are still being calculated as the assembly process is happening.

I don't have to document anything, talk to anyone, wonder if the diodes are reversed, wonder if 1% passives are being placed with 5% parts, etc. The direct to assembly option in-house makes up for the effort that goes into setting up and dealing with the machine.

My needs:
Small volume, short lead times, deal with design un-knowns (last second changes).

[mrpackethead]

To qualify that a bit better, I am only happy because it works and places the vast majority of parts that I need to place. It sits turned off most of the time.

My line gets used about 1.5 days a week on average.    Still well worth having.

When it is needed - it is generally an 'emergency'

I wont' say emergency, but theres always time pressure.   Someone wants somethign as quickly as possible.   I dont' want to have to keep truck loads of finished stock on the shelf.    And often the customers requirement is customised anyway, so it has to be manfuactured as its order.

I don't have to document anything, talk to anyone, wonder if the diodes are reversed, wonder if 1% passives are being placed with 5% parts, etc. The direct to assembly option in-house makes up for the effort that goes into setting up and dealing with the machine.

the talk to somone part.  Transfering all the nessary information is what takes huge amoutn of time, and huge amount of effort, and is where mistakes happen.
[/quote]


The cost for me, was a huge investment in having an end to end system that lets me design for manufacture the moment i start drawing a schematic.  ( for my line )

[D3f1ant]

Merge previous few posts and that's why we have our own machines too. In house proper CNC is next. We already have one of those little CNC routers, but we farm out a lot of custom enclosure work that could be done in house The CNC router gets used as often the PNP machine, boards always need a box with cutouts.

[rx8pilot]

Proper CNC is a rather serious learning curve, but I can't imagine my operation without.

Low cost machines are harder to use than a true commercial option - Haas and higher. All of them have a tendency to suck in time like a super massive black hole. Workholding is the big surprise learning challenge that surprises most newcomers. The expenses never really stop since you always need a different tool, a coolant filter, software upgrades, tool boxes, a bigger compressor, an air dryer, a saw, tool holders, vises, fixture clamps, metrology, surface plate, etc, etc, etc....

12 years in.... I am still surprised constantly at the never ending needs.

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

[mrpackethead]

Proper CNC is a rather serious learning curve, but I can't imagine my operation without.

Low cost machines are harder to use than a true commercial option - Haas and higher. All of them have a tendency to suck in time like a super massive black hole. Workholding is the big surprise learning challenge that surprises most newcomers. The expenses never really stop since you always need a different tool, a coolant filter, software upgrades, tool boxes, a bigger compressor, an air dryer, a saw, tool holders, vises, fixture clamps, metrology, surface plate, etc, etc, etc....

12 years in.... I am still surprised constantly at the never ending needs.

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

Like proper PNP.  You're always needing new feeders, new belts, nozzles.   Etc etc.    And its a rather serious learning curve to do it well, to get 1 in 10,000 placments error rate.            I bet the people who Do their on CNC probalby also do their own PNP.  Its all about attitude.   ( I also cut my own plastic in a laser cutter! )

[Gary.M]

I do small production runs of about ~100 boards. In the last years I have done this via various assembly houses in Germany, but always ran into issues. Main problem was the long wait time - sometimes 8 weeks.
So I thought I'd give one of the cheaper Chinese machines a try. Did not expect too much and was very aware this may just be something for prototypes, or pre-series runs. Took me about 1 week to get the machine running and already assembled 2 products, 100 boards each. Pleasantly surprised so far, let's see how this pans out..
Of course I don't have a wave soldering machine, so THT is a bit more complicated. Some connectors however can be soldered 'pin-in-paste', so that's less of a problem. Also there is this drag soldering tht cartridge: https://www.jbctools.com/c245669-cartridge-drag-soldering-125-s2-l-product-916-category-4-menu-4.html

Which machine did you buy?

[ubbut]

I do small production runs of about ~100 boards. In the last years I have done this via various assembly houses in Germany, but always ran into issues. Main problem was the long wait time - sometimes 8 weeks.
So I thought I'd give one of the cheaper Chinese machines a try. Did not expect too much and was very aware this may just be something for prototypes, or pre-series runs. Took me about 1 week to get the machine running and already assembled 2 products, 100 boards each. Pleasantly surprised so far, let's see how this pans out..
Of course I don't have a wave soldering machine, so THT is a bit more complicated. Some connectors however can be soldered 'pin-in-paste', so that's less of a problem. Also there is this drag soldering tht cartridge: https://www.jbctools.com/c245669-cartridge-drag-soldering-125-s2-l-product-916-category-4-menu-4.html

Which machine did you buy?

TVM802B, but there are several similar machines

[jmelson]

When it is needed - it is generally an 'emergency' and there is literally no off-site assembly option at any price that could deal with my needs. For example - a new set of PCB's arrives on a Friday and I need working PCB's by Monday. Perhaps 2-3 different designs, double-sided, 5 pcs of each.

Well, I really try to AVOID these situations.  I do have ONE board that I make for ONE customer, it is basically an 8-channel version of one of my standard boards.  Well, I don't keep any stock of these completed boards, and they had one go bad in the field.  So, that was a big rush to get them a couple boards whipped up.

But, otherwise, I try to calculate the rate boards get sold, and schedule to make a run so that I don't run out of them.  I really HATE to have big time pressure to cranks stuff out.  That's how you make mistakes that can lead to wasted boards or a lot of rework.

Jon

[jmelson]

My line gets used about 1.5 days a week on average.    Still well worth having.

HAH!  My P&P gets used more like 1.5 days a MONTH, and I STILL think it is WELL worth having!  Of course, not at new prices for a commercial-grade P&P, but for what I paid for it, YES INDEED!

Jon

[pauljmccain]

Just chiming in - I started with a TM802A as well for my business. Same reason, very low volume. It was also much easier than engineering for a contract assembler (though this isn't really a positive, it was just convenient for us not knowing a thing about assembly or contract assembly).

Now, 4 years later, we are running two lines with older Assembleon/Yamaha chipshooters, Speedline screenprinter, and AOI. Staff of two, 8 hrs a day every day. Obviously there's a huge investment now with component inventory, QC, training, machine maintenance, etc... but it's definitely still worth it, and I'm glad we do it vs contracting. Our batch sizes are still 100-200 panels max, running around 200 different products. We can turn things around very quickly, and adjust lead times as we see fit. We have looked at moving some of our higher-volume work to contract, but the best cost we've found is nearly double what we pay, even with utilities, rent, depreciation, etc factored in.

[rx8pilot]

Well, I really try to AVOID these situations.  I do have ONE board that I make for ONE customer, it is basically an 8-channel version of one of my standard boards.  Well, I don't keep any stock of these completed boards, and they had one go bad in the field.  So, that was a big rush to get them a couple boards whipped up.

If I could find a way out of this, I would take it. The nature of the industry and the product create a lot of last-second demand for products that are semi-custom. Not something I like, but something I have to deal with. The parts are too expensive to keep the shelves stocked with much. Whatever I put on the shelf seems to be a different configuration than what gets ordered.

[mrpackethead]

Well, I really try to AVOID these situations.  I do have ONE board that I make for ONE customer, it is basically an 8-channel version of one of my standard boards.  Well, I don't keep any stock of these completed boards, and they had one go bad in the field.  So, that was a big rush to get them a couple boards whipped up.

If I could find a way out of this, I would take it. The nature of the industry and the product create a lot of last-second demand for products that are semi-custom. Not something I like, but something I have to deal with. The parts are too expensive to keep the shelves stocked with much. Whatever I put on the shelf seems to be a different configuration than what gets ordered.

What i've found is that by consoldiating my BOM down to as few parts as possible,  ( in my case i have 172 core parts ), its possible not to have to change / move the PNP line configuration very much at all.. the stock parts can just stay on the machines.      This means that its just a matter of changintg the 'program' on the pnp's,  and putting a differnet stencil on.     Easy now.  Was a big job to get there.

Electronics production is not about PNP, thats somethign that happens down the track. Its about being super efficent in information managment.

[D3f1ant]

That's exactly what we do as well. Because we make a lot of automotive product, pretty much everything we build gets rated parts regardless. It's not worth the time swapping out a feeder save a few cents per board.

After a while it does become more about logistics of finding parts you know you have  somewhere, there is always odd values. We saw this thing first first hand at electronex and it seems like a great idea.

https://www.cluso.com.au/invman

[mrpackethead]

We build our own inventory system from teh ground up, its been key to making thigns work.

We use altium for design.   So all my librarys now are dblib based.    The same database that contains the information about the components, also contains the inventory information.          After i create a BOM in a design, i can quickly run a 'machine-pick' script that will generate a report for that tells me where stuff is.    It seperates the report into sections;   parts that are already on the machine lines,  parts that are in stock and need putting on the machines,  parts that will be needed but hand assembled, and then a list of parts that need ordering.

When i order parts, those orders get entered into the system, the parts get given a 'location' of 'ordered'..    When the parts arrive,  each reel/bag box of parts gets given an artibory unique QR code ( we used UUID's )..  This means we can preprint sheets of Barcodes on the laser printer onto low cost label sheets.   We bring up the list of 'ordered parts', find the part we are looking for.. and then scan the barcode we stuck on it, to register it..      From there, anytime you touch the parts you scan it..     We have stock locations.. ( i'm using small bins that hold about to 10 x 8mm reels ) mostly..   each has a barcode as well.   So you scan the part,   scan the "move" QR code  ( saves keystrokes ) and then scan the location code.        We dont' have to put stock in any particular location. Just an empty space,    It updates the database every time its moved.    If someone pulls a reel out and leaves it on top of the PNP machine.. (meh! ) you can just scan it, and put it away, and its not lost.

When we finish a PNP job we know how many parts where used, and we can then decrement the quantitys etc.

The barcode scanners cost about $35 for wired USB ones.  the wireless ones a bit more.   Bar code lables cost $30 for a box of 100 sheets of 70 per sheet.      Database and sdesign... LOTS of time!   Theres a lot more i'd like to do with it, but need time to do so.

[D3f1ant]

That sounds exactly like what I need. Pity I don't have the time build and develop something myself. Shortcut is to buy something off the shelf like the invman but it's fairly expensive. Complete system is around US$15k apparently. Not sure I could build it for that, unless I rate my time at 0.

[mrpackethead]

I probably coudl make mine public, but its so custom for us, that i'm not sure it would be that useful to anyone else. and its got lots of hacks in it, and things to fix..

[SMTech]

If you want a stock system that integrates that closely with multiple sources it will indeed be expensive and require some custom modules. Even $15k might be a conservative estimate, if you buy a big name package like Sage Manufacturing, the price is IIRC over double that. Just like office however you can pay for many of them as you go, and there are a bunch of open source ones if you have some time to put into setting them up, any package requires a fair bit of effort just to going just to get your data in. The primary acronym is an MES but these tools are manufacturing tuned implementations often offering the features of CRM and MRP so you can find manufacturing modules to add into packages that do that too.

A common package offered by distributors is Aegis Factory Logix , in the UK I have encountered several medium size IT support companies/software that have seemingly developed their own.
Some examples

http://www.manuonline.com/
https://qcadoo.com/en/
https://www.ogl.co.uk/profitplus
https://www.anagramsystems.co.uk/
https://www.xtuple.com/products/postbooks

loads of others, of course some of these can or do go way beyond what many small businesses want, planning lead times, servicing intervals. With open source versions there is always the potential of knocking up a script to integrate a data source of your own against the one the system uses.

[mrpackethead]

Mine was an evolution. It started out when i wanted to solve library sharing between people.  then it seemed an obvious addition to start tracking the stock items.      The wrirting modules to do things like creating PNP files directly for loading..

[D3f1ant]

Rather that corrupt another thread, what vapour phase oven are you using? I'm looking at oven options at the moment, somewhat keen changing to VP. A largish batch machine would be fine, but I see they do inline machines too...very spendy.

[mrpackethead]

Rather that corrupt another thread, what vapour phase oven are you using? I'm looking at oven options at the moment, somewhat keen changing to VP. A largish batch machine would be fine, but I see they do inline machines too...very spendy.

I'm now using somethign i built myself, after getting one from germany.  ( IEDMS ) .  The IEDMS machines are conceptually clever, but absolutely terribly made.    Avoid them like the plauge.   The controller is just crazy, and mine was electriclaly unsafe.  It also started falling apart.   AVOID one of these like the plauge.

i'll take some pics.  But basically its much much simpler than you think,,,  Baiscally just need to heat up some galden.

[Stavos122]

Have you considered a high power batch reflow oven?  Something like this: https://www.atco-us.com/products/pro-1600-smt-reflow-ovens

[rx8pilot]

I have been dreaming of a batch sized vapor phase machine.....like a while before I can afford it though.

It seems like a great option for modern power electronics that mix big and small devices. It seems to be much less stress on the components by bringing them all to the same temperature at the same rate.

[mrpackethead]

I have been dreaming of a batch sized vapor phase machine.....like a while before I can afford it though.

Time i publish my plans perhaps.   I think for about $300 and some time you'll be able to do it.    IT just dpends how much you want to optimize it.

[/quote]
It seems like a great option for modern power electronics that mix big and small devices. It seems to be much less stress on the components by bringing them all to the same temperature at the same rate.
[/quote]

Absolutely, i have a board that has a big Inductor on it, and 0402 and .4mm QFP.   

[rx8pilot]

Time i publish my plans perhaps.   I think for about $300 and some time you'll be able to do it.    IT just dpends how much you want to optimize it.

In general - I was hoping to avoid DIY but the price of professional solutions is very high. Maybe it is a good idea to try.
The best guess is that a modest performing vapor phase system is likely to be better than my convection batch oven.

Have you done any experiments?

[D3f1ant]

I still shopping for a new oven too, considering moving from a fairly large batch oven to a conveyor one. VP seems so awesome but the quotes I have for commercial quality batch machines are eye watering...making me really rethink the shift to VP.

[rx8pilot]

I have not yet figured out what drives the cost on the commercial vapor phase systems. Perhaps I am missing a lot of the details, but they seem remarkably expensive relative to the task.



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

[mrpackethead]

By trading off speed and efficiency the complexity comes down a lot and it can be much much simpler.

The commerical systems are keeping the vapour cloud going all the time, and have a carrier system that takes the board into the vapour and then pulls it back out.  They have preheat and they have to build the system in such a way that you dont' loose the 'soft vapour'  ( the 'thin' cloud above the main cloud ) as you put the board in/out.. So you need complex vapor proof locks etc. etc.     

With the 'heater in a bucket' its super cheap, but you essentially are putting your boards on a static carrier in a cold tank.  You then heat the entire thing up.. which takes quite a bit of time and energy.  Wiht some smarts you can heat it up on a sensible 'preheat',  ( following a curve )..   When it hits 230C, you then let it cool down, back to room temp, and take the boards out.        A cycle typically takes me about 20 minutes.   A commercail system will be more like 6.

I've found that i can stack boards up, several high. ( i've done as many as four panels deep ).. Runn several tanks side by side, and i shoudl be able to get to 50-60 panels per hour, ( panel is typically A4 Size )..

It just saves so much drama with profiling and overheating.. its worth the effort.
 

[coppice]

I have not yet figured out what drives the cost on the commercial vapor phase systems. Perhaps I am missing a lot of the details, but they seem remarkably expensive relative to the task.

Aren't most current commercial offerings of the continuous operation type? Those are quite complex. When I see the occasional ad for a small batch type VP machine it seems like the crude, made in someone's back yard shed, kind of thing that mrpackethead bought and regretted. The more professional makers seem to have given up on small VP machines for prototypers, which is ironic as VP is where the big growth seems to be in overall board production.

[Kjelt]

IIRC the commercial VP machines have at least two chambers.
One is where the Galden is heated and stored (lower part) and the other is where the boards are placed and removed.
So you place the boards, the top lid of the top chamber is locked, the under lid is retracted and the boards lowered to the height where the damp is.
Then they higher the boards , close off the hot Galden reservoir and then actively cool the upper chamber walls with an AC unit and small fan forcing the low amount of escaped Galden damp back in liquid state.
After a time period the fan stops, the top lid opens so the boards can be removed.
The (escaped) liquid Galden remains on the lid where it can be left without evaporating, next cycle it is drained back into the hot Galden reservoir when the cover re-opens. Something like that, was shown to me somewhere in 2004 I believe, and indeed without the AC unit it could be build for a couple of hundred $, which would be the price you could buy them for if two million people in the world wanted one.
The rest of the price at this moment is that they are is not mass produced machines and you need a ROI.

I always compare it to the sous vide machines, those can heat a waterbasin to 0,1^oC from 30-100^oC and those were ten years ago all over $2000. I build one myself. Today you can buy them everywhere for $300 and even the small standalone heaters you can put in any pan costs <$90.

[coppice]

IIRC the commercial VP machines have at least two chambers.
One is where the Galden is heated and stored (lower part) and the other is where the boards are placed and removed.
So you place the boards, the top lid of the top chamber is locked, the under lid is retracted and the boards lowered to the height where the damp is.
Then they higher the boards , close off the hot Galden reservoir and then actively cool the upper chamber walls with an AC unit and small fan forcing the low amount of escaped Galden damp back in liquid state.
After a time period the fan stops, the top lid opens so the boards can be removed.
The (escaped) liquid Galden remains on the lid where it can be left without evaporating, next cycle it is drained back into the hot Galden reservoir when the cover re-opens. Something like that, was shown to me somewhere in 2004 I believe, and indeed without the AC unit it could be build for a couple of hundred $, which would be the price you could buy them for if two million people in the world wanted one.
The rest of the price at this moment is that they are is not mass produced machines and you need a ROI.

That sounds like the machines which were common for small scale production in the 80s. These things have a long history, so whatever price they go for now isn't likely to change very much.

[Kjelt]

That sounds like the machines which were common for small scale production in the 80s. These things have a long history, so whatever price they go for now isn't likely to change very much.

How do modern VP machines preserve the Galden ?

[SMTech]

Vapor phase is great on paper and seemingly addresses many issues (as does Vacuum) however it does have a weakness, very few electrolytics can handle the process with most specifically excluding it as a suitable process in the datasheet. In the UK there are two visible VP phase brands ASSCON and IBL and while they are continuous operation machines the ranges are weighted to batch processing not in one end out the other. In fact my LinkedIn feed would suggest there is only one UK installation of a fully inline VP reflow process and it is new http://www.geminitec.co.uk/en/news-events-press-articles-a-more.

The other major reflow vendors either don't or barely offer VP, it has been the next big thing for years, hasn't happened yet. (IR by contrast is so dead it is barely worth mentioning).

There is something eerily similar about a number of the smaller lift in | out style machines like the Exmore and Roland Hecht which cost ~£20k which is very competitive with a convection reflow oven of potentially similar throughput (you can fill the 500x500 area up for each cycle).

The Roland Hecht IIRC quotes aloss of 1g/cycle of Galden

[D3f1ant]

I hadn't even considered if data sheets mentioned the process. If it's so safe why would device manufacturers specifically exclude it?

Does Galden ever get trapped (condense) in things like those shielded/covered modules or under parts? If so, does that cause problems for RF? Is there any residue on boards, if so how is that removed?
The machines don't seem to have vents for fumes, so presumably does that tend to coat the insides of the machine or cause contamination of the galden over time?

[coppice]

I hadn't even considered if data sheets mentioned the process. If it's so safe why would device manufacturers specifically exclude it?

I think its all about the time/temperature profiles. With vapour phase the vapour can be just 10C above the melting point of the solder, rather than the 30C or more you see in convection/IR systems. However, the peak temperature is sustained for longer. Many electrolytics react badly to the soldering temperature being sustained for too long.

[SMTech]

I think the problem is that VP work by getting every part of the assembly up to reflow temperature, electrolytics rely to some extent on the aluminum can protecting the inner. In VP the can cannot do this so the electrolyte gets a little toasty (boils). Electrolytics are one of those things that can throw up issues in reflow although I have never managed to destroy any myself..
Worth checking datasheet for any reflow advice, it can include other things like reflow once only or atypical reflow temps which might in turn affect which side of the board you could put it on for a double sided assembly.

[coppice]

The other major reflow vendors either don't or barely offer VP, it has been the next big thing for years, hasn't happened yet. (IR by contrast is so dead it is barely worth mentioning).

There seemed to be quite a lot of high vapour phase lines being installed in Chinese a couple of years ago, and the stated reason was generally energy efficiency. I don't know if that trend has continued.

[rx8pilot]

The Vapor Phase fluid is around $260/1kg from what I saw after a brief scan of suppliers.

That kind of shows the benefit of a closed system that recovers as much of the fluid as possible. I guess as @mrpackethead said - if you are not in a big hurry the entire thermal cycle can happen with the chamber closed. That, of course, allows the fluid to condense before the PCB is removed.

A 20min cycle would not be much of a problem with the way I work. Better than spending >$25k++ anyway.

[Stavos122]

By trading off speed and efficiency the complexity comes down a lot and it can be much much simpler.

The commerical systems are keeping the vapour cloud going all the time, and have a carrier system that takes the board into the vapour and then pulls it back out.  They have preheat and they have to build the system in such a way that you dont' loose the 'soft vapour'  ( the 'thin' cloud above the main cloud ) as you put the board in/out.. So you need complex vapor proof locks etc. etc.     

With the 'heater in a bucket' its super cheap, but you essentially are putting your boards on a static carrier in a cold tank.  You then heat the entire thing up.. which takes quite a bit of time and energy.  Wiht some smarts you can heat it up on a sensible 'preheat',  ( following a curve )..   When it hits 230C, you then let it cool down, back to room temp, and take the boards out.        A cycle typically takes me about 20 minutes.   A commercail system will be more like 6.

I've found that i can stack boards up, several high. ( i've done as many as four panels deep ).. Runn several tanks side by side, and i shoudl be able to get to 50-60 panels per hour, ( panel is typically A4 Size )..

It just saves so much drama with profiling and overheating.. its worth the effort.

So as the vapor cools it condenses down in the tank.  Is there any residue or droplets on the surface of the board when you are ready to remove it? 

[mrpackethead]

Aren't most current commercial offerings of the continuous operation type? Those are quite complex. When I see the occasional ad for a small batch type VP machine it seems like the crude, made in someone's back yard shed, kind of thing that mrpackethead bought and regretted. The more professional makers seem to have given up on small VP machines for prototypers, which is ironic as VP is where the big growth seems to be in overall board production.

I bought one of the IEDMS things and yes, it was very much amature hour.  Read my other post.   I see they are still selling them, i hope they have address much of the electrical safety concerns.    The guy did refund my money, and admitted that there were a lot of issues.     However while it was potentially lethal, and poorly constructed it did actually work.   There was a glimmer of hope for the process. 

[mrpackethead]

That sounds like the machines which were common for small scale production in the 80s. These things have a long history, so whatever price they go for now isn't likely to change very much.

How do modern VP machines preserve the Galden ?

My simple 'bucket' machines dont' loose much at all.   The main vapour cloud only goes about 50mm above the bottom of the tank,  theres a small amount of 'soft' vapour that just condences on the sides of the tank or the lid and falls back into it.     A very small amount sometimes is on the boards, but most of it has evaporated off the board, and it realy doesnt become and issue.

[mrpackethead]

Vapor phase is great on paper and seemingly addresses many issues (as does Vacuum) however it does have a weakness, very few electrolytics can handle the process with most specifically excluding it as a suitable process in the datasheet.

Interesting, I've done thousands of 100uF/10V Panasonic Electrolytic, and not had any issues.    I must have a deeper look into it.  The only thing i've seen a note about was some RF module that claimed it coud'tn use 'Damp Heat' Soldering.  ( not sure if that was VP, or not ).

For me, its been a whole lot better than 'great on paper'.  Its been great in practice.

[mrpackethead]

I hadn't even considered if data sheets mentioned the process. If it's so safe why would device manufacturers specifically exclude it?

Does Galden ever get trapped (condense) in things like those shielded/covered modules or under parts? If so, does that cause problems for RF? Is there any residue on boards, if so how is that removed?
The machines don't seem to have vents for fumes, so presumably does that tend to coat the insides of the machine or cause contamination of the galden over time?

I dont' use any modules like that, so cna't speak from expereince.  However my boards do tend to come out 'dry' as long as you let them cool down far enough ( around to 35C ) before you take them out.   At that point all the Galden has evoporated off the boards.. I'm not sure how it woudl work if it was inside a module.
One of the really nice things is how clean the boards come out.    The flux ends up in teh bottom of the tank, and after 50 cycles or so, i drain my galden, and filter it through coffee filters, which takes out the vast majority of the crud.    Some of it also sticks the sides of the tank.     Stainless steel wool and some IPA and a good old scrub is required.

[Kjelt]

My simple 'bucket' machines dont' loose much at all.   The main vapour cloud only goes about 50mm above the bottom of the tank,  theres a small amount of 'soft' vapour that just condences on the sides of the tank or the lid and falls back into it.     A very small amount sometimes is on the boards, but most of it has evaporated off the board, and it realy doesnt become and issue.

You have to weigh it I guess , but then if you have to weigh it , it is small potatoes.
BTW: this topic seriously needs some pictures 

[mrpackethead]

I hadn't even considered if data sheets mentioned the process. If it's so safe why would device manufacturers specifically exclude it?

I think its all about the time/temperature profiles. With vapour phase the vapour can be just 10C above the melting point of the solder, rather than the 30C or more you see in convection/IR systems. However, the peak temperature is sustained for longer. Many electrolytics react badly to the soldering temperature being sustained for too long.

My solder melts at 217, ( Henkel GC10 ), and I'm using Galden LS230.  ( 230C ).   I only hold the peak for about 20s, and then start cooling to try to get a 4C/s drop off to start with and then down to about 1-2C/s.   Cooling has been the hardest part though.    The old IEMES system had a set of fans that basically blew air over the outside of the tank,   rememebering that stainless steel is NOT a good conductor, and the way the fans where set up, they were mostly restricted.    ( I removed the metal grill, increase the outlet port sizes )..     I've tryed  using water which is pumped up to a leaky pipe, so water flows all over the outside of the tank walls.  Its taken a bit of effort as you have to keep the water off the electrical parts!      I'm thinking that it may actually be better replaced by simply putting a coil of thin wall copper pipe in the tank, and passing water through it when cooling is required.

[mrpackethead]

The other major reflow vendors either don't or barely offer VP, it has been the next big thing for years, hasn't happened yet. (IR by contrast is so dead it is barely worth mentioning).

There seemed to be quite a lot of high vapour phase lines being installed in Chinese a couple of years ago, and the stated reason was generally energy efficiency. I don't know if that trend has continued.

If you could run it continously, it woudl be verty efficent.   running it the way i do. Its not at all efficent!

[mrpackethead]

The Vapor Phase fluid is around $260/1kg from what I saw after a brief scan of suppliers.

That kind of shows the benefit of a closed system that recovers as much of the fluid as possible. I guess as @mrpackethead said - if you are not in a big hurry the entire thermal cycle can happen with the chamber closed. That, of course, allows the fluid to condense before the PCB is removed.

A 20min cycle would not be much of a problem with the way I work. Better than spending >$25k++ anyway.

yes, the Galden is NOT cheap.     But losses are low.  I've done 5000+ Panels ( A4 size ) in ours, and I've only lost about 2kg of material.

[mrpackethead]

My simple 'bucket' machines dont' loose much at all.   The main vapour cloud only goes about 50mm above the bottom of the tank,  theres a small amount of 'soft' vapour that just condences on the sides of the tank or the lid and falls back into it.     A very small amount sometimes is on the boards, but most of it has evaporated off the board, and it realy doesnt become and issue.

You have to weigh it I guess , but then if you have to weigh it , it is small potatoes.
BTW: this topic seriously needs some pictures 

The loss is low enough that its not worth worrying about.   5000+ panels and a loss of about 2kg

[rx8pilot]

yes, the Galden is NOT cheap.     But losses are low.  I've done 5000+ Panels ( A4 size ) in ours, and I've only lost about 2kg of material.

Sounds like around $.10 per panel - not exactly a financial stress point. Good to know!

Does Galden have a shelf life, pot life, cycle life, etc? Does it go bad sitting in the machine for weeks at room temp? Any maintenance like filtering, skimming, etc?

[coppice]

Sounds like around $.10 per panel - not exactly a financial stress point. Good to know!

Gee, you guys are rich. Do you realise how much negotiation most customers will engage in to drive just one cent out of their BOM?

[mrpackethead]

Sounds like around $.10 per panel - not exactly a financial stress point. Good to know!

Gee, you guys are rich. Do you realise how much negotiation most customers will engage in to drive just one cent out of their BOM?

Its for this exact reason i've followed through with using VP.  For my particular set of circumstances, it provides me the most cost effective overall solution.

- I dont' spend any time having to profile PCBs for Reflow.
- I get really good quality joints ( oxygen free environment )
- I have less issues.. 

Anything that saves time is the thing that saves me the most money.

[rx8pilot]

Gee, you guys are rich. Do you realise how much negotiation most customers will engage in to drive just one cent out of their BOM?

Yeah, totally. My Gulfstream 650 was in the shop, but I was able to use my spare jet to go to a party in Tokyo for the weekend. 

Seriously - I carefully consider my BOM cost like anyone but labor and rework tend to dominate. Anything that makes the process faster and more reliable is worth a lot more than $.10 per panel. Private jets are expensive, I have to keep my lifestyle moving at mach .94 and vapor phase will get me there, lol.

I wish I had logged the time I have spent dealing with solder problems this past year. Those problems are partially print/paste related and partially oven related. This conversation is encouraging me to look closer at VP, at least as an experiment.

[mrpackethead]

Vapor phase is great on paper and seemingly addresses many issues (as does Vacuum) however it does have a weakness, very few electrolytics can handle the process with most specifically excluding it as a suitable process in the datasheet.

I checked in with Panasonic on this topic today and found that they have covered this in their datasheet.


(6) VPS (Vapor Phase Soldering) reflow can cause significant characteristics change and/ or mounting failure due to
 deformation by acute temperature rise. VPS is acceptable provided that the process does not exceed recommended reflow profile and temperature rise
 is less than 3degC/sec.

Since i'm not dropping my boards into the vapour ( like the commerical machines do ), my Reflow profile is not goign to be nearly as steep' as it woudl be if you did just drop it in the tank, i'm nmore like 2/s

[rx8pilot]

From what I read - it seem that the commercial machines allow the system to hold a pre-heat temp that are suitably low to avoid thermal shocking parts. Then ramp at a chosen rate with the usual profile programming.

[D3f1ant]

Was having a look at the asscon machines. They have basic ramp rate profile for the entry level machines like the VP450 and 510, and what looks more standard profiling for the bigger machines like the VP800. Not exactly sure how they acheive this...but at E$40k I'm sure they pack some technology into them...to expensive to me
 

[Kjelt]

Look at youtube.
The asscon and IBL machines lower the board and monitor the temperature to follow the profile.
They certainly don't drop it in.
Perhaps the cheap machines that look like a horeca "au bain marie" machine do.

What is interesting is that both machine makers use vacuum at the end of the process.
I don't understand this since in vacuum a liquid will boil earlier at lower temperatures but perhaps this is not vacuum but lower pressure sucking the fumes from the chamber and cool the fumes down. It would be interesting to see/read more about how this last step is achieved.

[SMTech]

Well if you take convection ovens:

TWS £15-20k
Essemtec RO-400 ~ £20k
Similar Chinese but with Mesh AND conveyor  and extras same price.
Folungwin starts around £30k- paper spec rivals something like a Soltec/BTU
Basic Heller starts around £35k
BTU starts around £50k

The more expensive they get the bigger and more complicated, but the warranty also improves e.g lifetime heater and blowers on the BTU. Cheap ovens just blow hot air, they are simply a hot tunnel with controls. As you move up you gain all sorts of thing from complex air redistribution, logging, extra zones, filters, heat capacity, center support rails, UPS (the last thing you want is for your products to stop in the oven in a powercut), clever profiling software

[Kjelt]

Back to VP interesting read:
http://www.torenko.com/pdf/Todays-Vapor-Phase-Soldering-Tech-Paper.pdf

[mrpackethead]

Sorry, 'Drop it in' was not a good term.  I should have said.. "lower it into the vapour".    If you took a cold board and just dropped it into the vapour cloud quickly, the temp rise rate would likely be very high, but if you did it slowly in a controlled way, it would be quite differnet no doubt.

What i know about my 'bucket' heater.. ( its about 400mm deep ).      The amount of energy you stick in, will determine the height of the 'thick vapour'.     once its at 'temp', if i run it at ~300W it will create a cloud that is about 20mm thick.  If its 600W, it will create a cloud 60mm thick.  Above the thick vapour is a 'soft' vapour zone, where the temp is around 150-190C .. Above that is an area of warm air.   ( 80-120C or So )..  I think that in some of the commerical machines they use these varying temp zones to create a profile, by progressively dropping the board into the hotter areas.. ( counter intuivate, that the hotter area is at the bottom )..   They can keep the vapour cloud going, and control temp on the board by height.

In a bucket heater, you can acehive profile, by varying the amount of energy that you put into the system at any given time.  This is easy, you just need a thermocouple probe at board height, and a way to control the amount of energy you are putting into the system. ( if you are using an AC heater, you can just use some Zero crossing Triacs,  I'm controlling power in over 2seconds. ( 50Hz ) so i have 1% resolution.. Theres a lot of thermal mass in the system, so 2 seconds its not that long.     

Sorry been just really busy today making stuff, so will try to get some photos/info together, this weekend.    My new tanks are made from Stainless Steel.  It was made from sheet, water jet cut, and then folded, and welded..   

[Kjelt]

What is interesting is that both machine makers use vacuum at the end of the process.
I don't understand this since in vacuum a liquid will boil earlier at lower temperatures but perhaps this is not vacuum but lower pressure sucking the fumes from the chamber and cool the fumes down. It would be interesting to see/read more about how this last step is achieved.

Found the explanation:

A special form of vapor phase soldering is vacuum vapor phase soldering. After the soldering paste has completely melted, the air is pumped out in the process chamber. The negative pressure ensures that gaseous inclusions are largely displaced to the outside in the soldering points and thus separated from the soldered joint. The result are solder joints without voids. This technology is particularly advantageous when the solder joints are to dissipate heat, since air inclusions increase the thermal resistance of the solder joint. In addition, voids impair the mechanical load-bearing capacity.

So without the low vacuum , VP soldering has a high risk of introducing voids in the solderpaste, esp. critical for cooling pads and BGA's , see:

http://www.circuitinsight.com/pdf/void_free_soldering_new_vapor_phase_smta.pdf