Searching \ for '[EE] Reputable electronics manufacturing companies' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page: www.piclist.com/techref/index.htm?key=reputable+electronics
Search entire site for: 'Reputable electronics manufacturing companies'.

Exact match. Not showing close matches.
PICList Thread
'[EE] Reputable electronics manufacturing companies'
2011\05\09@200036 by Philip Pemberton

face
flavicon
face
Hi guys,

I've just finished prototyping and initial customer testing for one of my designs, and I have about a dozen beta-testers who have signed off on it -- "this is what we want, we like it". The hardware design is proven, the bugs have been found (and eliminated) and I want to move on with the project.

The catch is, it's currently taking two DAYS to make a single one of these boards. I just can't make them quick enough. Ramping up production would involve buying some seriously expensive kit... pick-and-place, solder paste placement, SMD ovens... you get the idea.

So what I want to do is outsource the manufacturing. The problem is, I've yet to find a reputable supplier who will even talk to me. The responses I've gotten so far basically fall into one of several categories:

  * "We don't deal with SMEs or freelance consultants." (shockingly common, even with an offer to pay up front via BACS or credit card)
  * "We're not interested in any less than <x>-off qty" (where x is between 1k and 10k units)
  * "We're already making a product for someone else that's similar to yours, and we don't want them to throw a hissy fit."
  * Utterly insane prices (£1800 per board if I supply PCBs and parts, no through-hole parts mounted, 100off qty, no discount for higher qtys.. I could buy a P&P machine and DIY it for less than that)


The board specs are:

  Two QFPs -- one 0.4mm-pitch 144pin FPGA, one 84pin 0.6mm pitch PIC.
  One QFN -- a TI TPS75003 power management chip
  About 50 passives, mostly 0603, one or two 0805. Only a couple of different values though; most of these parts are 10n and 100n decoupling caps.
  Four through-hole parts: a 40pin R/A IDC header with latches, a Switchcraft DC power connector, a resistor array and a B-type USB socket (the square one listed in the USB spec as "for target devices").
  PCB -- 160x100 double sided 1oz FR4, 7/7 thou track/space. Gold Phoenix design rules, EAGLE or Gerber/RS274X+Excellon format source files.

Does anyone know of any reputable manufacturers who could assemble something like this at a reasonable price?
Target quantity is 25 to 50 to start with (initial production run). Final intended quantity is around 250 units if sales go as projected...

Thanks,
-- Phil.
spam_OUTpiclistTakeThisOuTspamphilpem.me.uk
http://www.philpem.me.uk/

2011\05\09@212600 by Jesse Lackey

flavicon
face
Hi - I can recommend Meritronics, in San Jose, California.  Email Tam @ .....tnguyenKILLspamspam@spam@meritronics.com.  I realize you are in the UK, so it may not be cost effective / fast enough, but they have no problem with small quantity.  I use them for all my designs.  Tnt.com is fast and cheap international shipping, from california to the UK is 3 days.

Just a thought.
J


Philip Pemberton wrote:
{Quote hidden}

> Thanks

2011\05\09@221557 by David Challis

face picon face
>
> Does anyone know of any reputable manufacturers who could assemble
> something like this at a reasonable price?
> Target quantity is 25 to 50 to start with (initial production run).
> Final intended quantity is around 250 units if sales go as projected...
>
Try http://www.screamingcircuits.com  . They specialize in low volume production.
We've used them extensively with excellent results. As volumes grow, you can
move over to their turn key, high volume parent company, MEC Northwest.
They have an online quoting system so you can quickly get a feel for the
cost.

Dave Challis
QSI

2011\05\09@225802 by Oli Glaser

flavicon
face
On 10/05/2011 01:00, Philip Pemberton wrote:
> Hi guys,
>
> I've just finished prototyping and initial customer testing for one of
> my designs, and I have about a dozen beta-testers who have signed off on
> it -- "this is what we want, we like it". The hardware design is proven,
> the bugs have been found (and eliminated) and I want to move on with the
> project.

Here's (a UK based) one that has a quoting system too - if you lose the QFN/BGA the prices are not too bad.
http://www.pcbtrain.co.uk/quote-and-order-pcb/

2011\05\09@231552 by Mark Rages

face picon face
On Mon, May 9, 2011 at 7:00 PM, Philip Pemberton <piclistspamKILLspamphilpem.me.uk> wrote:
> Hi guys,
>
> I've just finished prototyping and initial customer testing for one of
> my designs, and I have about a dozen beta-testers who have signed off on
> it -- "this is what we want, we like it". The hardware design is proven,
> the bugs have been found (and eliminated) and I want to move on with the
> project.
>
> The catch is, it's currently taking two DAYS to make a single one of
> these boards. I just can't make them quick enough. Ramping up production
> would involve buying some seriously expensive kit... pick-and-place,
> solder paste placement, SMD ovens... you get the idea.
>

Based on your description, you should be hand assembling faster than 2
days per board.  Are you doing a stencil, hand place, reflow process?
I would expect to assemble and debug about ten boards / day based on
your description.

Regards,
Mark
markrages@gmail
-- Mark Rages, Engineer
Midwest Telecine LLC
.....markragesKILLspamspam.....midwesttelecine.co

2011\05\10@000332 by Mohit (Lists)

picon face
Hello Philip,

These guys are from Ahmedabad, India:
http://www.pcbpower.com/

I use them regularly for PCB prototyping and production. Although I haven't used their small-series assembly services, I can recommend them based on my satisfaction over their PCB services. Unlike the Chinese, I don't think you'll have much problem communicating with them - they speak and write decent English :-). Shoot them a mail to see if they can help you.

I also believe they provide manufacturing services to some 'front-end' PCB companies in EU and US.

HTH,
Mohit

2011\05\10@021630 by Sergey Dryga

flavicon
face
Philip Pemberton <piclist <at> philpem.me.uk> writes:


> Does anyone know of any reputable manufacturers who could assemble
> something like this at a reasonable price?
> Target quantity is 25 to 50 to start with (initial production run).
> Final intended quantity is around 250 units if sales go as projected...
>
Although I have not used PBFabexpress.com for the assembly, I am quite happy
with PCB manufacture.  They offer assembly at decent prices, small runs too..

Sergey Dryga
http://beaglerobotics.com

2011\05\10@044753 by Mike Harrison

flavicon
face
On Mon, 9 May 2011 22:15:51 -0500, you wrote:

>On Mon, May 9, 2011 at 7:00 PM, Philip Pemberton <EraseMEpiclistspam_OUTspamTakeThisOuTphilpem.me.uk> wrote:
>> Hi guys,
>>
>> I've just finished prototyping and initial customer testing for one of
>> my designs, and I have about a dozen beta-testers who have signed off on
>> it -- "this is what we want, we like it". The hardware design is proven,
>> the bugs have been found (and eliminated) and I want to move on with the
>> project.
>>
>> The catch is, it's currently taking two DAYS to make a single one of
>> these boards. I just can't make them quick enough. Ramping up production
>> would involve buying some seriously expensive kit... pick-and-place,
>> solder paste placement, SMD ovens... you get the idea.
>>
>
>Based on your description, you should be hand assembling faster than 2
>days per board.  Are you doing a stencil, hand place, reflow process?
>I would expect to assemble and debug about ten boards / day based on
>your description.

Ditto - I can't imagine how it can possibly take 2 days to make a board like that.

With a stencil (would need to be stanless for that pitch) and toaster oven, and a foot-operated
vacuum pen & strip holder for passives, you should easily get below an hour per board once you get
into the swing of things, but if you're into any significant quantity, subcontracting is definitely
the way to go.

As regards reccommeded subcontractors, I use CRS quite a lot for jobs on this scale and would
happily reccommend them http://www.crs-electronics.co.uk/
Having a local place is a significant advantage as nothing beats face-to-face meeting in the early
stages. Don't even think about going offshore for job on this scale, at least not  you've been
through the process a few times with a local sub & know all the issues.
Cost-wise, based on similar jobs I've done wth CRS   I'd guess you'd be looking at about £150 setup,
£170 for the stencil and £5-10 per board depending _very_ much on batch quantity - I'd suggest you
try to do a minimum of 50 at a time. This is for just the SMD assembly, one-sided. For the TH parts, doing it yourself may be easier/cheaper, although if the TH and SM are all on the
same side, most subcontractors will have flowsolder capabilities to do TH stuff - if it has to be
hand soldered, it can get expensive due to labour cost so you may want to DIY it.

You will need to make sure things are production-ready - A subcontractor is MUCH more likely to be
happy to work with you on small jobs if you show up with everything production-ready and know what
you're talking about, so it can go into their sustem with minimum work on ther part.
e.g.  Panelised PCBs with panel fiducials and local fids near fine-pitch parts, pick & place
location file,  parts on tape/tray with leaders & spares etc - it's often easer to get the
subcontractor to supply run-of-the-mill passives  as they probably already buy these by the  reel.
Most subs will also do complete kitting including PCB, although they charge for this, their
bulk-buying power and.or distributor discounts may offset at least some of this. If you're not familiar with  the productionizing process I very strongly reccommend you watch this
video before you do anything else:
http://www.eevblog.com/2010/11/15/eevblog-127-pcb-design-for-manufacture-tutorial/


As regards the "we don't deal with....." attitude, I've never experienced this, and you should
probably avoid trying to pursuade anyone with this attitude as they clearly can't be bothered with
your  business - the worst you should expect is having to pay in advance for the first batch,

The one thing you are  not going to find is a subcontactor willing to turn small jobs round quickly
- you'll be fitting inbeween their regular customers, and expect typical leadtimes of around 4
weeks. Some of this may be overlappable with PCB leadtime though if you book  the job in when the
PCBs are ordered.

2011\05\10@050232 by Mike Harrison

flavicon
face

>So what I want to do is outsource the manufacturing. The problem is,
>I've yet to find a reputable supplier who will even talk to me. The
>responses I've gotten so far basically fall into one of several categories:
>

>   * Utterly insane prices (£1800 per board if I supply PCBs and parts,
>no through-hole parts mounted, 100off qty, no discount for higher qtys..
>I could buy a P&P machine and DIY it for less than that)

...are you sure that price wasn't per batch of 100..?


This list of UK subcontractors may be useful http://www.buyers.electronics-sourcing.co.uk/results-cem-buyers.asp?orderby=manufacturer

2011\05\10@050740 by Mike Harrison

flavicon
face

PPS - My suggested approach would be to find some local places that look like being about the right
size, speak to them on the phone without being too committal on quantity, and arrange to go see
them. If they see you are serious and can see the product first-hand, this may help get past any
initial 'idiot filtering'. It also works both ways - you can tell a lot about a place with a quick
look round their production floor - any decent place will be more than happy to show you round - if
not, run a mile.
If you don't have a company name, make one up. No company name is an immediate turn-off for any
company offering industry-to-industry services.

2011\05\10@054402 by Oli Glaser

flavicon
face
On 10/05/2011 10:07, Mike Harrison wrote:
> If you don't have a company name, make one up. No company name is an immediate turn-off for any
> company offering industry-to-industry services.
>

I was wondering about this too, as I have never experienced what was mentioned either. If not registered as a company, this would be a good idea. As mentioned above, quite a few places simply won't bother with someone not associated with a company they can verify, especially the larger ones.


2011\05\10@054426 by Philip Pemberton

face
flavicon
face
On 10/05/11 04:15, Mark Rages wrote:
> Based on your description, you should be hand assembling faster than 2
> days per board.  Are you doing a stencil, hand place, reflow process?

Yes. The killer is the QFP -- the soldering failure rate on that thing is painfully close to 100%. It always solders down OK, but is almost always covered in solder bridges. I'm getting insane amounts of paste smear around that chip, and it's near impossible to align perfectly first-time, so I'm not exactly surprised...

> I would expect to assemble and debug about ten boards / day based on
> your description.

I can make a batch of six or seven up at the same time, but usually end up knocking a few components off one or more boards in the process, or the paste goes dry while I'm placing parts.

Assembly time is about six hours. This is for 99 total parts (6 T/H, 93 SMD, 47 unique parts). 30 of those are 10nF X7R decoupling capacitors...

-- Phil.
piclistspamspam_OUTphilpem.me.uk
http://www.philpem.me.uk

2011\05\10@054756 by Philip Pemberton

face
flavicon
face
On 10/05/11 10:02, Mike Harrison wrote:
>>    * Utterly insane prices (£1800 per board if I supply PCBs and parts,
>> no through-hole parts mounted, 100off qty, no discount for higher qtys..
>> I could buy a P&P machine and DIY it for less than that)
>
> ..are you sure that price wasn't per batch of 100..?

Pretty sure...

> 10day and 20days services not available at moment, due to maintenance.
> 5day is only available, we need to manufacture and assemble pcb's inhouse. Parts must be supply to manufacturing facility in China, UPS, all fees prepaid. We need UPS account to send you completed boards.
> Price is 2875 USD pr board.

-- Phil.
@spam@piclistKILLspamspamphilpem.me.uk
http://www.philpem.me.uk/

2011\05\10@060443 by Mike Harrison

flavicon
face
On Tue, 10 May 2011 10:44:21 +0100, you wrote:

>On 10/05/11 04:15, Mark Rages wrote:
>> Based on your description, you should be hand assembling faster than 2
>> days per board.  Are you doing a stencil, hand place, reflow process?
>
>Yes. The killer is the QFP -- the soldering failure rate on that thing
>is painfully close to 100%. It always solders down OK, but is almost
>always covered in solder bridges. I'm getting insane amounts of paste
>smear around that chip, and it's near impossible to align perfectly
>first-time, so I'm not exactly surprised...

How are you pasting it ? Paste printing quality is very important.
A thin stainless stencil woks wonders. Also maybe your paste isn't fine enogh - there are several
grades.

{Quote hidden}

You can easily do passives  at 15-30 per minute, by picking from tape with a foot-operated vacuum
pen (The finger operated ones with internal vacuum are useless, but a good base to DIY a proper one
from). I stick all the tapes down to small bits of thin MDF and peel just enough cover tape for each batch.

2011\05\10@060456 by Oli Glaser

flavicon
face
On 10/05/2011 10:44, Philip Pemberton wrote:
> On 10/05/11 04:15, Mark Rages wrote:
>> >  Based on your description, you should be hand assembling faster than 2
>> >  days per board.  Are you doing a stencil, hand place, reflow process?
> Yes. The killer is the QFP -- the soldering failure rate on that thing
> is painfully close to 100%. It always solders down OK, but is almost
> always covered in solder bridges. I'm getting insane amounts of paste
> smear around that chip, and it's near impossible to align perfectly
> first-time, so I'm not exactly surprised...

Hmmm, that sounds like something is maybe not quite right with your approach.
I would maybe try using normal solder and drag solder it. I don't consider myself to be too good at soldering, but I can manage a 100 pin QFP in a minute or two with this method. Can't remember having any real disasters - if I put a little too much solder on the tip (a pace miniwave - the smallest one, has "wave" at the end anyway) and get a bridge I just dab it with solder wick.
I have done a similar prototype board this way recently with FPGA, PIC, various other ICs, 50+ passives in around a hour and a half. I imagine others with some real skill can do it much quicker.
Are you using lead free solder? How exactly are you attempting this? (paste + heat gun or ??)

2011\05\10@063810 by cdb

flavicon
face
I used to do rework for a hearing aid manufacturer.

Tips.

Purchase the horrendously expensive  syringe of gel flux from RS - put small amount on all pads. This allows you to 'wipe' the bit and solder (make sure the solder is SMD gauge) the flux helps create surface tension and most of the solder will zonk to the pads, what bridges, can be removed with desolder braid.

If there is a hoof tip made for your brand of iron this is a worthwhile investment, you fill the hoof up with solder and just run it down the legs.

Even better by a QFP tip - Antex do them if you have an Antex iron but so do JBC and others.
Solder tweezers make chip type resistors and caps quite easy to align and solder, again some brands of iron also have chip bits.

Make sure you have the right flux to go with the brand and type of solder paste you are using. Make sure the pot of solder paste is kept cool - in a fridge if possible when not being used - not a food fridge obviously.

Colin
--
cdb, KILLspamcolinKILLspamspambtech-online.co.uk on 10/05/2011
Web presence: http://www.btech-online.co.uk   Hosted by:  http://www.justhost.com.au
 

2011\05\10@064124 by Djula Djarmati

flavicon
On 10-May-11 11:47, Philip Pemberton wrote:
> On 10/05/11 10:02, Mike Harrison wrote:
>>>     * Utterly insane prices (£1800 per board if I supply PCBs and parts,
>>> no through-hole parts mounted, 100off qty, no discount for higher qtys...
>>> I could buy a P&P machine and DIY it for less than that)
>>
>> ..are you sure that price wasn't per batch of 100..?
>
> Pretty sure...
>
>   >  10day and 20days services not available at moment, due to maintenance.
>   >  5day is only available, we need to manufacture and assemble pcb's
> inhouse. Parts must be supply to manufacturing facility in China, UPS,
> all fees prepaid. We need UPS account to send you completed boards.
>   >  Price is 2875 USD pr board.
>

Wow, does anyone ever pay these prices?

Some really good advices, especially from Mike Harrison, he wrote exactly what my thoughts are except making up a company name - if they don't want to work with a freelance consultant there is plenty of us who do..

Philip, can you please send me the BOM and the silkscreen layers for a quote? From the information you gave, I think the price for this should be less than 350 GBP plus shipping.

Still, I believe for 25 units the best way is to do it yourself, that is just a couple of days work. Get someone to help you preparing the components for you - when we do one-off prototypes by hand we call this arrangement Pick & Place.

Djul

2011\05\10@072932 by Philip Pemberton

face
flavicon
face
On 10/05/11 11:03, Mike Harrison wrote:
> How are you pasting it ? Paste printing quality is very important.

Pretty much how SparkFun show in their videos. A metal squeegee, solder paste (Multicore 318LF) and a plastic stencil.

> A thin stainless stencil woks wonders. Also maybe your paste isn't fine enogh - there are several
> grades.

Looks like I'll be buying a stainless stencil then... Newbury seem to have pretty good pricing on these (£30/ea for the Eurocard sized board I'm using).

> You can easily do passives  at 15-30 per minute, by picking from tape with a foot-operated vacuum
> pen (The finger operated ones with internal vacuum are useless, but a good base to DIY a proper one
> from).

I've DIY'd one... but my vacuum pump is a bit weak. It's basically a fish-tank air compressor with the one-way valve reversed. About as much sucking power as an asthmatic mouse breathing through a straw.

My biggest problem at the minute is smearing the paste with my hand...

"C1, C2, C... DAMN!"

I've been trying an "assembly line" type method -- install all the parts of a given value on all boards, then move on to the next value. This reduces the time-per-board, but I usually end up getting the Q-tips and solder paste out to re-coat pads where the paste has smeared or smudged when the boards have been moved...

Thanks,
-- Phil.
RemoveMEpiclistTakeThisOuTspamphilpem.me.uk
http://www.philpem.me.uk/

2011\05\10@073947 by Philip Pemberton

face
flavicon
face
On 10/05/11 11:04, Oli Glaser wrote:
> Hmmm, that sounds like something is maybe not quite right with your
> approach.
> I would maybe try using normal solder and drag solder it.

That's what I've started doing -- removing the solder paste, reflowing the board in the oven to mount the passives, then installing the FPGA with the soldering iron.

> Are you using lead free solder? How exactly are you attempting this?
> (paste + heat gun or ??)

Paste + PID-controlled oven (modified "mini oven" / toaster oven). The through hole parts are installed by hand with a soldering iron.

-- Phil.
spamBeGonepiclistspamBeGonespamphilpem.me.uk
http://www.philpem.me.uk

2011\05\10@075607 by Mike Harrison

flavicon
face
On Tue, 10 May 2011 12:39:45 +0100, you wrote:

>On 10/05/11 11:04, Oli Glaser wrote:
>> Hmmm, that sounds like something is maybe not quite right with your
>> approach.
>> I would maybe try using normal solder and drag solder it.
>
>That's what I've started doing -- removing the solder paste, reflowing
>the board in the oven to mount the passives, then installing the FPGA
>with the soldering iron.

I do it the other way round - fine pitch devices go on first, and have a few pins tacked down during
placement - as long as it stays aligned, almost any amount of shorts can be cleaned up easly with
braid, or by fluxing and reflowing a side at a time with a wide blade tip.

2011\05\10@083244 by Olin Lathrop

face picon face
Philip Pemberton wrote:
>    * "We don't deal with SMEs or freelance consultants." (shockingly
> common, even with an offer to pay up front via BACS or credit card)

So don't tell them you're freelance, and certainly don't offer to pay by
credit card.  They probably have no way of accepting credit card payment,
and just the mention of it will cause many to label you as more trouble than
you're worth and bail out.  They will tell you they can't do the job for
some reason or quote a ridiculous price.  Either way you go away and they
can stop wasting time on you.  The usual way to pay foreign manufacturers is
via bank wire transfer.  My bank charges $40 on my end of each transfer.
This is yet another reason to make the total order large enough.

If you are a unknown entity, then they may expect a large portion of the
cost up front.  That's fair if you are truly unknown and have no Dunn and
Bradstreet or equivalent rating.  From the manufacturer's point of view,
they are running the risk of sending you the finished goods and never
hearing from you again.

Even if you are known to the manufacturer, it's reasonable to pay some part
up front for large orders.  This should be about what it costs the assembly
company to buy the parts if you have them doing the kitting, since that's up
front cost on their end.  For example, we're having Djula doing a build for
us right now and we sent him $25K up front to cover parts costs.  We'll
still owe him another chunk on delivery after we've had a chance to inspect
the product.

>    * "We're not interested in any less than <x>-off qty" (where x is
> between 1k and 10k units)

Some manufacturers are set up for high volume only, and doing small lots
breaks their process.  Others understand that it's really just a matter of
money and price the small jobs accordingly.  If you want really small lots,
like under 100, then you probably need to find a local place and pay a
considerable premium.  You may need to find a different place to put the kit
together.  Some manufacturers are only that and aren't set up to do
purchasing and keep track of stock.

>    * "We're already making a product for someone else that's similar
> to yours, and we don't want them to throw a hissy fit."

That's not a problem in China ;-)

>    * Utterly insane prices (£1800 per board if I supply PCBs and
> parts,
> no through-hole parts mounted, 100off qty, no discount for higher
> qtys..
> I could buy a P&P machine and DIY it for less than that)

"We don't want to do small jobs.  Run along kid until you grow up and are
ready for a real manufacturing run.".

> Target quantity is 25 to 50 to start with (initial production run).
> Final intended quantity is around 250 units if sales go as
> projected...

Geesh, grow a pair and get at least 100 made.  If you're not serious, you
can't expect others to be either.  Anything less than 100 and it's
understandable why many won't want to bother.  At small quantities it will
be mostly setup charge anyway.  The incremental cost of going from 50 to 100
won't be that large if you get realistic prices for each.  The 1800/board
quote you got may well have been the 100 lot price divided by the number of
boards you asked for.

> Does anyone know of any reputable manufacturers who could assemble
> something like this at a reasonable price?

I you decide to get serious, give Djula a chance to quote the job
(TakeThisOuTdjulaEraseMEspamspam_OUTdatatehnik.com).


********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014.  Gold level PIC consultants since 2000

2011\05\10@084923 by Olin Lathrop

face picon face
Philip Pemberton wrote:
> Yes. The killer is the QFP

There should be nothing wrong with QFP for a real manufacturer.  However,
that brings up the question of whether you have designed this with
manufacturing in mind.

Do the thru hole parts really need to be thru hole?  Realize that this will
cost more and add process steps.  Are your board drawings, location index,
BOM, etc all ready to go?  Are there fiducials in at least two opposite
corners?  How is the manufacturer going to test the boards?  Do you have a
test jig?  How will all the programmable parts be programmed?  The procedure
may be obvious to you, but it needs to be spelled out in detail, preferably
with pictures, so that someone that can't spell "PIC" can do it.  If this
requires specific hardware and software (PIC programmer, cable, PIC
programmer app), you have to supply this and specify what kind of machine it
needs to be run on, how to install it, etc.

Putting together the manufacturing documentation can be significant work,
but it's just another part of the overall product development process just
like designing the schematic and routing the board are.


********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014.  Gold level PIC consultants since 2000

2011\05\10@085517 by alan.b.pearce

face picon face
> > Target quantity is 25 to 50 to start with (initial production run).
> > Final intended quantity is around 250 units if sales go as
> > projected...
>
> Geesh, grow a pair and get at least 100 made.  If you're not serious, you
> can't expect others to be either.  Anything less than 100 and it's
> understandable why many won't want to bother.  At small quantities it will
> be mostly setup charge anyway.  The incremental cost of going from 50 to 100
> won't be that large if you get realistic prices for each.  The 1800/board
> quote you got may well have been the 100 lot price divided by the number of
> boards you asked for.

To be fair, I went to a manufacturing fair at Farnborough recently where there seemed to be any number of assembly houses who were touting for small quantity assembly business. One told me he was quite happy to do singles, and would baulk if asked to do anything over 2k units.

Check here http://www.industry.co.uk/default.asp to see what shows they have in your area (I went to Southern Manufacturing 2011). You should be able to pre-book and get in for free.
-- Scanned by iCritical.

2011\05\10@091323 by Mike Harrison

flavicon
face
On Tue, 10 May 2011 08:49:58 -0400, you wrote:


>  How is the manufacturer going to test the boards?  Do you have a
>test jig?  How will all the programmable parts be programmed?  The procedure
>may be obvious to you, but it needs to be spelled out in detail, preferably
>with pictures, so that someone that can't spell "PIC" can do it.  If this
>requires specific hardware and software (PIC programmer, cable, PIC
>programmer app), you have to supply this and specify what kind of machine it
>needs to be run on, how to install it, etc.

For small runs I usually find it easier to get the boards back as-is and program/test myself - it's
typically way quicker than documenting the process. You ideally need to have tested a reasonable
number yourself to get an idea of typical problems and symptoms, to generate useful faultfinding
guides. If you're doing something that will be in long-term production, a good approach is to test
the first batch yourself so you can give the manufacturerer a more useful test setup for future
builds.

Another important thing to do is to build in as much selftest functionality as possible into the
product's firmware, so as soon as firmware is programmed it can test a reasonable proportion of the
hardware, bearing in mind that you will be primarily looking for opens (espeically lifted or dry QFP
pins) and adjacent-pin shorts.

Another benefit of getting boards untested, especially from a new subcontractor,  is that it is a
good indicator of their underlying process quality - if they are testing and reworking you may have
no idea of how many  boards failed the first time.

2011\05\10@092751 by Olin Lathrop

face picon face
Mike Harrison wrote:
> For small runs I usually find it easier to get the boards back as-is
> and program/test myself - it's typically way quicker than documenting
> the process.

That may be, but then you have little recourse when there are failures.  You
are also not giving the manufacturer a means of detecting problems with
their process, and therefore can't expect them to be corrected.

Doing a reasonably complete job of testing a board can take time, especially
without automated tools.  This is something you want done at the
manufacturer's labor rate, not yours.  Carefully documenting the process is
something you should be doing anyway, whether you perform it or someone
else.


********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014.  Gold level PIC consultants since 2000

2011\05\10@093818 by Wouter van Ooijen

face picon face
> I've been trying an "assembly line" type method -- install all the parts
> of a given value on all boards, then move on to the next value. This
> reduces the time-per-board, but I usually end up getting the Q-tips and
> solder paste out to re-coat pads where the paste has smeared or smudged
> when the boards have been moved...

interesting thread!

my experience:

I think I have done some 200 .. 300 of (the various variations of) my DB038's so far, http://www.voti.nl/DB038/. The bottom side has most of the SMD parts. Total is ~ 200 components. The TH's take most of the time. Assemly-line style pick-n-place is definitely faster than doing one board at a time. I prefer ~ 10 boards at a time. I need a big flat table to shift the boards around. I just grab a bunch of the SMDs from their container and put them on the table. I use metal tweezers for the smd (mostly 1208, some 0806, smallest pitch is an FT232RL). I use a plastic stencil now, but it has deteriorated too much.  Will switch to metal. I use a toaster oven with a (DB038 - based :) temperature profile control. Biggest drawback is unequeal heating (no fan). Will probably switch to a more professional oven. I can do ~ 8 boards in an 8 hour working day.

If I redo the board design I will make the board larger, replace more TH with SMD, and maybe try to put all SMD one one side. Even for hand-prototyping, SMD is much faster than TH.

I am looking at the suggested assembly-houses that have an on-line quoting system, very interesting. But in its current form of my board the TH takes most of the time, especially the connectors which I prefer to keep TH, so I am not sure doing the SMD assembly externally would save me much time.

--
Wouter van Ooijen

-- -------------------------------------------
Van Ooijen Technische Informatica: http://www.voti.nl
consultancy, development, PICmicro products
docent Hogeschool van Utrecht: http://www.voti.nl/hvu

2011\05\10@095115 by Mike Harrison

flavicon
face
On Tue, 10 May 2011 09:28:25 -0400, you wrote:

>Mike Harrison wrote:
>> For small runs I usually find it easier to get the boards back as-is
>> and program/test myself - it's typically way quicker than documenting
>> the process.
>
>That may be, but then you have little recourse when there are failures.  
Failures at this stage are at least as likely to be down to design as manufacturing issues - you can
easily investiagte both, the manufacturer is less able to. If they stuffed the wrong parts, and your
docs are right, then by all means send them back for rework, For an initial run quantity will typically be low enough that any rework isn't a major deal.

>are also not giving the manufacturer a means of detecting problems with
>their process, and therefore can't expect them to be corrected.

This can be fed back to them if necessary.

>Doing a reasonably complete job of testing a board can take time, especially
>without automated tools.  This is something you want done at the
>manufacturer's labor rate, not yours.  
Except that you will generally be able to diagnose faults _much_ faster than the manufacturer, since
you know the design inside out, and can do things like tweak firmware to help find faults, which is
why testing at least an initial batch yourself is useful to optimise the test procedure and be able
to do some 'if it does this, check parts x,y,z first' type guidance to save the manufacturer time,
and hence your money. And to debug any test software of course - a test system failure can put a major spanner in the
works for everyone's schedule.

Of course you should always discuss test with the manufacterer at an early stage (ideally before
final PCB layout) as they will probably be able to offer useful advice as to what will make things
most efficient for their process and capbilities.

>Carefully documenting the process is
>something you should be doing anyway, whether you perform it or someone
>else.

Absolutly, but when time gets tight, it's usually the first thing to get corners cut...

2011\05\10@095628 by PICdude

flavicon
face
Quoting Philip Pemberton <RemoveMEpiclistspamTakeThisOuTphilpem.me.uk>:
> Yes. The killer is the QFP -- the soldering failure rate on that thing
> is painfully close to 100%. It always solders down OK, but is almost
> always covered in solder bridges. I'm getting insane amounts of paste
> smear around that chip, and it's near impossible to align perfectly
> first-time, so I'm not exactly surprised...


I can hand-solder a 44-TQFP PIC with almost 100% reliability in under  5 minutes, most of that time being used to let the chip cool between  soldering the four sides.  And I don't consider myself a soldering  pro.  Perhaps you need to re-think the processes you use, leaving off  some parts for manual soldering.

Cheers,
-Neil.


2011\05\10@104911 by alan.b.pearce

face picon face
> >Doing a reasonably complete job of testing a board can take time, especially
> >without automated tools.  This is something you want done at the
> >manufacturer's labor rate, not yours.
>
> Except that you will generally be able to diagnose faults _much_ faster than
> the manufacturer, since you know the design inside out, and can do things like
> tweak firmware to help find faults, which is why testing at least an initial
> batch yourself is useful to optimise the test procedure and be able to do
> some 'if it does this, check parts x,y,z first' type guidance to save the
> manufacturer time, and hence your money.
> And to debug any test software of course - a test system failure can put a major
> spanner in the works for everyone's schedule.

The other trick, if using a chip like a PIC24 or any others with JTAG is to use the JTAG interface for the board test at the assembly house - doesn't need any software in the chip if you do it right (AIUI, haven't tried doing it), so your IP doesn't go AWOL on a 'lost' chip.

I got this from talking with a supplier of JTAG test gear, at the previously mentioned show I went to. It seems they have libraries of components that can be interfaced through JTAG, and help the board house set up the test program on the JTAG unit.

OK, you lose a handful of pins to the JTAG interface, but these can be brought out to pads for fitting resistors to indicator LEDs after the JTAG test, so the pins don't necessarily need to be lost permanently, you always need some indicators.
-- Scanned by iCritical.

2011\05\10@105832 by Philip Pemberton

face
flavicon
face
On 10/05/11 14:13, Mike Harrison wrote:
> For small runs I usually find it easier to get the boards back as-is and program/test myself - it's
> typically way quicker than documenting the process. You ideally need to have tested a reasonable
> number yourself to get an idea of typical problems and symptoms, to generate useful faultfinding
> guides. If you're doing something that will be in long-term production, a good approach is to test
> the first batch yourself so you can give the manufacturerer a more useful test setup for future
> builds.

My test procedure (derived from a dozen prototypes, one complete write-off, eleven usable) is -- roughly:

  * Power up with a 50mA limit and 3V supply. Ramp the voltage up to about 9V. Abort test if current exceeds 50mA at ANY point in the testing. Confirm that 5V, 3V3 and 1V2 are within range (5%).

  * Install BOOT jumper. Connect PIC programmer via Millmax header. Increase current limit to 500mA and program PIC with Bootloader and initial firmware image. Check that MCU STAT LED is blinking rapidly. Disconnect power and programmer, remove jumper, power up from standard PSU.

  * Connect to PC. Put hardware into "Secret Squirrel" (Production Test and Debug) mode. Run full suite of tests:
    * FPGA programming
    * FPGA <=> MCU communication
    * Static RAM test
    * Board-mounted LEDs and jumpers
    * I/O test (requires switch-and-lamp test board)
    * Final read-write test (connect a cable kit and drive, see if the read-write and seek circuitry works).

A complete test takes a couple of minutes, and the Python script I'm using auto-detects new hardware, runs the tests and assigns and records the board serial number if the tests pass. It's entirely possible (in theory!) to hook up a USB hub and run multiple board tests at once, but I haven't tried that.

All I'd expect an assembler to do is the power test... as long as the power comes up, it's reasonable to expect that the board is at least borderline functional and -- at most -- needs a few opens resoldering around the FPGA bus, I/O buffers or SRAM.

The tests have advanced to the point where they can detect individual stuck bits on the SRAM address and data bus, and single or multiple stuck bits on the FPGA comms bus. This has proven extremely useful when building the prototypes!

-- Phil.
piclistEraseMEspam.....philpem.me.uk
http://www.philpem.me.uk

2011\05\10@111219 by Josh Koffman

face picon face
On Tue, May 10, 2011 at 9:37 AM, Wouter van Ooijen <EraseMEwouterspamvoti.nl> wrote:
> I am looking at the suggested assembly-houses that have an on-line
> quoting system, very interesting. But in its current form of my board
> the TH takes most of the time, especially the connectors which I prefer
> to keep TH, so I am not sure doing the SMD assembly externally would
> save me much time.

My current project has a few TH components that can't really be
changed to SMT easily (terminal blocks, etc). To help, I've milled a
block of plastic into a sort of negative mold for the board. After
toasting on the SMT components, I plan to populate the board with TH
bits, put the plastic block on top of the board, flip them both over
and solder. I'm hoping that makes it less finicky to ensure that the
TH devices are seated correctly.

Josh
-- A common mistake that people make when trying to design something
completely foolproof is to underestimate the ingenuity of complete
fools.
        -Douglas Adams

2011\05\10@114225 by PICdude

flavicon
face
Quoting Philip Pemberton <RemoveMEpiclistEraseMEspamEraseMEphilpem.me.uk>:

>> A thin stainless stencil woks wonders. Also maybe your paste isn't  
>> fine enogh - there are several
>> grades.
>
> Looks like I'll be buying a stainless stencil then... Newbury seem to
> have pretty good pricing on these (£30/ea for the Eurocard sized board
> I'm using).

Your choice of stencil will make a BIG difference.  Typically 5-6 mils  is common, and you can control/reduce the amount of solder by  shrinking the pin apertures a bit.  I do get occasionally bridging  with a 6-mil stainless stencil for a 28-SSOP PIC, so I'll be going  5-mils on the next stencil.

The lower-cost stencil sources are appealing (~$100), but getting the  same stencil from a more established source will usually get you some  design experience with mods to prevent problems like bridging.

Cheers,
-Neil.

More... (looser matching)
- Last day of these posts
- In 2011 , 2012 only
- Today
- New search...