PCB assembly versus shared equipment
Moderators: adafruit_support_bill, adafruit

Please be positive and constructive with your questions and comments.

PCB assembly versus shared equipment

by bootstrap on Thu Jul 10, 2014 9:30 pm

I've been doing other projects for nearly 4 years, and now finally just about ready to get back to building devices with PCBs with fine-pitch SMTs. At worst I need equipment to support 0201 discretes and BGA/QFN/etc with 0.50mm or finer pitch (0.30mm would be great).

When I was last at this point I was scandalized at how much assembly houses charge for prototype PCB assembly, which got me to look into buying pick-and-place, reflow-ovens and stencil machines... or building something myself. While I had some great ideas for DIY, I didn't have enough time for "yet-another-project".

Now that I'm ready to dive into this issue again, I decided I should first ask whether PCB assembly houses have become more affordable during the past 4 years. Have they? Can anyone answer that?

One alternative is to share the cost of those 3 pieces of equipment with someone else who also wants to assembly PCBs from time to time. I figure the three pieces of equipment will cost roughly $50,000 to $60,000.

Another alternative is to establish a special-purpose high-end makerspace of sorts where several people share the costs. Perhaps such a collaboration could be set up with 3 or more kinds of members, namely:

#1: contribute money and access the machine personally.
#2: contribute money and let other members assemble their PCBs.
#3: cannot contribute money but assemble PCBs for others in exchange for equipment access.

In theory I could buy all the equipment ($60K), but that's a bit crazy since I will probably only keep the equipment busy one week per year (2% utilization). With 2 people, that's $30K each. With 3 people, that's $20K each. With 4 people, that's $15K each.

Given a specialized makerspace approach with dozens of members, the price could become almost affordable. In fact, if there are some time-rich but money-poor digital wizards out there, they could be doing the PCB assembly for many "out of town" (or even "out of country") members in exchange for free access to the equipment for their own projects. This might be a way to drive the cost of PCB assembly down to [what I'd consider] reasonable levels.

The equipment that seems best to me:
pick-and-place : LE-40V
reflow oven : GF-12HT
stencil printer : SPR-45VA

My reason to chose these models is primarily... high-precision placement for fine-pitch components (but speed not crucial), reliable solder-paste application (which many say is the most problematic step), and reliable solder reflow with lead-free pastes and no damage to sensitive components. But I'm open to alternative equipment. These items are designed and manufactured by a company in the USA with a good reputation (as far as I can tell), so that might avoid problems down the road (parts, questions answered, etc).

I'm looking for general advice as well as anyone potentially interested in a collaboration or [remote-controlled] special-purpose makerspace for [at least] design and assembly of electronic devices. Note that I have other equipment that will be useful in such a setting, like a 4-channel 5-giga-sample-per-second PC oscilloscope, stereo-microscope and various other measurement gizmos and tools.

In case it matters, here are links to a couple of my PCBs that I currently need to assemble. They are probably fairly representative of the kind of PCBs that I will be designing and making in the future. They are 8-layer PCBs with 4-mil trace/space and 0.50mm pitch QFNs and BGAs (maybe the finest BGAs are 0.65mm... not sure). For scale, the larger PCB is about 150mm square and the smaller one is about 75mm square:

photo 1
photo 2
photo 3
Posts: 73
Joined: Wed Jun 02, 2010 8:47 pm
Location: phobos

Re: PCB assembly versus shared equipment

by alphatronique on Sun Jul 13, 2014 5:00 pm


DFM = Desing for manufacturing ,then you will have decent assembly price

get rid of 0201 you have lot of space so noting to justify this
then you Qoute will probably lower by 20 - 35 %

whit other DFM trick you may probably reach 50% off or even more..

shared equipment ? if something fail who paid ?
as exemple some one remove feeder while machine run and hit the head that end whit 15,000$ repair bill
did it will paid the bill or make like noting happen en let some one else find the issue ?

also best way to crash buisnis or good project was do "project" into "project"
run PCB assembly was a full time job/project on it own and you will probably surprise that most run
whit no more of 15% profit markup

so if cost to much to assembly ,stop blame fab , just go back to drawing board and this time work closely whit the fab
for have good DFM ... and decent price

sorry if look rude but was the reality

Posts: 229
Joined: Fri Jun 25, 2010 8:30 am

Re: PCB assembly versus shared equipment

by bootstrap on Sun Jul 13, 2014 9:23 pm

I worked very hard to make my PCBs designed for manufacture.

But some components are only available in fine-pitch packages, and a 0402 or large capacitor is simply too big. Frankly, at this point in history, 0201 should not be considered "overly small or difficult". Note that I'm not talking about tiny little simple PCBs that hobbyists might need. For those kind of simple PCBs, probably 0402 or 0603 can indeed be adopted. However, you just have to accept the fact that not every PCB is a "tiny, simple, hobbyist type" PCB.

I have no idea what the average reader of this forum does or builds, so maybe I'm on the wrong forum. But I tend to design rather high-performance, high-speed, elaborate devices, often with signals approaching or exceeding 1Gbps.

As an example of making the PCB easier to manufacture, note that I had the holes under the BGA balls filled and surfaced with conductive material. This is so 0201 capacitors could be placed between the fine-pitch BGA pads. Without this, it would have been just about impossible to even create a layout for this PCB, because the caps really need to be very close to the BGA ball or induced noise becomes a significant factor.


About your comments about shared equipment.

What happens if you own the entire piece of equipment and it fails, or you stick your hand or head into the running machine and break your face, neck or parts of the machine? Answer: You pay 100% of the price.

With a shared machine, any product failures outside warrantee are shared by the owners. If someone is stupid and breaks the machine, then obviously they should pay for repairs. What happens if they pretend the machine just fell apart without any stupid action of their own? Then we all get ripped off a bit when we share the repair expense.

How many repairs do you expect for brand new equipment that is operated only 5% to 20% of the time?

I understand English is not your native language, so I apologize but I can't understand your next paragraph about "project into project". You appear to be discussing how much markup assembly vendors make (15%). That may be true, I don't know. However, I can immediately see many ways our own machine will be cheaper to operate over the long run.

For example, consider what an assembly house must face! They get PCBs from dozens of PCB manufacturers, they get assembly data from dozens of different software packages, they get oddball components they've never seen before, and this happens all the time, every day. In contrast, if we have a shared machine, I suspect we'll all lay out PCBs with the same schematic and PCB layout software, so the assembly data will always be the same, and always be familiar. Furthermore, we will all try to adopt the same components whenever possible (especially capacitors, resistors, etc), and so the variety of "oddball components" will be much more limited. This also saves on cost, because we all gain the benefit of buying common components at higher volume == lower cost. And so, the efficiency of our shared machine will likely be much better than an assembly house could match.

PLUS... you're forgetting the MAIN COST... and that is labor. We will be assembling our own PCBs, and not charging ourselves for the labor we expend to assemble our own PCBs.

I do not "blame fab". I do blame the one specific company that gave me quotes, then tried to charge me MANY TIMES the price. And I was completely honest when I supplied component data to get the quote. They knew how many BGAs, QFNs, 0201s, etc. They knew everything about the PCB when I got the quote. And then they wanted MANY TIMES as much to actually do the work. So yes, I do blame that company. They were liars, they were corrupt, and they totally screwed up my prototype budget.

Since I needed to make 50 of the prototype devices, the difference in their original quote and what they tried to charge me completely screwed up my budget. Hey, a 20% error? I wouldn't say a thing, and I certainly wouldn't buy my own equipment to solve the problem. But they wanted many thousand of dollars to assemble those prototypes... about 1/3 what it would cost to buy the freaking brand new pick-and-place machine capable of 0201s and 15-mil pitch components and 0.001" placement accuracy. When one prototype project costs 1/3 the cost of the pick-and-place, that's definitely when I say to myself, "I'm gonna buy this equipment myself".

You are not rude, but you are unknowingly trying to defend a corrupt company. You are also too quick to assume that I made no effort to design the PCB for efficient manufacture as best I could given the requirements of the circuit. You made that judgement without knowing what signal speeds are involved or much of anything else. That's a mistake, but not rude. Just a mistake.

In the end, you say I shouldn't blame the assembly company, but you're happy to blame me without knowing the facts.

That makes me wonder whether you are an assembly company, and just too sensitive. Don't assume all assembly houses are 100% reasonable and perfect, and all product designers are morons. Life is not that simple.
Posts: 73
Joined: Wed Jun 02, 2010 8:47 pm
Location: phobos

Re: PCB assembly versus shared equipment

by alphatronique on Sun Jul 13, 2014 11:04 pm


i work for last 10 year for GE Smart Grid Division doing high end acquisition system
full of FPGA , Gigabit LAN ,DDR3 Memory ..
and right now i work on PCB of one of the best open source SDR radio

guess what i never use anything smaller that 0603 even on tiny sensor board on in BGA decoupling
need to carefully place cap under ship and carefully manage powerplane impedance
ok for some 0.5mm pitch BGA 0402 work bit better for decoupling , but still far from 0201

not a hobbyist or professional question here it simple dam DFM (IPC)

Yest today machine place 0201 BUT yeild was better on bigger part ,optical inspection (automated or not) was more easy
Rework was also more easy , reflow have lot less problem on bigger part and so on ..
each time you go down one generation assembly issue increase by 2 or more

also typical SMT shop have more that one SMT line ,whit 0201 you need to run on it newest line for have low defect (no one what to rework 0201)
and naturally it also the most expensive (my local competitor just put 10M$ on a new line so run ot it = $$$$
mutch better to run on it bit older line that was already paid and whit less demand on it so cost = low
( by old line i mean machine less that 8 year old normally big shop never go more that 10 year )

all this have big impact on assembly cost i knot since i pre-retired from GE last year in run full time a smt assembly fab .. (ok a small one but i pre-retired anyway ;-) )

as for own SMT machine you will quickly learn that it expensive to run and it must run , let it sit and good luck ween start it

you right every day i got new job to make and yep i make a new package 1 -2 time week remain was already see in the past
i have now bit over 1000 different package , but yes is you run machine 2 week a year you will have to make near all the package ;-)

as for Labor once again you wrong , programming was the expensive part , once programmed monkey may operate it ..
smt fab house buy and use expensive cad system for program machine that you will not got whit your machine
whit my in house tool i take 15 - 20 min to program a new run ,by hand it will take you a full day for sure ..
so what you paid for was running cost of the machine + operator ~20$/ hour
i have calculated that run my old mydata cost 300$/ hour ,a new machine may easy reach 600$
yup it seem very big but calculate term cost by machine run hour / consumable part / maitencen / local / 3 comercisl phase electricity and so on operator salary come near the last ... (did i said SMT lime was a money pit ?)

as for your assembly it have ton that float around so change to new company that honest (see 4pcb in usa)
and if you what to play and validate DFM you may have fun whit my online Qoute tool
http://www.alphatronique.com/instant-quote/ ( it not handel 0201 since i do not do it even if machine was able to do it )
but if some one Qoute you more that 50% more that me it have a problem ,since by standard i was quite expensive
and no i not what to assemble your board it try to focus on local job and my schedule was quite full anyway ..

also not forgot that pick place price was not the big part ,think feeder
my feeder collection cost > 3x the price of my pick place and each aps gold feeder cost > 800$ each
as exemple i was a small shop by current standard and i have bit over 200 x 8mm feeder (200 x 800$ = 160K $)

not to sensitive i just see customer do same reflection as you all day long ,and must said that until now no one have
save money ,pick place was a money pit and if you really what to place reliably 0201 need 1-10M$ budget
under this you will run in project into project (let me explain)

project in project was expression when you build your project (PCB of your Project/product) you need to lost time
doing make work the tool and that expend on many smaller project on it own , so you dilute your time and money
up to point that it remain noting expect ton of small uncompleted project whit noting really work
that what happen if you go into SMT

trust me after 10 year do it near full time it still scratch my head at least one a week for find a solution to a issue
when it not on how to fix some machine ,calibrate it , rebuild feeder and so on ..

so yup i try to avoid you make mistake , but at end you do what you what ;-)

and yes if you wish i may help you reduce your assembly cost by Study your BOM and top ,bottom solder mask

and finally i just find it bad that admin not come here for talk about in smt adventure
that primary reason wly i no longer monitor that forum or very rarely it do open source but look very "closed"

best regard

Posts: 229
Joined: Fri Jun 25, 2010 8:30 am

Re: PCB assembly versus shared equipment

by bootstrap on Mon Jul 14, 2014 3:05 am

I appreciate that you try to address all my questions and comments. A few comments.

First I want to say something about "design to manufacture". That is not all we need to pay attention to! We also need to worry about "design to reliability". For the components I chose, the manufacturer of the ICs clearly states that above a certain frequency (bit rate), the capacitors must be placed directly from pad to pad directly under the BGA balls, otherwise the component may not be reliable due to induced noise.

I understand that possibly they might be overly conservative, but guess what? I'm not inclined to design that way. In fact, I am inclined to design the opposite direction. I'd prefer to make a PCB slightly less optimal from a "design to manufacture" perspective if the resulting circuit (and product) are more reliably. Perhaps at super-duper high volumes (like iPhones) it is important to cut the cost-per-item down to the absolute minimum, in which case several PCB layouts can be created and thoroughly tested to figure out the best trade-off "all things considered". No product I create will be that high volume. Most will be hundreds or thousands, maybe sometimes tens of thousands, but probably never in the hundreds of thousands. Those are the size markets I am familiar with from 30 years of product development, so I can only speak about these kinds of products (high-end, high-tech, high-speed, high-performance, modest volume).

When you say "optical inspection"... is that before soldering or after (or both)?

How many errors per thousand components do you typically find by means of optical inspection (at each stage) for 0603, 0402, 0201, 0.015" pitch, 0.020" pitch, 0.025" pitch, etc? What is the source of imprecise placements? The pick-and-place machines I'm considering are "dual camera" systems and claim a placement precision of 0.001". Why would these machines place components badly in your experience? What is the reason? Mistakes the software makes when looking at the images of the component contacts? Or what? I have a couple very high quality stereo microscopes. Is that what we need for "optical inspection"?

I have heard from other folks that applying the solder paste is by far the biggest source of manufacturing errors, not precision of pick-and-place machines (at least, not for pick-and-place machines with "dual cameras" and high-precision encoders). Do you disagree? Because people claim this, I was planning to spend extra on a fancier fully automatic paste printer system. Do you disagree that part of the process is the most error-prone? If so, why do people say this? What is your advice about this stage of the process?

It is possible that I am making a mistake. But you know what? In my 30+ years of designing and prototyping electronic devices, I have had nothing but trouble when I contract out work. After which I buy high quality equipment, learn to operate it myself, and from then on I experience almost perfect reliability. This is happened on many aspects of projects, but especially on making wire-wrap prototypes (which I stopped doing many years ago because my designs ALWAYS work, and almost always without a single patch wire, and because WW is no good for [near] gigahertz circuits), but also on PCB layout and other aspects of the development process. I am very good at being alert, aware, careful, triple-checking... and then usually having zero problems with my designs and prototypes.

So my tendency is to assume my experience with SMT will follow my previous experience. HOWEVER, I understand that learning from past experience in this case might blow up in my face. Unfortunately, I have no way to know... until and unless I try.

Somehow I suspect a $10,000,000 assembly line costs so much because they need speed and throughput AT LEAST as much as to achieve precision placement. Let's hope so!

If SMT machines need to be run now and then to avoid getting rusty (or whatever), then that's another good reason to share with other people! :-)

Unless I am reading the documentation on my software and pick-and-place machine wrong, the PCB layout software outputs everything the pick-and-place machine requires to perform the layout process (possibly involving a conversion from one file format to another). The PCB layout software I run requires every component be precisely defined... the exact location and size of each pad (including range of variations), the exact outside dimensions of the package (including range of variations), and so forth. So the PCB layout software certainly does have all the information the pick-and-place machine SHOULD need to perform the placement. Of course, I might be forgetting something that is not obvious to me, but then again, I'm an expert programmer so I can easily write converters from one file-format to another... assuming the required format definitions exist.

You say "the programmer is the expensive part"... but "programming IS labor"
Posts: 73
Joined: Wed Jun 02, 2010 8:47 pm
Location: phobos

Re: PCB assembly versus shared equipment

by bootstrap on Mon Jul 14, 2014 3:15 am

..... continued .....

I don't know why, but it cut off the last half of my previous post. Fortunately I had a copy! The following is the rest of the previous message.


You say "the programmer is the expensive part"... but "programming IS labor". But also, I don't know what you mean by "programming". As far as I can tell, the PCB schematic and layout software creates files the pick-and-place machine can convert to its own format files to perform the assembly process.

The machines I am looking at are not 3 phase... fortunately. That's one advantage of less expensive machines.

Yes, I understand the feeders are expensive. They seem to cost about 1/3 as much as the pick-and-place machine for an adequate set. I already added something like $10,000 to $15,000 for feeders (can't remember exactly). However, we are not so spoiled that we would refuse to change components in a feeder between PCBs. The only reason anyone would need so many feeders as you want is... if they want to allocate a feeder to one specific component, and never change that. Obviously we cannot even begin to think that way!

I hope the machines do not break once per week. If they do, paying an assembly house is DEFINITELY a wise move. However, I don't see how the manufacturers could possibly offer a 1+ year warranty if their machines break down every week! Unless they lose their shirt for the first year, then earn huge profits thereafter when the customer must pay for repairs. Needless to say, I hope you are greatly exaggerating the problems with these machines! Maybe you bought the WRONG BRAND. And hopefully I don't too! :-o

I will try out your online quote, and also get a quote from 4pcb.com. I can't remember any more, but I think the company that screwed me over was something close to "abcpc"... though... I could be wrong.

I hate to ask you to do a lot of work looking at my BOM and PCB layouts. But if you want to take a cursory glance, you can download the BOM and some layout files at these links:

ice_eye gerber file (view with gerbv)
ice_quad gerber file (view with gerbv)

The "BOM" lists parts for both ice_quad PCB (2 pages) and ice_eye PCB (last page).
The "ice_quad" PCB is the larger one (about 150mm square).
The "ice_eye" PCB is the smaller one (about 75mm square).

You should consider this a "very slow" PCB compared to most I'm doing. Most of the signals on these PCBs are only about 250Mbps to 500Mbps. The ice_quad contains a 1 gigabit ethernet PHY that requires very clean signals because the PHY is looking at the exact shape of every signal, not just capturing high to low and low to high transitions like most digital circuits.

However, my newer PCBs contains 2.50Gbps and 3.125Gbps signals, and presumably my future PCBs will contain faster signals (gulp). So... while you may decide the PCB posted here could be designed less "carefully", my other ones in the queue certainly cannot. They have multi-gigahertz differential signals going in and out of FPGAs, and also passing through differential drivers and receivers...and to connectors. So even if you argue the current circuit can be relaxed, the newer ones cannot, and there's no point in buying any equipment if I don't continue on with my queue of PCBs and products.

I'm afraid I don't believe you when you say we need to spend $1-million to $10-million to place 0201 components! Sorry. The pick-and-place machine costs $30,000... so you're saying we need $970,000 of ADDITIONAL equipment. Here I suspect you are grossly overstating the price for a prototype shop (non-production) environment. What? You have a $900,000 stencil machine? 1000 component feeders? Not necessary!

You know what? Somebody should write a nice big up to date book on designing PCBs with SMT and assembling PCBs with SMT, including all the facts (and horror stories) about stencil-printers, pick-and-place, reflow-ovens, optical inspection, cleaning PCBs with SMT, repairing PCBs with SMT... and so forth. That would be a great book to read!

PS: You say your machine can't place 0201s [reliably]. So no wonder you don't design with 0201s. That makes sense! Perhaps a machine with "dual cameras" and 0.001" placement precision (assuming that's not a lie) puts someone in a different situation than you. I sure hope so.
Posts: 73
Joined: Wed Jun 02, 2010 8:47 pm
Location: phobos

Re: PCB assembly versus shared equipment

by alphatronique on Mon Jul 14, 2014 10:34 am


will try to make short since today have lot of job up front ;-)

1-) yes solder past printing was one of the most issue ,reason is simple it were you have the most "variability"
temperature ,consistency ,blade pressure ,alignment all this will influence the quality of the print
so need a good very rigid printer ,personalty i take one automated that modified for manual use
then you need to automated inspect it before put on pick place ,whit small pad like 0.5mm BGA and 0201
any dust particle or juts for the fun the solder remain trap on stencil hole and was not remain on your PCB
then the aperture fill and that defect will present on all remain of the production

2-) dont beleive marketing guy no way a 30,000$ machine will place reliability 0201 part no non and no
i have a 150,000$ mydata machine and yes if you talk to sale staff it will tell yes yes yes but real life was no no no
you ask how many fail / 100 that really hard to answer and delicate question for a business to reveal
but i may said that 0603 = 1 defect on 500 part 0402 = 1 defect on 100 part (may go 2 -3 time more if it bad day )
0201 = on really good day , new nozzel ,fresh calibration lot of setup tuning 5 per 100
of will let you tine to recovery form fall of your chair ;-)

wly so big number ?? reason was 1-) short run so not time to fine tune it (Feeder centering ,vision etc etc )
will go bit more in detail now ...
1-) feeder was plain mechanical stuff that may remove and replace , whatever the system it have mechanical slack on it
whit use/wearing feeder also become bit sloppy and finally tape pocket was bigger that part so it may move on it
static or humanity will change way part stick to cover tape or to pocket and so on lot of thing may go wrong
2-) mechanical stability of machine on morming power off to end of your run machine will grow by near 1mm
yup 8 feet long of steel that heat up will dilate and expend that mutch more that part size oupss once again problem come
3-) for fix above low cost machine rely on vision ,but machine vision was pain ,cap color change form batch to batch
plated solder terminal shape was never similar ,and one bust particle on camera lens and it not longer "see" the 0201
4-) hear place force was another very critical thing ,bare PCB was not flat it have up to 1mm flex on it
so head go down whit part it may hit the board and broke the tiny 0201 part or make vibration that will make move off the past another part at other side of the board ... so expensive machine never toutch the PCB but measure board height
then suck part into nozzel and for place it it shoot puff or pressure air for move part on the solder-past
5-) need to pre-reflow inspect board for missing part and defect buy AOI AOI that see and "really work" for 0201 cost alone 30,000$ alone and i not tlk here if the guy use black pcb that same color as the resistor ...
6-) reflow on 0201 it have so tinny mass that it one side have bit bigger thermal mass it melt before and pull on pad make part tombstoming ,non bigger part part mass reduce that but on 0201 it purely no mass so if have 4 thermal releve and on other side 1 trace it will fail ,for avoid this it need really good reflow 10 zone whit true hot air convection (that were that 3 phase come into play that baby take 30KW ) whiteout this forgot it
7-) post reflow AOI that for catch solder defect
8-) cleaning even if use no clean that required for be able to make final inspection ...
9-) final AOI pass ant this stage i have no more that 1 defect on 5000 part whit 0402 and 1 defect on 15000 part on 0603
but yess duriung process i have to manually fix small thing and that was really expensive since it not monkey operator but skilled person whit experience (so i me )

what to skip the 30K$ of the aoi ok but need to make it under binocular microscope and good luck ;-)

ok let talk of machine programming will try to make it short ..

the CAD package generated pick place was so usefull that near all pcb shop not event read it ..
because it not have any standard on that file format unit used and 0,0 reference point and on many cad
if user not have make propeley the lit pick point was not present or wring position ...
part rotation will never fit a point that IPC tell to not include it any more whit the file
aegis and some other make program that take file figure out the cad system and generate output file that fit your machine
but that cost near 20,000$ / seat ... yes that very ridiculous
so here i end up to hire a programmer and one year later i have my own in house tool that permit to make it under 15 minute
then you have package and part to program into machine
package was just like cad library tell machine body dimesion , number of lead , pitch , height etc etc
then have part so NE555-ND , timer bla bla bla whit use soic-8 package

so rember never trust the sale guy that will tell you it come whit software for do it this ...
so on spot whit you file on usb-key (not send it before) sit whit it on front of machine and check how long it take to make machine run and run right .. if guy refuse go away ,not forgot that you will drop 30K$ on the table no place for mistake

personaly i will go buy a used Universal GSM (< 30K) or a mydata MY15E (have one on ebay now for 65K$)
but need really good reflow oven that 10k$ used and stencil printer may find someting nice for 6K
start whit manual inspection ..

as for feeder not rare that it need 80 x 8mm feeder for one board alone whit analog board whit filter bom explose..
then you need "special" feeder for part like FPGA that need 48mm whide tape you will put 1000$ for buy feeder that
will used for a singel part on a singel desing since remain next desing may need only 32mm tape
need also lot of 12mm and so on that quicly become nightmare ,trust me the big money was on feeder
that wly need to consider this first ,since if thing go good and you grow you will what ot keep feeder
that wly i use mydata , 20 year old feeder still fit it 2014 machine ...
aps-gold have no really big and nice machine so feder will become use less later

so for resume even if machine have 1mil resolution and 10 camera that change noting to yeild of 0201 part
lay of physic also limit placement rate to +/- 4800 part/ hour / head
what really Mather was got well build and reliable machine that not broke whit time (funny my 2,3 ton mydata machine still vibrate on the floor so yes HD not last more that a year and have often conection failure and that a good machine
i have hear horror storie on small MDC machine that make whit 80/20 aliminium not ask wly it have donate for free to a hacker space that what it wort ... and yup sale doc said it place 0201 but for more that prototype ummmm that other storie

ok all my prod was late now so need to go feed the monkey ;-)

ha a n about my 1M$ budject .....

1 X mydata MY100E = 250K
1X heller reflow new = 30K
1X Stencil printer = 30K
Feeder box for 0201 12K / 16 x 20 = 192K
AOI 5 Camera = 45K
X-ray microfocus = 65K
wasing system closed loop = 25K

so was a near 700K$ and noting fancy yet only bare minimum for do "relaible" board
yes a toaster oven will melt the solder ,but solder was metallurgy if what it reliable (grain structure and so on )

have fun .. will look doc bit later today

Posts: 229
Joined: Fri Jun 25, 2010 8:30 am

Re: PCB assembly versus shared equipment

by alphatronique on Mon Jul 14, 2014 11:00 am


just a small thing that one of my decoupling ...

1 st trick i use BGA package that have mutch more ball that what i need
cost bit more but save on asembly allot ... that permit to full take adventage of first 3 row of ball
and keep center clean for power decoupling

very hight speed dif pair from both ADC was on top so now impedance bump cause by via
and it very close to FPGA so keep trace very shot and minimize reflection so avoid expensive impedance controlled PCB
even if it was calculated from PCB fab preferred stack-up ..

i will make you happy, when worked @ GE that kind of board was build in very low volume ~10 year (very specialized stuff)
and whit 600 part on it and 90 bom line assembly was very expensive then we start slowly to buy equipment from 2008 crash
now in 2014 i leave GE but keep the assembly shop that now my primary income source
so yes your idea was doable i have do it but it try to same you from error i do
my first 5 pick palce have end up a scrap yard and money put on it may have used to buy mice machine

ha i forgot i have originally planed 20K for build the shop ;-)
now i have for 250K$ of used equipment and i was still marginal and not even think doing real job whit 0201
even if all the equipment was rated for ,as said ton of other factor come into play that make yeild decrease
and not forgot that every one me included need 100% working PCB at end of the line

best regard ..
decoupling.jpg (517.33 KiB) Viewed 3274 times

Posts: 229
Joined: Fri Jun 25, 2010 8:30 am

Re: PCB assembly versus shared equipment

by alphatronique on Mon Jul 14, 2014 12:37 pm


based on quick estimate in QTH of 10 i may build the ice Qube for 138$ each (assembly) + 318$ programming (one time)
and 250$ for the stencil ... (one time)
ok programming may lower and even free based on costomer and cad it use , i have make custom program that extract
and generate mydata nativer file format from cad input that may end to 0 programing on some cad (Zuken ,p-cad ,altium ,pulsonix )

in 100 i fall to 98.00$ each and as said my price was quite hi compared to competition (4PCB was near -30% from me)

so not see how you will save so much money for build you own in-house shop

i base my qoute on

Board order quantity:10
#of BOM line item:72 ( got idea of number of reel to setup and put on machine)
Two side SMT assembly: Yes ( tell me if need 2 pick place and 2 refow run add 20% if yes)
Stencil Supplied: No
0402 Part or Smaller?: Yes ( have a single 0402 part make cost by part to increase by near 40% since yeil lower)
Number of SMD parts:392 ( total SMT part to placed by machine i factorise price per part including all cost and labor asociated to it )
Number of fine pitch parts:7 ( BGA or QFP need human optical inspection and many time place part by hand since no special feed on hand so normally it add 3 to 5$ per part that fall on this category ..)
Number of BGA/LGA parts:0 (same as above but sligly less cost / part since BGA was better yeild )
Number of PTH pins:40 (very quick estimate of trouh hole pin to solder by hand

whit all that number i may make inhouse formula that make the price Qoute ,yes on some job i lost money and on other i make lot but on average the price was good and that permit to have insten qoute instead of take 2-3 hour to check all by hand whit gerber ...

as for programming stuff it simple as (number of reel) x (fixed price) that cover cost of programming part into machine
here it have room to negotiated since it likely that machine have already on it database passive part so it just need to copy and past it whit new name ... but after all i need to make profit somewhere for be happy to ..

please note that price for setup reel on feeder was take into account on the price per board not in programming
since if you redo a run later i will to redo feeder loading whit reel ..

stencil was my supplier cost .. only trick i do here was put top and bottom on a single stencil for have 1 instead of 2
that wly i hack my printer for make it manual ,so i print only the section i need instead of the full 15" x 15"
that also save lot of solder past (no joke solder past expense was very high lot of lost past that end on metal slavager bin)

Best regard

Posts: 229
Joined: Fri Jun 25, 2010 8:30 am

Re: PCB assembly versus shared equipment

by bootstrap on Mon Jul 14, 2014 6:47 pm

I have no doubt your post contains much wisdom and experience. However, I also see some statements that are wrong.

For example, you talk about expansion and contraction of the pick-and-place machine due to temperature changes. Of course that is true, because most metals have high co-efficient of thermal expansion. However, the encoder scales are typically Silicon Dioxide (quartz) or Titanium Dioxide (called ULE quartz), and those materials have very close to ZERO thermal expansion coefficients. I know this stuff because I used to design telescope optical systems that had to retain their shape within 100 to 1000 angstroms during entire nights observing outdoors while temperature is constantly changing. And I've designed products with linear and rotary optical encoders too, so this is familiar territory for me.

So the position measurements of the PNP are based upon the quartz scale linear encoders, not the metal parts of the PNP machine.

But that's not all. Anyone who is not stupid puts fiducial marks on their PCBs, especially if they contain tiny components and/or fine-pitch contacts. When the cameras center over those fiducials, the machine knows exactly where those positions on the PCB are... in real time. And the software can re-check the location of those fiducials as many times as it wants during the assembly process. Frankly, a smart pick-and-place machine can chose just about any tiny through-hole on the PCB as a stand-in fiducial mark, so even fiducial-free PCBs should be capable of precise positioning, unless the software in the pick-and-place machine is seriously stupid.

These two facts, and other common techniques, make precision much better and easier than your posts imply.

I also note your comment about the Z-axis precision... about errors in PCB height. Any machine that doesn't have at least 2mm or 3mm of "Z-axis resiliency" is stupid! What I mean is, the placement tip should have at least 2mm to 3mm of spring-loaded or air-pressure-loaded travel. And the machine should detect the first 1mm of tip travel. Then the tip can be removed (with a simultaneous "puff" of air-pressure to assure the part does not cling to the tip).

If current machines really are as stupid and crude as you imply... they should be tossed in the dumpster! And furthermore, if they are that stupid, well, now I understand why you have so many troubles!

BTW, I sent my information to http://www.4pcb.com and they refuse to assemble PCBs unless they manufacture the PCB too.

Before I continue, understand this. I don't complain about prices. People are free to charge whatever they want. What I complained about was that company giving me a quote, then when I asked them to build the PCBs they quoted, telling me I had to pay 10 times more than their quote. That is FRAUD. And in my case, that completely destroyed my budget calculations (to make my prototypes, not production runs). I don't have unlimited budgets like many companies. That's why I got quotes in the first place!

I believe everything you say about your personal experience. However, I cannot assume that everyone else, with every other set of equipment will have the same experiences. I would hope that new equipment does better than some old piece of junk equipment that people abused for years. This is another huge variable! Some people take care of their equipment, and run it slower to reduce wear and tear, and maybe that would be good equipment to buy used. But other people take no care of their equipment, they abuse it endlessly, and when they sell it used, it is all screwed up and nothing but trouble for the poor fellow who purchased it from them.

I would also hope that new equipment is better. I would hope that all companies learn smarter ways to build their equipment every year... by observing problems in their old equipment, by observing how other companies build their equipment, by collecting statistics on where errors arise (and then focus attention to improve those areas).

You have talked a lot about errors in the assembly process. Yet the manufacturers love to talk about "maximum speed". From my point of view, I would prefer to run the machine at 1/10 speed to increase precision, reduce assembly errors, and make the machine last longer. That won't work for a company that runs all day assembling PCBs, because they'd have to purchase 5 or 10 times as many pick-and-place machines.

Think about how much time is required to inspect PCBs, find errors, rework PCBs to fix errors, debug PCBs that are not working (if that is even possible, which it isn't unless the PCB is your own so you know how it works). Frankly, the kind of people who are smart enough to debug fancy electronic circuits cost so much that it is almost always cheaper to throw away PCBs when they don't work.

So everything you say makes me want to be careful at every step (look carefully at the solder paste on the PCB before you place in the pick-and-place machine), slow down the pick-and-place machine, make sure the reflow oven is optimal, etc. My plan was to have a multiple zone oven, for exactly the reasons you stated... to improve reliability.

You had to throw away your first 5 pick-and-place machines. My guess is, they were used machines. Of course it is possible to get lucky and buy a machine that was very well designed, very well taken care of, and be happy with your choice. I am too afraid to attempt that. I will only buy brand new, up-to-date machines. That approach has treated me well in the past, and so I will continue the practice. Maybe I'm wrong, but my experience tells me to take this approach.

$200,000 for a "feeder box for 0201 components"? What the hell is a "feeder box"?

As I understand this, the MYDATA machines are all-day every-day production machines, not relatively slow machines designed to build prototypes quantities. Which is probably why that machine is 5 times as expensive as the ddmnovastar machine I'm looking at. If we include feeders, it appears like the mydata is $250K and the ddmnovastar is about $40K~$45K. On the other hand, apparently you have about 10 times as many feeders as I plan to have!!! So maybe the entire difference in cost is feeders, not the basic machine! Hahaha.


Though I think you exaggerate some issues, I take your comments seriously. I am thinking about how to approach this situation. Tell me what you think of this approach. What if I approach the ddmnovastar pick-and-place company where I was planning to get ALL my equipment (automatic stencil printer, pick-and-place machine, 3-zone reflow oven) and offer them the following deal. I will buy your machines IF... we run 25 of my ice_quad PCBs and 50 of my ice_eye PCBs and all components are placed correctly and all solder joints are good. In other words, no assembly errors.

What do you think would happen? Do you think they would refuse? What if they accept? Am I asking them for too much? Should I allow 1 or 2 faults out of those 75 PCBs to be fair? I'm serious. If these companies are BIG FAT LIARS, then why not find out? Why would they refuse?

Another variation of this would be to ask for a recent customer with the same equipment, then I could call them up and ask them to build my PCBs while I watch them. Of course, I'd pay them for doing the work, but I would say I'm thinking about buying this equipment and want to see it work before I decide.

Because... if your stories are even close to real experience (with brand new machines)... then I'm not even slightly interested in buying this equipment or assembling PCBs.

Frankly, your messages got me thinking. I've been designing and prototyping high-tech electronic devices for over 30 years now. Back before SMT was necessary, I'd build all my own PCBs by hand, and they'd all work. ALL of them. Okay, maybe not the first year I started, but after that when I had enough experience. Unfortunately, now most of the components I need are only available in SMT packages, so I have no choice but design with SMT.

But you have me thinking. My old PCBs were also about 300 to 400 components (though on larger PCBs). The reason they worked was... because I could personally see what I was doing. Which means, a very alert, careful, observant person could reliably build PCBs with virtually 100% success every single time for year after year after year.

So I wonder... is there a way to make a system today that works the same way, with the same reliability, for PCBs full of fine-pitch SMT components? I already tried a fully manual approach, and that doesn't work. Even though I had plenty good vision with my stereo microscopes, other problems made the process impractical. For example, there is a slight adhesion (or static cling) of components to tweezers (no matter what material they are made of). So once you release the grip on the component and move the tweezers away... the cling will drag the component a bit. I'm sure you know all about the troubles of trying to place components manually. Sure, it is possible to place an 0201 or fine-pitch QFN component with great effort and plenty of time (and multiple attempts), but that is too time consuming for building a whole PCB.

But... if a technique can be found to manually verify placement component by component, with as much time as necessary for the human viewer to inspect each placement... that is probably sufficient for assembling small prototype runs (like one or two or a few PCBs of each type).

I guess my first thought along these lines is the following. You set up the pick-and-place machine to pause an array of 5 cameras over each component after it places the component, but before it picks up the next component. The human viewer can then look at the component from directly above, and from four 45-degree angles (from N, S, E, W). I would guess that 90% to 99% of bad placement or goofy-looking solder-paste situations would be visible and caught by the human operator (assuming they are me, not some bored jerk of an employee at a regular company).

Now, the first response might be... "that's too labor intensive".

My answer to that would be... everything you wrote! Because if what you wrote is true (of new machines too), we ALWAYS spend an ENORMOUS amount of time finding and fixing screw-ups of the automatic processes.

My final comment is this. If SMT assembly really is as bad as you portray, then SMT was one of the most enormous screw-ups in the history of modern technology! The effort should have gone into finer pins and through-holes... or some other approach to increasing package densities. Because what you describe is ABSURD compared to previous generations of electronics.

It is a bit difficult for me to believe that the entire SMT equipment industry is one gigantic FRAUD. Of course, that is possible, for there are a great many huge frauds in the world today, so I don't rule that out entirely. I certainly need to find out whether that is the case, or perhaps you just suffer from problems associated with old models and badly misused used equipment.

Thanks for all your observations.
Posts: 73
Joined: Wed Jun 02, 2010 8:47 pm
Location: phobos

Re: PCB assembly versus shared equipment

by alphatronique on Mon Jul 14, 2014 7:53 pm


not assume that all machine have glass scale encoder
(like mydata that use pair of sealed encoder and pignon and rack this was for be finger print and dust safe)
fidicial was on pcb (ok many not put use it but anoter issue) but it not have fiducial on feeder
and on part tape , feeder spoke and hole in take have more play that size of a 0201
and your glass scale or pinon and rack was on head mot on machine base ere feeder was located
yes tou teach feeder position but in may move ,vibration and so on ..
so placement acuracy on the pcb itself was not the issue for any machine were it complicate was part pick-up / feeder

as for Z i have Said that machine normally never touch the PCB , spring or air system was way to stiff
need mutch more delicate force sensor for avoid touch PCB (my data have one nice self learning algo for this)
but really not sure aps gold handle something like that LE40 was quite crude machine like all under 60k$

as for my data i build for prototype low volume machine any one in industry knot that it not a big production machine
that wly i suggest you a used GSM-1 that a work best build like a tank and aviable refurbish to original MFG specification

when i talk about feeder box i refer to my data a feeder box take 16 x 8mm reel and yup new it cost 20k$
so around 1200$ per 8mm reel so it not big deal for that kind of feeder (Agilis that load in few second)
so if you need to handle 200 8mm tape you need 10 of that box

and mydata @ 250K was machine alone no feeder just like you le40 that will need a good 40k$ in feeder
just board i qoute for you need 80 reel last time i check aps gold it feeder was 800$ each (was was 10 year back)
so 800 x 80 reel = 64,000$ just for feeder of you old board ,houston we bust the budget ;-)

as for my 5 juki that end in scrap yup was used got it for dam cheap but have permit to learn allot before make move to real machine specially that the pick place was not the key machine

as for defect i not exagerate that the true but no one will tell it for not lose customer 0 defect at end of reflow was only valid for long run on short prototype run you have now yay time to fine tune it and trow away first bard board
so if customer send 10 board you must return 10 working board even if you knot that fisrt one will need many rework

and aps will never accept such deal ,whitout AOI whit 0201 this was simple not possible and if it do it
you may sure that i sell all my ship and but it system , i have review it 10 year back and put it on cheap machine like MDC
and it not have good reflow here , you need a heller that was at lest 10 feet long ,in aps gold if you open lid you see the heater that first bad thing if you see the heater that mean that IR was not shield , so black plastic IC bony will absorb it and heat , the solder metl and become shinny oups so it no longer heat since it mirror like so reflow do the damp contrary that what it made for ... so it junk .. good heller reflow heater was hidden in 2 or 3 aluminium box and fan force air thoud all the box like a labyrinth then air take heat from aliminium surface then it push on the pcb then come back into the box for next loop absolutely impossible to any IR leak so every thing will metl in same time and not heat damage black plastic connector ..

as smt the process was really good and weel knot process was as alway the PCB guy it loke make pcb that look HOT and pass for a king ,kind he look i am a pro i use 0201 and the new 0.4MM BGA whit buried via ... that the problem
most of my customer do industrial stuff that need to work ans i able to reach also 100% yeild
honestely i have 2 -3 return a year but that board was make whit 0805 and 0603 part so it easy to make and to inspect
peopel forgot that 0201 was make for cell phone and laptop mass market stuff that it buy a top of lime milion $ machine
that come whit it own tech and 2 year layer i trow away machine to scrap that how foxconn work ... (it trash human to but that another thing)

but yes in final if you able to find some one that run aps gold machine and you may meet it in person whit no sale rep present you will knot the right storry and see if it work in day to day like sale rep said ..
the segond option was try to make aps gold staff do one of oyur bord (i never accept more that this) and i will make it on double side tape covered board no past invoke for sure ( remember tape will make yield better for at least 4 X the actual one)

anot not gotgot that if oyu take a used GSM-1 a good selle will make traning and put warenty on it and if it fail it have ton of part aviable you may buy a complete machine calibrated for 10K$ so spare part was not a issue

and yup LE40 was same machine aps try to sell to me 10 year ago so did it have all recent technology on it ????
i think it also change name since backrout 2 year ago ?

ha i have one of it manual stencil printer + universal frame if you what it 200$ + shipping

Best regard

Posts: 229
Joined: Fri Jun 25, 2010 8:30 am

Re: PCB assembly versus shared equipment

by alphatronique on Mon Jul 14, 2014 8:17 pm

Hi that a good exemple a local compagny ...

take from wiki ...

In November 2006, Lyrtech released a small form factor software defined radio development platform in partnership with Texas Instruments and Xilinx.[1] Its Innovator division, acquired in 2006, for nearly $3M, was sold back for roughly $300,000 in late 2007 as part of a restructuring plan designed to save the company.

Lyrtech was left without assets and operations and has been delisted from TSX-V.[4] Last trades were at 0.005 (1/2 cent) per share,[5] giving the company valuation at about $200,000

the Innovator division was a montreal assembly house that was subcontractor of Lyrtech ,someday it start think the assemble do the money so it buy it the remain was history a good health company have die by try to go into assembly

Posts: 229
Joined: Fri Jun 25, 2010 8:30 am

Re: PCB assembly versus shared equipment

by alphatronique on Tue Jul 15, 2014 9:08 am

Hi you may also try to reach http://www.smallbatchassembly.com/

the come here last year take lot of my trick then build a shop and make advertise here
that also wly i delete all my older post when i explain in detail how i operate internally..

so anyway i same size and same kind of "low end machine" that you what to buy
it may a good thing to see what result it have or buy back it equipment at low price ..

Best regard

Posts: 229
Joined: Fri Jun 25, 2010 8:30 am

Re: PCB assembly versus shared equipment

by bootstrap on Tue Jul 15, 2014 7:36 pm

Alphatronique wrote:Hi you may also try to reach http://www.smallbatchassembly.com/

the come here last year take lot of my trick then build a shop and make advertise here
that also wly i delete all my older post when i explain in detail how i operate internally..

so anyway i same size and same kind of "low end machine" that you what to buy
it may a good thing to see what result it have or buy back it equipment at low price ..

Best regard

Thanks, but they only assemble PCBs with components on one side and 0603 is their smallest discrete.
Posts: 73
Joined: Wed Jun 02, 2010 8:47 pm
Location: phobos

Re: PCB assembly versus shared equipment

by alphatronique on Tue Jul 15, 2014 7:50 pm


whit it nice toaster oven that not a bid surprise ;-)
on it website it show the shop ..

Posts: 229
Joined: Fri Jun 25, 2010 8:30 am

Please be positive and constructive with your questions and comments.