Prototyping Surface Mount Components??

Greetings,
I am new to surface mount components, so let me seek advice from
those who are well knowlegable. What is a good way for a hobbyist to
prototype these components? When I used to do bread board type work
with the HC11, I would stick the chip in the programmer, then bring it
back to the board, and test it. The wiring was easy enough to change,
and reprogramming the controller was easy. How does this work with
surface mounts?

As I understand, you most likely have to build a pcb for the circuit.
What is a cheap way to do this? Are there any good references out there
for equipment that can be used at home instead of in an invloved lab?

How do you modify the program on these units that are surface mouned
microcontollers? Programmers and evaluation boards, usually have one
chip from the family, but that may not be the one you want to use. It
may have a different memory size. What does one do about that? Or does
every srface mount MCU have its own evaluation board? Should I get a
separate programmer for these devices, since the chip on the eval board
cannot be removed? I am thinking of using the Atmel AT91RM9200 for
example.

As you can see, I am new to this and any references, hints and pointers
will be of great help.

Cheers,

A

 

You might be surprised what is possible, with just hand tools. A
fine-point soldering iron, and one of those illuminated desk magnifiers,
will take you quite a long way. Tweezers to position components.
Most packages except BGAs can be done this way. You'll find solder paste
easier to use than solder wire, as things get smaller.
Even BGAs have been done at home, using an oven or (horrors!) a toaster.
But this is definitely an advanced technique.
As to programming, don't consider a CPU that cannot be reprogrammed in
situ. Most modern CPU's can be programmed (often debugged also) via a
JTAG port, or the like. Always provide for this in your PCB layout.

For PCBs, I have given up making them at home. A local fab shop offers a
really good price for prototypes, if you don't need them yesterday. (He
waits till he can fit them in the corner of a panel that's going through
anyway.) The finish, accuracy & general quality saves me so much stress
in putting it together.

 

I would go a little further and suggest a low-power microscope for under $100.
For 58 year old eyes, I need all the help I can get.

Besides JTAG, there are vendor-specific In Circuit Serial Programming
connections, like the 5-wire connection used on Microchip parts. Low end micros
are more likely to have these than JTAG.


Robert Scott
Ypsilanti, Michigan

 

Solder paste is overrated. Joke, but there's a serious point to it. To
solder a chip with two rows of very fine pin spacings on each end, I've
used a normal iron and solder. I tried all sorts of things at first, then
found that the old way works better by far than anything you'd expect to
use with SMT. The trick is to keep the tip very clean, apply the solder
liberally, and use gravity and cohesion on the iron to draw the solder away
as you move along the row. This leaves shiny joints that are not starved,
and a bit of fluxed braid can remove the solder bridge that is usually left
on the last two or three pins, and if you get it right you won't even need
to do this.

Close viewing is vital, it will show you the results of your technique in
merciless detail. > I've found that the same idea is good for SOT23 and
various tiny capacitors and resistors, I place the acute angled edge of the
tip against the PCB, just clear of the part with the flat of the tip
facing it, make a small bead of melted solder there, then quickly bring it
to the part and back away, and the resulting flow makes a very neat raised
fillet. I anchor the part with a tweezer or jewellers screwdriver for first
contact, then do the other end properly as I decribed, then finish with
proper flow to the first end.

In all cases I've found that the best way to get good SMT contacts is to
use plenty of solder, not to worry about too much heat, because this is the
way to make the contact fast. Being fast is a better way to avoid excess
heat than trying to reduce the amount of hot solder.

I do have quite a bit of waste solder collected though, over the years.

 

For the Atmel AT91RM9200, they have an onboard serial bootloader that
be activated based upon a jumper, so the programming of your board is
done in-system. To program it, you will need a serial port on your PC,
a terminal program like minicom or hyperterminal and off your go.

However, I would recommend you get one of Atmel's eval board, either
the DK or the EK. I have had very good experience with the DK eval
board, and it certanily reduce our bring up time on the whole project.

 

I have not used a bootloader before. I would assume it is straight
forward to load the new program after you get the serial port up and
running. Am I correct?

Yes, I am thinking of getting one of these if I can come up with the
money to do so. Does anyone have any recommendations of good
evaluation kits for the AT91RM9200?

 

[snip]
Whatever works for you
I have tried paste & wire, and I prefer paste. Just my $0.02

 

In comp.arch.embedded,

Not on this chip. The bootloader can only download code to the internal
SRAM (16KB but can only use about 12K download size) and start that.
Usually you will load a small loader program that sets up memory etc.,
downloads the real code and programs it to flash. The loader can use
the internal ROM services, so doing an X-modem download is very easy.
I believe the eval kit comes with samples of this but you should also
be able to find something on the internet.

 

I had a hell of a time trying to get all the paste to melt, had to clean up
a load of unmelted goop and start over. Fluxed wire and a standard tip
worked very well. I already decribed how to do it so I won't repeat it.
The results are extremely good though, before I did it I had some terrible
things to contend with.

One nice example was the poor heat conduction with paste or inadequate
solder by wire. I overheated a track on a tiny row of 80 IC pads, and it
was dragged by the iron, half turned around its axis. The easiest way to
repair was turn it fully over, scrape and tin the other side, then use a
needel tip to align it and flatten it ready for the IC replacement. That
worked, but it was a nightmare task, by far the most delicate bit of
'surgery' I have done. Making easy contact by lowing a generous bead of
solder from a fluxed wire is a good way to avoid anything like that
happening. It's also the cheapest way of amking sure that no solid ever
comes into contact with the work, risking damage to it while applying
solder. As solder flow is uased in commercial manufacture, it makes good
sense to use the same machanics on a small scale to get good contacts made.

I still keep paste, I'll probably want it to make small SMT layouts where I
heat the whole board carefully to join all parts in one operation, but I
don't like it for small repairs. You can't melt it with an iron tip easily,
because the flux melts vigoruously, carrying the metal powder away too
rapidly to allow melting and cohesion. All that results is a mess that
needs cleaning up. That same standard tooling gets good results with wire
though.

If anyone reads this and wonders, just try it. Try both, and if you post
about a way to use solder paste with a stanard iron on a cold board without
damage to pads or parts, I'll read it and try it, but without that, my own
experience will convince me more than anything else.

 

As I understand, you most likely have to build a pcb for the circuit.

As Brooks pointed out, these days a PCB fabrcation house
can do it at a lower cost than you can at home.
http://groups.google.com/group/comp...arebonespcb+Shanghai+PCBex+*-quality+futurlec

If your time has no value and you want to try it for the experience,
Mike Harrison's site has lots of DIY details:
http://66.102.7.104/search?q=cache:...stuff.co.uk/pcbs.html+Mike-Harrison+Copperset

There is a Yahoo group on this subject:
http://groups.yahoo.com/group/Homebrew_PCBs

The lowest cost DIY method is the glossy clay-covered paper method
described most thoroughly by Tom Gootee.
Short version:
http://groups.google.com/group/sci....ems+long-soaks+hot-water+180-*+toner-transfer
Long version (He has other cheap tricks here too):
http://www.fullnet.com/u/tomg/gooteepc.htm

 

Lower cost? Some copperclad, a sheet of Press-n-Peel or glossy paper
and a few cents worth of etchant and toner does not add up to $33+. I
could 'pay' myself twice the minimum wage and still be ahead.


Tim

 

As Brooks pointed out, these days a PCB fabrcation house

Sorry to hear you're only making twice minimum wage.

You're not factoring in that
the 1st time it's not "a few cents worth of etchant";
you have to buy an entire bag or a whole jug (+hazmat shipping ?).
You also didn't mention the basin.
(Most experienced folks would heat the solution;
Harrison rig shows a Pyrex dish.)

Surface prep, ironing, remove PnP, inspect,
correct flaws in toner deposition, mix etchant, heat etchant, etch[1],
drill for non-SMT items[2], remove toner, inspect, clear flaws,
dispose of (or re-bottle) etchant, clean up work area.

It may be a breeze the 2nd time;
the 1st time, it's a good chunk of a day consumed.
If you're using clay-covered paper, add even MORE time.

The 1st time is where you learn all the little tricks
and most folks don't want to do it themselves a 2nd time.
..
..
[1] If you don't have a mechanical agitation system,
you have to stay close by and do it manually.

[2]The whole *vias/PTHs* thing is a discussion all unto itself.

 

....

I don't recommend it but at Tektronix in 1989 I had zero budget,
available time, a 750 Mhz counter timer, and desperately needed to
demonstrate higher frequency applications with this.

So I begged a few 2.8Ghz divide by 4 parts from one of the vendors,
got a postage stamp sized bit of double sided clad circuit board,
got out the dremel tool, and went at it.

The divider was only 8 pins on .05" centers. I only needed a few
surface mount resistors for termination. And I only had a single
signal in and signal out peltola connector.

With a very calm morning, a very steady hand, a very tiny dremel
bit and a few scrapped boards the prototype was built and worked.

I don't recommend it, and I wouldn't even consider it on any of the
REALLY fine pitch parts I see today. But if I only had a few pins
I might try soldering short bits of wire wrap wire onto the pins
and connecting those to the destination. "bo bo" wires are always
fun.

Don't try this with 940 pins

 

How long does it take you to drill and connect through the holes?
What does it cost for a mini drill press that can handle 20 to 30
mil carbide bits? (Regular steel bits last through about a dozen
holes in PCB stock).

How much extra design time will you waste in adjusting your
design to minimize the number of holes drilled for components
and vias?

If you're in engineering and working for only twice the minimum
wage, you ought to consider a change in careers.

Mark Borgerson

 

-----------------------------------------

(Just for the sake of accuracy, and maybe more "completeness", for
people who might need to decide what the best way to go about getting a
pcb made is, in their particular situation

The etchant CAN be dirt cheap: I now use two parts common 3% Hydrogen
Peroxide plus one part common Muriatic Acid (i.e. 20-something%
Hydrochloric, IIRC), which are both very-widely available, in small
quantities, very cheaply. [Using a fresh batch, each time, it's less
than $0.25 per board; maybe way less.]

It etches a 1oz board at room temperature in about five minutes, with
hand agitation (i.e. a hand in a disposable latex glove, gently wiping
surfaces of PCB with a small wad of paper towel). [Less than $0.10 per
board.]

I etch in a cheap plastic food container from WalMart: [Less than $1.
Divide that by the number of boards etched in your lifetime.]

Copperclad, even pre-cut to specs, within .01", is extremely cheap,
from the right vendors. [Certainly less than $0.50 per board, for,
say, 4"x6" or 9"x3", 1 oz, 2-sided, .082"; and WAY less for
cutting/sizing to specs within +/- 1/16".]

Surface-prep requires an abrasive nylon pad (e,g, a "Scotchbrite" pad).
[Reusable for at least 20 boards: Less than $0.10 per board.]

I don't use "Press-n-Peel" (PnP). It's just one more thing that I
can't get locally within 10 minutes, and is kind of pricey, too. I run
glossy inkjet paper through an old laser printer. Even in very small
quantities, I think it's less than $0.10 per sheet. (And you can buy a
whole truckload of old laser printers for $50.) My old LaserJet 4's
toner is super-expensive, around $80 or more, for something like 4000
sheets printed. Assuming only one board's patterns are printed per
sheet, that's about $0.02 per board.

And I don't believe that using clay-coated (i.e. "glossy") paper would
take MORE time than PnP, even for 1st-timers. Not at all. And I think
that the results, using glossy paper, are AT LEAST as good as those
with PnP.

The TOTAL time required COULD be as short as 20 minutes from "pattern
on computer screen" to "start soldering components", for simple boards
(i.e. with very few drill holes; Surface mount would be even better.).

Drill bits: I use used, "re-pointed" carbide bits, purchased in
quantities of about 100 at a time for about $25.00. If my drill press
(and my technique) don't break too many, we can probably say way less
than $0.25 per board.

Well, I probably forgot to mention some of the replenishable materials'
costs. But you can probably see the trend and do the math.

For anyone who wants a prototype or one-off PCB, or even a small
production-type run, my method is obviously way less expensive, even
just in terms of out-of-pocket expenses, even (or maybe especially) the
first time.

Each person (or their company) would have to calculate their own time's
value, to them or their company, to see if a board-house makes more
sense, at that time, for their needs and their situation. HOWEVER, if
you need a pcb or two, VERY QUICKLY, e.g. if an engineer's or
technician's time is going to be more-or-less wasted until the board is
in hand, then you had better also take that cost into account. Delays
can be detrimental or costly in many other ways, too. So you might
also have to compare the cost of the time spent doing a toner transfer
pcb to the large "express turnaround" fee that a board house would sock
you with, and then still take a day or more to get your pcb to you,
plus the cost of that delay.

Regarding the "1st time" or startup costs, and the extra time required,
etc: Most of the costs were covered above, except, of course, for
things like driving to supplier locations (or ordering on-line), which
also takes some time (and money), as well, plus the "opportunity cost"
value of a person's time.

You are probably correct about investing a whole day (or at least a
half-day), the very first time you attempt to make a toner transfer
PCB. (And if someone is setting up to be able to make small
production-type runs, it might be longer yet.)

BUT, hey! How does that compare with the 1st-time costs, and cost of
time spent, when using a PCB-fab vendor?? Worst-case might be someone
who's never done either method. Are you going to take into account the
cost of the pcb-design software they would need, and the time to learn
it, to be able to generate the files required by the pcb-fab guys?
(Toner transfer files could be created with MS Paint; probably very
quickly for small or simple boards.) Even for those already owning,
and experienced with, pcb-design/layout software, just finding a pcb
fab house to try, reading their website, readying the files to send to
them, emailing them, getting a quote, arranging and sending payment,
etc, etc, would probably eat up a good half-day, and maybe more,
especially the first time. (And, for those people, the artwork for
toner transfer might already be done and ready to print.)

You are certainly right about plated-through holes, and vias, being a
HUGE problem, for all DIY pcb-making mehods. It uses up a lot of board
space, to have to make pads on both sides, for example, for each via.
And then a wire has to be soldered through them (or special eyelets
crimped in). And I do greatly-dislike not having plated-through holes.
If I use, for example, a DIP socket on the top side of a board, then I
basically HAVE to approach every one of its pins with a bottom-side
trace (or make a "via" pad nearby), or else they can't be connected by
soldering. And if board real-estate is getting tight, all of that can
be a REAL pain.

SO, it should be noted that, if making a choice between DIY pcb-making
and using a board house, then the PCB LAYOUT, itself, would have to be
compatible with DIY limitations, for DIY to even be an option. For
existing board layouts that were intended for professional manufacture,
DIY is not likely to be an option, without re-layout work, except for
the simplest of boards.

And, of course, most people limit DIY pcb-making to single or double
sided boards. SOME people do make "multi-layer" pcbs. But they
generally just use very thin pcb stock and then glue the layers
together. Seems like another "can of worms", and more "hoops to jump
through". Never tried it, although I have been very tempted, at times.

OK. Sorry to have blathered-on for so long, about all of that!

I hope I didn't sound too biased, and that I wasn't actually unfair or
inaccurate in my comparisons, et al.

The bottom line is that I think that, in general, there might be times
when DIY toner transfer pcb-making could make more sense than
out-sourcing, and wanted to try to shed some light on some more of the
factors that might be involved, for someone who might be faced with
choosing among the available methods for coming up with a printed
circuit board.

(Whomever mentioned my website, in this thread, thanks!)

Regards,

Tom Gootee

tomg at fullnet.com

http://www.fullnet.com/u/tomg

-------------------------------------------

 

About 30 minutes last board. That was 60-odd holes, 4 or 5 vias (which
took another couple of minutes to solder).

That was indeed an investment I ignored in my analysis, and one can
spend a good few hundred dollars to get a decent drill. My investment
specifically for electronics was $60 or so for a drill-press thingy
for my Dremel. My HSS bits last 50 or so holes through 1.6mm FR4.

Nice objective use of the word 'waste' there. But anyway: less than I
spend thinking about impedance, cross-talk and other layout criteria.
I've found that the components with leads need holes and those without
don't, so I don't spend much time optimizing that. But I do spend some
time optimizing the size of pads and tracks to fit the relatively
crude process. Vias that don't go to ground mean a slot in my ground
plane, so avoiding them would be a consideration no matter how the
board was to be made.

I'm not a professional engineer and that's not the context of this
thread. I'm an electronics hobbyist and twice minimum wage is
substantially less than I earn while at work, but it's substantially
more than I earn while not at work. DIY boards are a daft idea if
you're doing it professionally. If I did this for money, I'd have
someone else make the PCBs. In fact, if I was trying to minimize my
hobby costs I wouldn't do any electronics at all, but then I wouldn't
do anything in my spare time at all, because I wouldn't have any.
Spare time is uneconomic.


Tim

 

It sounds like you get paid over twice minimum wage in your spare
time! I wish I was that lucky. While I earn over twice minimum wage
while at work, I get paid nothing while I'm not.


Tim

 

it is a good idea to factor in that while you're waiting on the
prototype boards to arrive in the mail you're sitting there "getting
paid" to do nothing. sure, we all have other things to do and can
occupy our time writing invoices or something, but what we're really
doing is wasting time waiting for prototype boards so we can get back
to the "real work". the best reason to make boards is because you can
have them in an hour (or a few), and in terms of labor costs it's
cheaper to pay someone 200$us/hr for a few hours to make a 30$us board
than it is to pay someone 200$us/hr to sit around tinkering for a few
days waiting for boards to arrive.

 

as far as i am concerned tom said it all right here, at least in spirit
if not in word. the key reason to do your own pcb's is so you don't
have to wait around for them to be mailed. it would have probably been
cheaper to have pcb's made in the "old days" when everything was DIP
too ... but who would do that ? why didn't everyone order their pcb's
then ? it's not because they couldn't, or because nobody was around to
make pcb's. the reason was because there wasn't a market for them,
because it isn't hard to get a through hole prototype board and solder
some chips on it, or use wire wrap. the reason it isn't done now is
BECAUSE IT IS MORE DIFFICULT, and people don't want to go to the
trouble of etching their own surface mount boards.

OR ... you need multi-layer boards and it might not even be feasible to
do those yourself. that is a valid argument to make.

if there was some brilliant magical way to just print pcb's off of your
printer, i mean, without having to transfer the toner pattern to a
board, etch it, etc ... then of course everyone would do it themselves,
nobody would waste time sending their design off to a pcb shop and
waiting a few days for boards to come back, even if the boards cost 5
times as much to do yourself. nobody cares about saving money on pcb's
when they're prototyping something, everyone's time is more valuable
than that. what people who are sending board designs off to pcb shops
are really doing is saving TROUBLE, because it's hard to print the
design on a laser printer, iron it on to the board, mix up the acid,
get the gloves ... etc, etc, etc, people are basically lazy lol. and
lazy is fine, but let's not bullshit ourselves about it by hiding
behind the cost of the boards lol.

 

JeffM wrote:
[snip]

i used this method.

i think tom's page makes it sound even harder than it really is, and
even he makes it sound easy. i believe he was afraid that people would
not believe this worked so he went to great pains to be very careful
about how he laid out his directions to insure that the people using
the method would get the same results that he was getting. truth is, i
found i could cut corners even on his easy procedures and STILL get a
terrific result. the key thing about this method seems to be making
sure you don't rush the part with the iron, that is, insuring that you
really do pass the iron over every part of the printed image and really
pressing down on it hard and letting it heat up enough to transfer the
image. but some of the things he talks about such as the trouble of
getting the paper off, etc, i found to be very easy. for me just
putting the board under water when i was done was all it took, the
paper just came right off in one whole sheet, i didn't have to fight
with it at all.

this method is terrific. the only limitation i found with it so far is
that in the resolution of my printer! or maybe it's really the
resolution of my software. i can't seem to make traces small enough
for one of my surface mount chips, the pins are just too close
together. for 99% of chips i don't think this is a problem. my other
limitation is my own knowledge of pcb layout and my own experience
level, which is very inexperienced. more experience with the layout
software would probably even solve my problem with the resolution.

but inexperience or no, i have successfully etched boards with this
method and it worked great. just for fun i even printed a picture, a
photograph that is, in black and white, and transferred that to a
copper board and etched it, it turned out really pretty! lol.

the best thing about tom's method i think is the speed. you just print
the image, go out and prep your board and use the iron to transfer your
image over to the board, clean the paper off, etc, mix up some acid,
get your gloves, etch the board, and clean it up. i mean what could be
better than that ? it takes basically no time, is easy, cheap, etc,
you get your board done and get back to working. and if that board is
screwed up, fine, print another page and do it again. that's the
beauty of it ... the key word in the phrase "rapid prototyping" is ...
RAPID ... lol.