Surface mount prototype work - help!

[Gizmo]

Ok, well, I've confirmed it's my crap reflow work causing the problems. I just built up a board by placing the larger 0.8mm pitch components on the solder paste along with all the passives then reflowed all of that including the small pitch component pads unpopulated.

Once reflowed I then placed the smaller pitch components onto the solid solder and reflowed that. Works fine now! So better placement and making sure the solder paste is neatly applied seems to be the key.

[Gizmo]

Yikes!  Definitely add weak pull-ups (10-100k?) so all enables reach a defined state!

It's always good practice to assign pin states, even with a weak pull up/down, because a floating CMOS input will permit erratic behavior and unusual current consumption.  When the pin voltage is in the transition region between V_IH(min) and V_IL(max), the P/N input transistors are both conducting, and drawing excess supply current; the voltage can drift into this range due to input protection diode leakage, or ambient fields (electrostatic, or RFI, including merely switching nearby pins).  Unused MCU pins, if not pulled with physical resistors, should at least be assigned either as weak pull-ups or active outputs in software.

Tim

Thanks, I will do in future.

I already tie all unused inputs to a known state as I was aware of that problem, but I didn't think to pull inputs which can be left floating during programming / reset time to a known state. I realise now this is just asking for trouble...

[SirNick]

I'm fairly new to SMT myself, but I'm having amazingly good luck.  Here are a few things I've found that help, FWIW:

- If you're doing anything smaller than 0805 and huge pin-to-pin spacing on ICs, get a stencil.  OSH Stencils and the like are making them for peanuts now.  Take advantage of that.

- Make sure the stencil is FLAT on the board.  This one is a killer for tiny packages.  I have some 4mm x 4mm, 10-pin dual-output linear regulators with a large ground pad on the bottom -- these things were a bit of a challenge, since the pads need to have some solder paste directly on them (meaning, good stencil registration), but not too much (meaning tight coupling of the board and stencil, and adequate control over paste application.)  Tape your board down, tape your stencil to the board, and make sure it's stretched out enough to remove any waviness.  If your paste is smearing, you should work on your application process.

- If you have solder mask, a stencil, and good paste application, you won't get bridges.  If you still get bridges, work on fixing those three things first.

- Using a hot air station, warm your whole board at ~100C for a bit, then ramp it up to operating temperature.  Your parts will align nicely if all sides are able to reflow around the same time.  Placement isn't nearly as critical then.  When you get the hang of this, it's so quick and easy with a good air wand that I don't see any reason why anyone would ever want to go back to a soldering iron.

- If you can develop your board in such a way that you can assemble it in pieces, that's best for the reasons already given -- particularly, being able to test the PSU section before populating the rest.  I have cut stencils into "areas" and built those areas one-by-one, but it takes some thought during layout to make that possible.  On dense boards, it's often all or nothing, since you don't want to thermally-stress neighboring stuff too much, or risk unintentional (potentially botched) reflowing.

As for your circuit design, always think about what happens before the entire thing is up and functional.  Think about what happens while the PSU is ramping up; if the PSU is under-volted; if the MCU locks up; if the MCU hasn't initialized; etc. etc.  Make sure the default state, before the whole thing is under software control, is safe and sane.  Pull-ups / downs are cheap insurance.  I often take all the IC reset pins and bring them to a central pull-up with MCU override, so they're all held in reset until the software is ready to bring them online.  If the MCU has brown-out protection, this helps to make sure they're not allowed to initialize (or stay active) under less-than-optimal PSU conditions also.

[Emil]

@rx8pilot
Where do you get the small black (presumably anti-static) trays in your first picture? What are they called?

[bronson]

rx8pilot, by stencil printer you mean one of the $200-$350 frames on ebay?  Buy a stencil, lock it in the printer, tape down your PCB guides, then use a squeegee to drag paste over the stencil?

Or do you have a more automatic setup?

[mark03]

Don't give up on 0.5-mm pitch parts by hand with an iron ("drag soldering").  Once you get the hang of it, it's not so bad.  I've even done QFN this way in a pinch.  Extra flux is key.

I wanted to comment on board bring-up strategy.  Whenever I am prototyping a new design, I plan ahead and assemble, then test, in stages so as to have the least number of parts stuck on the board if something goes wrong.  So e.g. on an FPGA dev board with a multi-rail switching power supply, I put together the power supply first and made sure it was working.  Then the auxiliary chips---some I2C stuff, and so on.  And finally the FPGA last, when everything else was verified working.  By planning this out ahead you can minimize the pain when you uncover that inevitable mistake.  Most of my rev A boards have at least one fly-wire to fix an error, no matter how careful I was in design and layout.

Obviously, this takes a lot more time than just stenciling paste and shoving it in the oven.  It really depends on the complexity of the design, and my confidence level, whether I go that route or the piece-by-piece route.  After a while you get a feel for it.

[rx8pilot]

@rx8pilot
Where do you get the small black (presumably anti-static) trays in your first picture? What are they called?

Those little guys came from All Electronics in Van Nuys, California. They are non-catalog random pieces they had in the stor. Last I looked, they still had tons of them at about $.10/ea USD.

rx8pilot, by stencil printer you mean one of the $200-$350 frames on ebay?  Buy a stencil, lock it in the printer, tape down your PCB guides, then use a squeegee to drag paste over the stencil?

Or do you have a more automatic setup?

I started with the MANNCORP RT2100 which is the more expensive of the cheap  starter printers. They call it "entry level" and they are not kidding. I ended up custom machining some magnetic PCB holders that make it WAY easier to use. The PCB holders that it comes with are true comedy - useless. The way it is setup now only take about 2-5min to get it aligned for .4mm devices. I get the framed stencils from Stencils Unlimited which are good quality/price/speed. You could get the same result with some of the lower cost printers as long as you are will to fiddle and potentially make some customizations.

Don't give up on 0.5-mm pitch parts by hand with an iron ("drag soldering").  Once you get the hang of it, it's not so bad.  I've even done QFN this way in a pinch.  Extra flux is key.

I wanted to comment on board bring-up strategy.  Whenever I am prototyping a new design, I plan ahead and assemble, then test, in stages so as to have the least number of parts stuck on the board if something goes wrong.  So e.g. on an FPGA dev board with a multi-rail switching power supply, I put together the power supply first and made sure it was working.  Then the auxiliary chips---some I2C stuff, and so on.  And finally the FPGA last, when everything else was verified working.  By planning this out ahead you can minimize the pain when you uncover that inevitable mistake.  Most of my rev A boards have at least one fly-wire to fix an error, no matter how careful I was in design and layout.

Obviously, this takes a lot more time than just stenciling paste and shoving it in the oven.  It really depends on the complexity of the design, and my confidence level, whether I go that route or the piece-by-piece route.  After a while you get a feel for it.

Great advice - limit your losses. And yes, it takes patience and practice for sure. Iron assembly is a critical skill to have even if you do mostly paste/oven assembly. I regularly have to replace components on a PCB that is already done. Having solid hand soldering skills becomes essential at that point.

[Zero999]

If it's just a prototype why bother with SMT, unless you have no choice?

Stick to through hole for most parts and goo old strip board. If it's only available in SMT, then you can get SMT to DIL converter boards pretty cheaply off ebay or from RS/Farnell for a ridicules price.

Having said that SMT capacitors and resistors occasionally come in handy for soldering between tracks on strip board. 

[Howardlong]

Depending on the project, if it's a one-off with under about 100 parts, I hand solder except for the QFNs which I use a hot air iron on after tinning the pads and finish off with an iron. For BGAs, it has to be reflow in an oven. I have managed to do a low pin count BGA with a hot air iron, but you need to have a heated board, expect failure, and have patience.

More than 100 parts, or when doing a panelised set, I use a stencil and solder paste, and reflow the lot. My maximum rate of manual parts placement in a single day is 3,600 when doing panelised boards, 120 parts per board, single sided. I've only done that once. More typically a rate of 1,500 per day is sustainable of you want to maintain your sanity.

I don't use a vacuum tool because it just doesn't work with 0402, my standard passive size, in my experience. If you use bigger parts, you may find a vaccum tool more useful. My placements are generally very tight too. I use the 45 degree tweezers for everything. I can get away with an illuminated magnifier when placing on paste, but when hand soldering I do it with a stereo microscope. The benefit of using an illuminated magnifier is the increased field of view.

A full visual inspection after reflow is essential, there are often errors like tombstoning, so you need to be reasonably adept with your hand soldering skills to correct these anyway.

For SMD hand soldering, I have four irons; a 0.25mm tip, a 2.4mm chisel tip, a tweezer iron and a hot air iron. Trying to fanny about changing tips is a PITA, having the right irons ready is essential. For through hole I also have a hot air pencil and a desoldering iron, but I rarely use them, as I do almost no through hole stuff.

Expect failure, and be patient. You will get it. It's an experience thing.

It takes ten or fifteen minutes for my body to settle in to a placing session, once relaxed and in the grove, placement is much easier.

One final key point. I limit my coffee intake to one cup before I start.

[janoc]

I've tried it, I make a COMPLETE mess of it. Not a chance. 
I'll sort out my paste and reflow issues.

I tried that multiple times also on TSSOPs with limited success mostly a few bridges left.

Bridges are normal and no problem when drag soldering. Simply get the solder wick out and suck them off once done. Problem solved.

If it is only for prototypes, then I wouldn't bother with stencils and paste. Stencils are expensive for one-offs and keeping paste for the odd bit here and there is a pain - it needs to be kept refrigerated and has a short shelf life. I can hand-solder most of the things, even QFNs - put a tiny amount of the solder on each pin and also on the pads, then place the component and reflow using hot air and flux. It takes a bit of practice, but it is certainly doable.