Surface Mount Soldering

[dlh]

There was some discussion of SMT soldering buried in a ZX-40 forum discussion of 64KB EEPROMs. I thought it might be better discussed here. (Perhaps Don can move or duplicate some of the other posts.)

There's a company in Las Vegas that will make low cost, laser-cut MYLAR stencils that are suitable for low volume SMT solder paste application.

• www.SMTstencil.com

[DocJC]

I love the ZX series, but for some small tasks it is overkill, particularly if one is spinning off some small tasks for sub-processors. My other experience is with Basic Stamps, which seem incredibly slow and expensive after looking at PICs and Atmel's micros. I purchased an AVR Butterfly board, (ATMega169), ~$21, to see what it takes to program one, breadboard one, etc. I'm using Basic, as I'm not a C programmer...

Looking ahead, I see the chip has pin spacings on 0.5 mm centers, quite small even when I use a loupe.

I have seen several discussions on manually soldering SMT components, hear and on SparkFun, etc.

Is it reasonable to manually solder a 64 pin TQFP package? Is it possible with a "normal" soldering iron, or is a hot air gun needed? Is this impractical and hence an electric skillet or hot air oven is needed?

I understand that still using through holed components carbon dates me with dinosaurs, but I'd be interested in hearing some thoughts.
If this is too far off based for the ZX forum Don can delete it.
JC

[dlh]

Is it reasonable to manually solder a 64 pin TQFP package? Is it possible with a "normal" soldering iron, or is a hot air gun needed? Is this impractical and hence an electric skillet or hot air oven is needed?

It's certainly possible. I think one of the first tutorials on SMT from the Seattle Robotics Club showed manual soldering (and removal) of this or a similar chip. I think it's impractical and borders on unreasonable, especially for more than one off production. I find a toaster oven works quite well as long as the pins are exposed so you can see and fix any solder bridges. The toaster oven and a low-cost laser-cut mylar stencil work very well for small production runs - see Spark Fun's tutorial on this. I do not like DFN or other packages where the connections are beneath the package and I've read that these are a problem even for board houses that specialize in SMT. (I think Screaming Circuits has a whitepaper on this.) For any kind of production, I would have the boards made and populated in China. For resistors & capacitors, it adds very little cost aside from the one time charge for the stainless steel stencil. I'm approaching 70 (as slowly as possible) so old dogs can learn new tricks. I have a serious spinal cord injury at the base of the neck and flexing my neck and shoulders takes a toll. I actually find it easier to do SMT using a toaster oven than to manually solder through-hole components. Consequently, I try to use SMT where possible and affordable (some SMT components cost far too much).

This may just be personal preference but I think designing boards with an automatic ground plane where most of the copper stays on the board, helps keep SMT parts aligned in the oven. I've attached a picture of such a board design.

[dkinzer]

Is it reasonable to manually solder a 64 pin TQFP package? Is it possible with a "normal" soldering iron, or is a hot air gun needed? Is this impractical and hence an electric skillet or hot air oven is needed?

As with most things, it is easier with the right tools and there is a learning curve. You'll definitely need a low wattage iron with a small tip. If you've been doing through-hole soldering for years it is likely that your iron will be too big to be practical. The iron that I use is a 15 watt Antex with a 3/64" tip and even that seems too large at times, especially with 0.5 mm pitch since that tip width spans almost three pins.

If you search the Internet, you'll find several good articles on manual SMT techniques. You'd probably be well advised to start with some larger devices, 0805 or larger and 0.65mm pitch or larger, to develop your skills and technique before tackling the smaller parts.

[DocJC]

Every time I look at the chip I think I must be crazy for even contemplating manually soldering one...

Never the less I am inspired by your comments, and encouragement.

Fortunately my projects are generally prototype / proof of concept, or gadgets for myself, and hence the number of boards is minimal.

I'll give it a try with my next design.

JC

[mikep]

I purchased an AVR Butterfly board, (ATMega169), ~$21, to see what it takes to program one, breadboard one, etc. I'm using Basic, as I'm not a C programmer...

It sounds like you are using Bascom. This has some limitations I believe unless you buy the full license (~$150). You may find that learning C is not that much harder and then you can utilize the free tools such as the AVR GCC toolchain and associated libraries.

If you are looking for an integrated breadboardable solution, then you might want to look at the Oak Micros om128. This integrates a RS-232 converter, voltage regulator, port monitoring LEDs, and optional external RAM with a ATmega128 on a compact 2 x 0.9" circuit board that can be breadboarded. See my Oak Micros website for more details (end of shameless plug).

Is it reasonable to manually solder a 64 pin TQFP package?

The 100-pin AVRs use 0.5mm centers but the lower pin count ATmega128, ATMega644 use 0.8mm spacing which is eminently solderable. In fact I like the drag technique best. This involves soldering two corners of the chip and then using some liquid flux and dragging solder and the tip over multiple pins. This technique does not require an ultra-fine tip. With practice you can avoid solder bridges and solder a 64-pin TQFP package is less than a minute. The smallest I have soldered so far is 0.65mm but now I feel confident that I could do 0.5mm as well. Using a PCB with solder masks helps tremendously with the solder flow and eliminating bridges.

I bought a $35 themostatically controlled iron and extra tips several years ago and it has been great. This iron is a cheap clone of a Hakko 936 iron that retails at $94. Tips are around $5-6 and I keep several different sizes around depending on the soldering task (SMT, thru-hole or desoldering). I can buy tips in my local Fry's Electronics which saves on minimum orders and shipping.

[dkinzer]

Every time I look at the chip I think I must be crazy for even contemplating manually soldering one...

Yes, it is a daunting prospect but if you have a modicum of manual dexterity, a reasonably steady hand and ample tenacity, you'll master it.

If you know others that have access to SMT boards, you might see if you can get some scrap boards to try your hand. If nothing else, you might post a message on the Electronics 101 and/or ExpressPCB message boards soliciting unneeded and scrap boards for practice work.

[dlh]

Another tip - use solder paste rather than wire solder. It frees a hand, making it easier to hold the chip in place while touching an iron to a pin. And, like Don, I use a 15W iron when manually soldering SMT components.

[DocJC]

A new iron with a small tip, some solder paste, and some dinky small parts... I guess I know what I'll be getting for Christmas this year!

Mike, I've been following your product line, and like Don's, I've been very impressed. For my current endeavor, however, the goal was to see what it takes to use the small micros, (Pic or Atmel, Atmel chosen), as a pre-processor, essentially making smart sensors, which then decreases the workload on the main processor. The digital equivalent of replacing the 555 timer chip. I certainly like breadboardable prototyping, but the om128 blows the budget for this task.

I almost took the plunge and learned C about the time Don came out with the ZX-24, programmable in Basic, saving me yet again! Never-the-less, every time I read Circuit Cellar or Nuts & Volts it is readily apparent that the industry is C based. I still have Teach yourself Visual Basic.net and Java books collecting dust... If only there were more hours in the day!

Venturing into new areas, with new learning curves, is both challenging and rewarding. It is comforting, however, to have guidance from those that have blazed the trail before!

JC

[stevech]

I'd like to find a source for some "practice" or "training" SMT boards and components. So I can make a lot of mistakes at little expense in perfecting how to use solder paste. It really seems daunting to be able to hand-solder things like 1 mil and less, e.g., page 7 of

http://www.meshnetics.com/ZigBit_OEM_Mo ... asheet.pdf

I have some of these thumbnail sized modules with the SMD connections so small I can't see the insulating gaps without magnification!

But- a fellow where I work says he does it now, after lots of practice and refining his techniques. And with a good microscope and itsy-bitsy soldering iron tip and the right iron.

[GTBecker]

I'd like to find a source for some "practice" or "training" SMT boards and components...

Easy. Disassemble some old PCI boards or mobo, and old radar detector, digital camera... and try to remount some of the components.

It's great practice and - if you're successful - you'll gain a few usable parts, too.

I think you mean 1mm, not 1mil, right?

[Don_Kirby]

It really seems daunting to be able to hand-solder things like 1 mil and less...

Surface tension is your friend. The use of flux, and the right amount of heat (however it's applied) allows you to solder things you can't even see. Solder mask is a must though.

Personally, I find it much easier to solder fine pitch QFP packages than to solder 0806, 0603, and 0402 parts. Probably due to the fact that when they go flying out of the tweezers, I can actually find the bigger parts on the carpet...

-Don

[stevech]

yes - 1mm, not mil.

Thanks for the idea of dismantling SMT junk. I'll look into that. I have a collection of stuff too good to throw away.

[mikep]

Another tip - use solder paste rather than wire solder. It frees a hand, making it easier to hold the chip in place while touching an iron to a pin.

I think paste has its own problems too. I find I always put too much on a pad (i.e. hard to control) and it has a shelf-life especially if you leave it on the bench for extended periods.

I prefer water-soluble solder for most tasks. Some people recommend 15mil but I'm getting by with 20 mil. For soldering components (and I use 0603), solder one end and then the other - see SMD online tutorials for the details. Apply extra liquid flux as required. This method works quite well and doesn't require a "third" hand. I only use solder paste for components where it is hard to simultaneously reach the pin and pad (i.e. SOJ RAM chip and Hirose plug/socket).

[dlh]

I think paste has its own problems too. I find I always put too much on a pad (i.e. hard to control) and it has a shelf-life especially if you leave it on the bench for extended periods.

I haven't found that to be much of a problem. I use hypodermics and change needle size to control how much is dispensed. The biggest problem is cleaning needles afterwards so I don't clean them. I push a piece of small diameter wire (resistor lead) in it to clean it and act as a plug and store everything in the refrigerator. Then I just pull the wire out for the next session. There are formulations that have six-month life at room temperature - longer when refrigerated.

While a bit more than most will want to spend, this prototype stencil package is a fairly good value considering what's included.

• http://www.stencilsunlimited.com/produc ... ucts_id=28

The accompanying articles are worth reading whether or not you buy their stencils.

Sparkfun has a tutorial and links to a cheaper mylar stencil from a company in Las Vegas.

• http://www.smtstencil.com/

For one-off prototypes there are also adapter boards which convert SMT chips to DIP. They don't eliminate the need to solder the SMT chips but do make breadboarding possible.