Reflow advice

Forum 2005-2010 (read only) Hardware Development
1

I jumped from the iron into the skillet and tried cooking up a PCB with SMDs, but it hasn't turned out that easy, and I'm suffering from information under/overload. I've watched the SparkFun video (like everyone else), and have checked out some of the YouTube offerings too. I hoping for a little advice from others that have tried making home cooked boards to clarify some of the finer points.

Background:
The PCB maker told me that the boards did not need to be stenciled and that I could simply place my parts and bake the board. Emboldened by a little knowledge, I decided to give it a try and hope for the best.

I bought a thermocouple amp from SparkFun, but it's output suddenly falls to 0v above 130C, and I've rebuilt the circuit twice (so simple it should be a no brainer) trying to figure out whats wrong (I suspect the chip), but the bottom line is I am at least temporarily abandoning a control relay for the skillet.

My multimeter came with a type K thermocouple which I never used till now, and I am very glad that it lets me read temperatures. I saw one video with the thermocouple on the skillet itself, others with thermocouple pads taped to the boards, infrared non-contact thermometer hand measurements, and plain old eyeball techniques to see if it looks melted enough. It seems like everyone has a different technique.
I decided to place my thermocouple on an unpopulated part of the board and measure the temp from there. The skillet I got has a glass lid with a hole in it that facilitates inserting a probe.

My first experiment was to see if I could fix an Ethernet shield I had that was missing a 805 resistor. I quickly learned that the header pins melted and lost alignment. Another pass with a piece of tile holding the pcb failed to flow the resistor.
Figuring that better heat contact was needed, and that a shield with header pins was impractical, I decided to try and repair an old Seeeduino board. After heating it up, I pried off what I suspected was a dead FTDI chip. Once the board cooled down, I noticed that the through-hole components had generally been pushed up since the board went down as the solder melted. The plastic on the headers deformed as well. Lesson #2, same as lesson #1: not advised to try re-reflowing boards with non-smd components.
I reheated the Seeeduino once again with a small piece of solder on top of the PCB to act as a visual indicator. The solder melted somewhere around 180C and I let the temp get up to 215 before taking the lid off, the solder hadn't really flowed onto the new leads, but a careful tap with some tweezers settled the chip down fully onto the pads. Afterwards the new chip seemed to work fine, but my board looked like R2D2 after a ride on the back of an X-Wing fighter.

So that was my warm up, trying to avoid trashing new prototypes, but now for the serious part

Take#1: Placed all parts on PCB, placed PCB on top of piece of metal placed in skillet to help spread heat evenly, set skillet temp to 275F, let thermostat kick off for about 30 seconds to let the hot spots disipate a little bit, then turned it all the way up and let the thermocouple hit 220C.
I pulled the lid and checked the board. The heat certainly makes it hard to get a close look. My components showed no signs of fillets from solder flow. I carefully tapped several components hoping to seat them in the thin layer of tin on the PCB. When the board cooled down, only one resistor stuck to its pad, and it was easily rubbed off.
It did seem that the skillet temperature rose close enough to 1C/sec that I didn't need to worry too much about not matching a "profile." Then again the lack of fresh solder paste probably makes all that irrelevant as well.

Take #2: I have one leadless LGA component with incredibly tiny pads. Well aware that too much solder would cause a bridge, I carefully used an iron to apply solder to each pad and then used a wick to take the extra off. I hand soldered all the other smd components. Board is heated in skillet as before to 220. This time I could see the hand soldered components flow. I tapped the LGA chip to help seat it, then let things cool.
Inspection reveals that the chip is slightly off center, and it quickly pops off when touched. The pads reveal that only a couple had any solder connection.

Take #3: I add solder to the pads again, but am careful to leave the tiniest of pillows. This time the thermocouple tip moves and drags an 805 resistor all over. I try to focus on the chip and clean up the resistor afterwards. The chip ends up slighty off center again.

Take #4: I heat the board up and after laying some towel over the edge of the skillet so I don't burn my arm, I try to slide the chip over slightly. Next thing I know I've accidentally flipped it over, in addition to cooking my hand and eyes.

Take #!@%5: I've tossed everything else and just put the board in the skillet. Hand held thermocouple is nearing 200C when hot spots begin making the resin at the unpopulated end of the board boil out and sizzle. Mmm, tasty. Slight darkening of board seems harmless, and this time it seems good enough that I'm willing to add thru-hole components and test it.

Prepping a second board by using a tip to add solder to the LGA pads I accidentally lifted one of the pads, fortunately a N/C, so that may not be such a great technique.

Questions (under the context of I'm only making a couple boards right now):

  • Does this work for anyone else without adding solder (like I was told it would)?
  • Should I use a solder paste syringe instead of hand solder?
  • Should I use flux to help the solder flow? Liquid or paste?
  • Should I get some high temp tape for the thermocouple, get an infra-red thermometer, or just stick with what I've got?

Thanks!

2

In the paste I have used a skillet but only to remove SMD components. You do have to be careful as the heat coming off the skillet can almost burn your face. I had my thermostat on 375F to melt the solder and remove the chip. I think 220C is too hot; 190 should be sufficient but only if the complete board touches the base of the skillet i.e. no thru-hole parts. I now have an inexpensive hot air rework station which is still hot but much more controllable.

You definitely need extra solder of some kind to solder parts to a board. As you have found there is simply not enough solder on a tinned board. You should be able to do it by either putting solder either on the board or on the pads of a SMD chip. You should use plenty of flux when adding solder to a board. I use a water soluble flux pen and then ultrasonically clean the boards. Add the solder quickly with a stroke of the iron across multiple pads. This is the technique I use for a QFN chip before applying heat with the hot air station. So far every chip has worked fine with no rework. I only lifted N/C pads on 1 board out of 25.

I currently only use paste for a Hirose connector that is difficult to solder without it and then heat the paste with a soldering iron. I think the hot air station will melt the plastic on the connector. There is definitely an art to putting paste on a board without a stencil. You need a very fine nozzle and a way to quickly stop the flow for a given pad. The hardest seems to be ICs. In my previous attempts I put on too much paste and then cannot see where to position the chip as all the pads are covered. Now I have a hot air station this is one of things I going to play with next.

I would say to ditch the skillet and get a hot air station - it makes life so much easier. Buy some paste but treat it well as you can easily waste $30 by keeping out too long. My first tube got all dried out. My second tube is still going 18 months later.

3

Thank you for a very enjoyable read and I hope you won't take my laughter the wrong way. ;D

To answer your questions:

  • Yes, you need solder
  • Yes, you should use a syringe
  • No, you should not use flux if you use solder paste as it has flux already in it
  • Yes, you should use high-temp tape (e.g., Kapton) and make sure the thermocouple is making contact with the board. It is not enough for it to do so when the board is cool -- as it heats up the tape "relaxes" and the thermocouple can easily lift up off the board. You really have to try hard to keep that thing on the board at all times.
  • Yes, you should ditch the toaster oven and get a hot-air rework station and/or batch reflow oven :slight_smile:
4

The right way to do it is with solder paste, which you can get for a good price from Zephtronics (www.zeph.com). Their paste is formulated for proto work, so it stores well at room temperature and doesn't need to be shipped overnight on ice or the other rot you have to go through getting paste from the usual suspects.

Use a syringe to apply it -- you need surprisingly little.

The only parts you don't have to add solder paste for are BGAs, because they have solder balls already there. You just need to add flux. Zephtronics sells specially formulated flux for the purpose.

I just put a heavy washer on top of the TC to hold it in place. However, I mostly use that as a reference -- I watch the color and texture of the paste as the flux starts to boil and the solder melts.

The trick tool for this kind of work is a hot air rework station. I've seen them for as little as $175. Basically, it's a precision heat gun for working with SMD and solder paste. Makes soldering on even the smallest leadless SMD parts dead easy even for a complete klutz like me.

5

Well I was well aware of just how stupid some of my questions might be, and the fact that it all gets immortalized by posting them, but I was desperate for answers so thanks for laughing.

I'm getting over the shock that I was able to hand solder a C0402 capacitor that was smaller than my solder diameter, what at amazingly small component. I skillet cooked a second pcb yesterday that went better. I basically hand soldered everything possible and then cooked it to let things settle down. It still needed touch up because I hit a couple of resistors with the hand held (doh!) thermocouple. I realize the leadless package would work better if I had some flux.
So far I've been experimenting on a couple shield pcbs with only a few components. That way it costs less than $20 to trash a board. I've got a more ambitious board that will cost about $70 per experiment.
I've now got two copies of a prototype board, using a chip I've never used and have no code for, fabricated with dubious technique, and I've got to find out if they work! ...and it's all westfw's fault for suggesting I make a board with the MMA7455L ;D Just kidding, but this is quite a long process. Six months ago though, I'd never worked with embedded code and pcbs so I'm having a lot of fun.

I see a hot air rework stations on ebay for $100 with shipping, any experience with these?
I've been using a plate to put the pcb and smds on while I use tweezers to place the parts. Is it ok (from a soldering perspective) to do the hot air work on the plate to help avoid jiggling parts out of place?

6

you can try by manually applying solder onto the smt pads and then use one of those hot air rework station and manually insert the smt parts using a twizzer. it definitely beats using the oven unless you are dealing with lots of smt parts.

if you need to use the oven, get a aluminium solderpaste stencil made and then evenly apply the solderpaste onto the pcb. the solderpaste will be able to hold down the parts when they reflow in the oven.

7

I've been going through all of my old kit looking for stuff to salvage, thought i'd have a practice with an iron lifting smd parts, mixed but encouraging results so far, broke a leg on the first set of pins I tried lift being too aggressive too early with a blade, other side came of perfect and clean, I wanted to see what I could actually do with some of the parts once I salvage them (some useable sized eeproms amongst other things) attaching bellwire is out of the question really isn't it? :slight_smile:

Glad I came across this post as I had read through the sparkfun tutorials but don't have the kit yet to mess around with skillets etc. I did find this UK site http://buyaoyue.com/index.php?main_page=index&cPath=4 if anyone else over here is looking, they look reasonably priced but are any of them any good or to what level parts are they useable? For £50-75 ($80-120) I'm hoping they'll be good enough for what I want, I noticed the Sirocco Series 6028 is ideal for working around plastic components. I guess it also depends how many boards you need to make?

I've got loads of old disposable or broken kit, laptops, routers PCI cards, phones etc, with re-useable memory, crystals, LCDs, IC's etc. all over it, no point filling landfill or someone else reclaiming the parts for free. + I have a birthday coming up, this looks like a reasonable unit
http://buyaoyue.com/index.php?main_page=product_info&cPath=3&products_id=14
or are things like the vacuum sucker a gimmick?

I've just been given a PSP-1004 with a broken LCD, as I already have an original PSP 1000 I'd like to hardware mod the 1004 which is going to have some small parts to work on, like swapping out the UMD drive for flash of some kind. The screen will be relatively easy to replace but it also has serial I/O on the headphone remote socket so could possibly be a portable platform for developing/communicating with the arduino, wirelessly. I'll stop rambling about.....now