SMD reflow with a blowtorch (2016) | Hacker News Reader

SMD reflow with a blowtorch (2016)

lab.whitequark.org

49 pointsluu20 comments

Lovely hack :) I used blowtorches in my poor days to rapidly desolder components. Take a bucket, fill it with water to about 3/4 full, hang your PCB upside down over the bucket. Heat evenly with blowtorch. Part will desolder themselves, gravity pulls them down as the solder melts and when they're free they drop in the bucket for instant cooling. You can get quite aggressive like that without damaging parts. This trick works for both SMD as well as hole through components, but not when the legs have been folded over.

In no time you'll have a good pile of parts for experimenting or building new stuff, you can even de-solder very large ICs using this.

Well, if you do not really want to damage parts, you better not cool them too rapidly (in the water), better leave them to cool slowly.

The heat is mostly still in the connectors at that point which are all metal. That won't suffer from the shock of cooling down whereas the residual heat in the pins could easily damage a sensitive chip if not cooled.

I used to whack PCBs onto the table to get things off (after heating something with some shitty underpowered soldering iron.) Just because I didn't have proper tweezers, and the solder would quickly solidify again...

People looked at me with serious "WTF?" faces whenever I soldered somewhere away from home ;D

Still do whack the soldering iron itself to get excess solder off. Obviously not with the tip but the handle, so the solder just continues onto the mat and makes a nice small splat. (No it doesn't fly off, it quickly solidifies on landing due to the splat increasing the contact area for heat transmission. Doesn't even damage the mat...)

You just gotta know how to help yourself with the tools available!

I do the same with the iron, but first i melt a bit of fresh solder on.

Mitch Altman taught me that trick! I use it all the time now.

I went up to him to ask for some desoldering braid at a conference, and he showed it to me. Life-changing! I don’t know why I didn’t think to do it myself.

Wouldn't contact with water make them unsuitable for future soldering? (some ICs are shipped in airtight bags with moisture indicators. or they are just overcautious?)

Whitequark's lab notebook [1] is sheer pleasure to read through (if you're int that sort of things, obviously).

[1] https://lab.whitequark.org/

(Don't mean this to sound negative)

I guess you make use of what you have, but if you are at the level of doing smd reflow, there are significantly better tools at a low cost (hobbyist-level too).

For smd rework I'd say better to use a cheap hot air station. Plenty of cheap ones that work ok, at least much better than such an imprecise tool as a blowtorch.

For prototyping, you can buy kits with a toaster oven + controller that works pretty well for reflow. You plug the power cable of the oven to the controller, and the controller switch it on and off according to the temp profile. Works pretty well, we use one at work for small runs.

It's cool to know that it's 'doable' even though I guess it require acrobatic levels of control and high charring rates :)

You kind of missed the point here, which was that it is surprisingly doable using a ghetto-ass blowtorch when the correct tools aren't available, which is counter-intuitive. I would have expected this to go spectacularly wrong and leave a charred PCB, which it most certainly didn't ("When holding the tip of the flame ~15cm away from the PCB, it took over a minute to get the solder to melt and wet the plane.").

One hallmark of the 'true hacker' is that they get the job done, no matter what the tools at hand (or even no tools, in some cases). This post is one of the very few posts on the homepage right now that to me makes HN worth reading. Using a brute force tool to do something extremely delicate.

> if you are at the level of doing smd reflow, there are significantly better tools

What do you mean by "at the level of doing reflow"? Techniques like this demonstrate that "the level" of doing reflow is much more attainable by resource-constrained hackers than people sometimes think.

Many years ago, when my personal time-vs-money tradeoff was different, I made a number of surface mount PCBs by doing reflow over an ordinary candle or paraffin lamp. I would balance the PCB on a piece of metal shim to act as a heat spreader and keep the board from getting sooty. It worked really quite well up to a couple square inches — the trick was learning how to follow a preheat profile to keep the small passives from tombstoning.

This is remarkably similar to the DIY version of a high-end soldering technique I've just learned about called Vapor Phase Soldering.

I'm waiting for a bottle of Galden LS230 to arrive so that I can try this DIY method on some whacking great big Xilinx BGAs: https://m.youtube.com/watch?v=czwRntpEzgg

> Something to note is that torch exhaust contains a substantial amount of water vapor, which readily condenses on cold areas of the PCB.

Reflow + wet PCB/parks === no go. If they will have to redo that in an hour or so, they will get popcorn on PCB or components.

Can you elaborate? I never heard the term popcorn used in that way.

https://en.wikipedia.org/wiki/Moisture_sensitivity_level

If you order parts from a major distributor (arrow, avnet, digikey, mouser, etc.), everything comes with appropriate humidity indicators where needed. Pretty much anything that is something + plastic around it will be affected to some extent.

I haven't seen things go "popcorn", but broken off edges and cracks across a package... yeah.

Add: that said, moisture from a blowtorch probably won't have such an effect; you need to get it nice and deep into the package and unable to come back out.

Add(2): this is also why ziplock bags make good outer semiconductor packaging. Make sure you properly zip them back up whenever possible.

Add(3): a quick look through my semiconductor stack (SMT ICs, mostly QFN, some SOP, no BGAs) is about 30% MSL3, 45% MSL2 and 25% MSL1 or unmarked.

This is great to know! One question I have is what the actual technique the author used is (ie distance and time), because they mention that it took a minute for solder to melt at 15cm and I'm not sure if that was what worked best in the end.

I thought that was Jeri Ellsworth. Thinks things through from first principles. But no, it's a another red-haired woman who thinks like Feynman.

> it's a another red-haired woman

Why is it significant that this post was from a woman or that she has red hair? This comment sounds like a backhanded compliment, at best.

Reply on news.ycombinator.com