3 ms·
> One, fumes. It's trivial to experimentally demonstrate that solder fume contains almost no lead, the quantity is negligible. Claiming the contrary is the ele
by segfaultbuserr 3y ago
> One, fumes.
It's trivial to experimentally demonstrate that solder fume contains almost no lead, the quantity is negligible. Claiming the contrary is the electronics equivalent of saying HTML is a programming language. Please don't do that again. The fume is indeed toxic, but it's due to the VOCs from the flux core, not the lead in the alloy.
A more solid (no pun intended) argument can be the hazards of debris. Furthermore, in my opinion, a newer and more serious problem of leaded solder today, in a workshop setting, is its use in solder paste. Solder paste and a reflow oven are required for prototyping any circuit boards with surface-mount components (SMT) - basically any modern circuit board today. Solder paste is a tube of toothpaste-like chemical mixture that contains tiny, micrometer-sized metal particles, mixed with sticky flux. If they're used without care, a solder paste spill is a sure way to contaminate the floor or work surface of your workspace. The sticky paste is also hard to wash away from skin.
Unfortunately, reflow soldering of surface-mount components can be really challenging, even more so when doing it by hand. Thus, classic lead-tin alloy is often used to reduce difficulties of assembly during workshop prototyping due to its technically superior properties. Switching to lead-free is only possible when you have a tightly-controlled and consistent work flow.
If you want lead-free, for small-scale prototyping and rework, a non-toxic bismuth-tin alloy is sometimes a good alternative to standard SAC305 lead-free solder thanks to its low melting temperature, which is one main reason that makes most lead-free alloys difficult to use (it even has considerable popularity in mass production of LED devices, as they are heat-sensitive). But its surface tension is slightly different, weakening the self-alignment effect of components during reflow soldering, increasing the chance of defective joints - a concern in prototyping. Its brittle nature also increases failure rates in the field, among other caveats.
Lead is really a gift from the devil.
- petertodd 3y ago> Solder paste and a reflow oven are required for prototyping any circuit boards with surface-mount components (SMT) - basically any modern circuit board today. This is incorrect. You can easily do small scale work with almost all SMT components without solder paste. Solder paste is required for automated assembly processes. But almost anything done by hand can also be done with conventional solder. Source: I worked as an electronics designer for a few years, and assembled prototypes and small production batches by hand with SMT parts (0603's, TTSOPS, etc) every day. The one exception is BGA devices, because the solder pads are underneath the device. But doing those by hand requires precise alignment that is difficult enough that few people do it. Also, for smaller BGA devices with fewer pins, skilled operators can still solder them in place with a heat gun by covering the pads with solder and flux and just melting them into place.
- jacquesm 3y agoI know two people (one of which is here on HN) who do fairly large BGAs by hand, I wonder if that skill is really that rare. I couldn't do it myself though, I did try but for some reason I can't make it work reliably enough to risk it on stuff that matters.
- segfaultbuserr 3y ago> This is incorrect. You can easily do small scale work with almost all SMT components without solder paste. Solder paste is required for automated assembly processes. But almost anything done by hand can also be done with conventional solder. I disagree. I don't consider 0603 passives and TSSOP packages "modern" anymore. Of course these components can be hand soldered with ease (possibly at top quality with the aid of a microscope). Unfortunately, the industry is gradually abandoning TSSOP and QFP in favor of DFN, QFN, and LFCSP in recent years. For anything that does high-speed signaling or multiplexing above 1 Gbps (which is old by computer's standard) like USB 3.0, PCIe 1/2, QFN goes without the need for a mention (short of using BGA). But the thing is, even in simpler ICs like DC-DC controllers, you can see the same trend. Simple RFICs are another source of heavy users of these packages, reduced circuit parasitics is certainly a factor. These packages are all leadless, and frequently with thermal pads at the bottom. An older term for leadless packages is BTC - Bottom Termination Components. [1] After a few successive and multiple failed QFN soldering attempts, I switched to ordering stencil, solder paste, and a hot plate. It worked perfectly on my first attempt, so I never looked back. Unless you have top 10% soldering skills, which I don't (experienced smartphone repair technicians seems to have mastered the art of QFN), I found solder paste is required for maintaining your sanity with leadless packages. Furthermore, without reflow soldering, prototype assembly can be very time-consuming and takes hours, especially when you need 3 or more prototypes. Occasionally, leadless packages also have optional difficulties turned on, completely eliminating the possibility of hand soldering, such as multiple bottom pads for different nets (to minimize parasitic inductance), or having two layers of contacts, one row on the exterior and on row on the interior. > You can easily do small scale work with almost all SMT components without solder paste. [...] The one exception is BGA devices And DFN, and QFN, and LFCSP, and... Thanks to industrial and automotive users, some ICs still have QFP versions for these markets (due to their vibration resistance) that are friendly for hand operation, but you have to pay a premium. Finally, even plain-old QFP chips have bottom thermal pad these days (in that case, you can manually apply a blob of solder on the PCB and reflow again with a hot air gun, but manually apply a drop of paste is easier to work with). --- [1] But these days it would make people think it's some kind of a Bitcoin mining ASIC. BTW, the last time I've checked, these ASICs are indeed QFN, so one can say they're BTC BTC chips...