The non-eutectic lead-free solders are unworkable for hand-soldering (and rework), you have to get the 4% silver type with lead-free and then its fine (yes its more expensive, but it works - the other stuff is for machine soldering I think), though it runs a tad hotter than leaded - and you don't risk lead poisoning. I'd never understood the existence of 60/40 when 63/37 is better - pasty joints are never satisfactory.
I agree with that. I've never found 60/40 particularly hard to work with though. That's not eutectic. I do appreciate that joints made with 93/37 are either molten or hardened. None of this pasty stuff. For me that only comes into play when I'm soldering wires onto connectors which isn't that often. But still. If I was buying a new roll of solder I'd go for a eutectic alloy. There's no compelling reason not to.
Tom
Because no-clean flux is very hard to remove. A friend of mine calls it "can't clean" for that reason.
I've written quite a bit on this topic. You can find my thoughts here: Choosing Solder for Electronics - Neurochrome.
Flux is hydrophilic so over time it will attract water and cause performance shifts and in some cases corrosion. The no-clean fluxes are less hydrophilic so the problem is less, but still there. Also, flux can wreck the DC performance of a circuit as I show in this video:
I prefer 63/37 (not 93/37 as I typed above) Sn/Pb with water-soluble flux. In applications where I can't wash the circuit after (not all components are wash tolerant) I'll build the majority of the circuit using solder with water-soluble flux, wash it, dry it, and then finish the build with no-clean flux.
I use RMA flux in cases were I might come back and clean up the flux after. At least I have the option to do so. RMA flux can be left behind though it is a little more chemically active than no-clean.
And I use lead-free if the circuit goes into the hands of a customer. The 99.3/0.7 Sn/Cu alloys are awesome. Almost as good as the leaded solders.
From what I understand, RMA is used in military/space applications as well. They also use leaded solder. But they also use 99.9% isopropyl alcohol to remove the flux and avoid contaminants. So there's still a need for RMA flux.
Oh, and before I forget: Chip-Quik has a flux type that they market as "no-clean, water washable". Basically the idea is to have a flux that can be left behind but that you can also clean off the board if you want to. All you need to do is to hit the board with 60 ºC water and the flux washes off. Sounds great, right? Except the flux residue is really slimy and there's lots of it. It gets on your fingers. It gets everywhere. They should market it as Chip-Quik SoluSlime®. It does wash off nicely, though.
Tom
I've written quite a bit on this topic. You can find my thoughts here: Choosing Solder for Electronics - Neurochrome.
Flux is hydrophilic so over time it will attract water and cause performance shifts and in some cases corrosion. The no-clean fluxes are less hydrophilic so the problem is less, but still there. Also, flux can wreck the DC performance of a circuit as I show in this video:
I prefer 63/37 (not 93/37 as I typed above) Sn/Pb with water-soluble flux. In applications where I can't wash the circuit after (not all components are wash tolerant) I'll build the majority of the circuit using solder with water-soluble flux, wash it, dry it, and then finish the build with no-clean flux.
I use RMA flux in cases were I might come back and clean up the flux after. At least I have the option to do so. RMA flux can be left behind though it is a little more chemically active than no-clean.
And I use lead-free if the circuit goes into the hands of a customer. The 99.3/0.7 Sn/Cu alloys are awesome. Almost as good as the leaded solders.
From what I understand, RMA is used in military/space applications as well. They also use leaded solder. But they also use 99.9% isopropyl alcohol to remove the flux and avoid contaminants. So there's still a need for RMA flux.
Oh, and before I forget: Chip-Quik has a flux type that they market as "no-clean, water washable". Basically the idea is to have a flux that can be left behind but that you can also clean off the board if you want to. All you need to do is to hit the board with 60 ºC water and the flux washes off. Sounds great, right? Except the flux residue is really slimy and there's lots of it. It gets on your fingers. It gets everywhere. They should market it as Chip-Quik SoluSlime®. It does wash off nicely, though.
Tom
Flux is hydrophilic so over time it will attract water and cause performance shifts and in some cases corrosion. The no-clean fluxes are less hydrophilic so the problem is less, but still there.
Well Tom,
The whole idea of no- clean is no clean. Iirc I tried cleaning the glow-core with ipa, it just got sticky like rma does.
I do think that’s debatable in performance shifts that affect a circuits operation. A lot depends on the circuit specifics to begin with, the environment and conditions it’s operating in.
I assume you use a subcontract mfr, so what’s their smt/ tht solder/ flux / clean process ?
Last time i did a big design, iirc it was no-clean lead free process at sci/sammina? for a 10gbit test card , it was a multiple bga type design. The intel 10g Serdes was a 0.8mm pitch full bga array. Not so easy to clean flux under a bga is it? A part needing X-ray for solder joint inspection. So for bga I think that you should only use no clean flux right? Or am I out of date
knaw not much has changed in 20 years other than no clean flux has gotten better probably
Ask your contractor what process they use for bga designs?
Well Tom,
The whole idea of no- clean is no clean. Iirc I tried cleaning the glow-core with ipa, it just got sticky like rma does.
I do think that’s debatable in performance shifts that affect a circuits operation. A lot depends on the circuit specifics to begin with, the environment and conditions it’s operating in.
I assume you use a subcontract mfr, so what’s their smt/ tht solder/ flux / clean process ?
Last time i did a big design, iirc it was no-clean lead free process at sci/sammina? for a 10gbit test card , it was a multiple bga type design. The intel 10g Serdes was a 0.8mm pitch full bga array. Not so easy to clean flux under a bga is it? A part needing X-ray for solder joint inspection. So for bga I think that you should only use no clean flux right? Or am I out of date
knaw not much has changed in 20 years other than no clean flux has gotten better probably Ask your contractor what process they use for bga designs?
Last edited:
I agree, BGA is not the generally prefered package for analog but they do offer a LM358 in a 1.3x1.3x0.5mm DSBGA-8, damn that's small.
Are analog circuit more prone to malfunction than digital circuits due to slightly conductive flux?
Both no-clean and aqueous are viable process
So it comes down to application and environment as to the type of solder flux that is used and if it needs to be removed.
Are analog circuit more prone to malfunction than digital circuits due to slightly conductive flux?
Both no-clean and aqueous are viable process
So it comes down to application and environment as to the type of solder flux that is used and if it needs to be removed.
Eutectic tin/lead is great if your inspection criteria is "shiny".
However, eutectic freezes such that the tensile stress at the copper/tin intermetallic boundary is greatest. This is caused by the 15% shrinkage by volume that the solder does when it changes phase. Unfortunately, since the air contact surface of the solder is the first to freeze, the stress is built into the joint.
This tensile stress will make the joint susceptible to temperature cycling fatigue failure. For example, relay socket copper pins through holes, the solder over time will detach from the intermetallic surface, ending in a burn at failure.
Off eutectic solders will freeze differently. As you pass through the liquid line on the phase diagram, the precipitate that comes out will be high lead (assuming tin/lead 50/50 for discussion), and the liquid will become more and more tin concentrated heading towards eutectic. (google tie line for a good explanation). When the joint falls to 183C, it will have very little tensile stress at the interface. This is exactly how 50/50 tin/lead works.
Same shrinkage/tensile stress applies to tin/silver, but at 221C. Tin bismuth is great in that it doesn't shrink at freeze, so joints are better at surviving low cycle fatigue typical of top of telephone poles temperature swings in the desert (Bell labs, 1950's).
Non wetting to a pcb surface is all about the flux. If the flux is unable to clear the oxide and dirt of the surface, the solder will never wick. Scotchbrite is your friend. If you need RA flux, get some neutralizing solution to clean the joints. (stained glass work used that a lot, as stained glass work tens to use water based RA, and if you don't neutralize it, the solder will be white in a few months. They call it mold, but it ain't..
John
However, eutectic freezes such that the tensile stress at the copper/tin intermetallic boundary is greatest. This is caused by the 15% shrinkage by volume that the solder does when it changes phase. Unfortunately, since the air contact surface of the solder is the first to freeze, the stress is built into the joint.
This tensile stress will make the joint susceptible to temperature cycling fatigue failure. For example, relay socket copper pins through holes, the solder over time will detach from the intermetallic surface, ending in a burn at failure.
Off eutectic solders will freeze differently. As you pass through the liquid line on the phase diagram, the precipitate that comes out will be high lead (assuming tin/lead 50/50 for discussion), and the liquid will become more and more tin concentrated heading towards eutectic. (google tie line for a good explanation). When the joint falls to 183C, it will have very little tensile stress at the interface. This is exactly how 50/50 tin/lead works.
Same shrinkage/tensile stress applies to tin/silver, but at 221C. Tin bismuth is great in that it doesn't shrink at freeze, so joints are better at surviving low cycle fatigue typical of top of telephone poles temperature swings in the desert (Bell labs, 1950's).
Non wetting to a pcb surface is all about the flux. If the flux is unable to clear the oxide and dirt of the surface, the solder will never wick. Scotchbrite is your friend. If you need RA flux, get some neutralizing solution to clean the joints. (stained glass work used that a lot, as stained glass work tens to use water based RA, and if you don't neutralize it, the solder will be white in a few months. They call it mold, but it ain't..
John
Last edited:
Surely the copper wll draw the heat out of the molten solder better than air(*) and solidification occurs at the copper/solder interface first, so internal stresses aren't present there. You can certainly see the solidification spreading from the part being soldered outwards into the air.
(*) ratio of heat conductivities is 4 orders of magnitude different! Also clean metal surfaces don't radiate heat well so conduction back along the wire or trace should dominate.
(*) ratio of heat conductivities is 4 orders of magnitude different! Also clean metal surfaces don't radiate heat well so conduction back along the wire or trace should dominate.
For critical applications (life support electronics, aerospace, defence) lead free solder is not allowed. Generally eutectic tin/lead. Reason is that lead-free solders are prone to whisker growth which compromises reliability.
In practice for domestic stuff that does not require extreme reliability, or is not used in challenging environments, it doesn't matter so much. But in applications where a solder related fault can cost lives, or where it is impossible (such as satellites) to do a service visit, tin/lead eutectic is the solder of choice.
In practice for domestic stuff that does not require extreme reliability, or is not used in challenging environments, it doesn't matter so much. But in applications where a solder related fault can cost lives, or where it is impossible (such as satellites) to do a service visit, tin/lead eutectic is the solder of choice.
Remember heat capacity. You have to fill the copper with energy to get above melt temperature. Cooling all that heat capacity doesn't happen from inside the hot copper.
NIce thinking however. I can think of instances where you are correct.
I've had to solder large copper things where I needed to heat the entire structure to 150 prior to using a soldering iron, so the air does do it first. Relay socket pins that are thick square copper will retain the heat such that the surface solidifies first.
John
Last edited:
Agreed. However, it appears that tin/silver eutectic seems to stall or prevent whiskers. We use it exclusively in our big machines that run at 4.5 Kelvin in helium.
In our application, even a small flashover will ionize helium gas and the expanding plasma ball will not stop until all the energy is gone. 5 henries and 5500 amps is 75 megajoules, so when one starts, we do tend to notice, as would seismologists in other states.
The biggest issue I am finding is even if we specify tin/lead for assembly of PC boards, most of the components available now are tin plated leads. I occasionally autopsy a failure and find whiskers all over the place.
Tin pest was also a worry for cryogenic use, but that is absolutely not an issue.
John
I used to be Engineering Director of Oxford Instruments in the early to late 90's. So indeed I know a bit about the hazards of superconducting magnet quenches; we used huge back to back Schottky diode/resistor voltage limiters across different magnet sections, and even across different superconductor grades to prevent arc formation during a quench.
We had a guy who designed a rather large 1 metre bore mineral separation magnet without voltage limiters. First quench - pow. Once warmed up around a kilo of the end of the magnet was completely missing - vaporized. Hell of a mess. Bit of a inquest into that.
But yes - with the sort of energies involved in superconducting magnets, tin whiskers do not have a chance.
The interesting thing is that space grade parts have gold plated ends - both through hole and surface mount. The gold is only there as a protective layer to prevent oxidation of the component end. However the problem is that if you solder to the gold plating, the gold dissolves in solder, and causes joint embrittlement. So all parts have to be de-golded in a special solder bath. The solder bath gold content is assayed on a regular basis so that the concentration of gold is known and the solder in the bath changed. After de-golding the parts are held under dry nitrogen until soldered in place.
We had a guy who designed a rather large 1 metre bore mineral separation magnet without voltage limiters. First quench - pow. Once warmed up around a kilo of the end of the magnet was completely missing - vaporized. Hell of a mess. Bit of a inquest into that.
But yes - with the sort of energies involved in superconducting magnets, tin whiskers do not have a chance.
The interesting thing is that space grade parts have gold plated ends - both through hole and surface mount. The gold is only there as a protective layer to prevent oxidation of the component end. However the problem is that if you solder to the gold plating, the gold dissolves in solder, and causes joint embrittlement. So all parts have to be de-golded in a special solder bath. The solder bath gold content is assayed on a regular basis so that the concentration of gold is known and the solder in the bath changed. After de-golding the parts are held under dry nitrogen until soldered in place.
Hello,
Several years ago I bought a few rolls of surplus Multicore CRYSTAL 400 63-37 Solder, much the same as recommended by Mark Johnson above. Turns out this is very good solder, if everything is clean it flows nice and the the zero Halide flux is non-corrosive.
It is a absolute must to wash off the Kester 44 flux as it is more than a little corrosive.
Thanks DT
https://au.element14.com/edsyn/su8250-s5/solder-wire-fsw34-0-8mm-250g/dp/3236870
Tin Lead Copper ? difference over silver content
https://au.element14.com/stannol/631902/solder-wire-lead-free-1-0mm-100g/dp/840031801
8Use this wire with high silver content 3.8%, in my speaker crossovers and i found issues was infact the solder on the board some good some bad
63/37 that i use now for everything
https://au.element14.com/duratool/spc22158/solder-wire-63-37-sn-pb/dp/2527485
little bit left of Chip Quik tin lead silver 2%, saved for interconnects.
I have to redo my Celestion SL-6 crossovers, if was most likely why the speakers just stopped working, as when the woofer and tweeter were out they read the correct ohms, so not blown.
I would need to buy more of the Chip Quik or would the Duratool 63./37 be fine to use? crossover has silver plated wire to the speakers and woofer and terminal. Why i thought of getting more of the Chip Quik, both flow very well. with lovely shinny joints as opposed to the Stannol i used a few years back
Tin Lead Copper ? difference over silver content
https://au.element14.com/stannol/631902/solder-wire-lead-free-1-0mm-100g/dp/840031801
8Use this wire with high silver content 3.8%, in my speaker crossovers and i found issues was infact the solder on the board some good some bad
63/37 that i use now for everything
https://au.element14.com/duratool/spc22158/solder-wire-63-37-sn-pb/dp/2527485
little bit left of Chip Quik tin lead silver 2%, saved for interconnects.
I have to redo my Celestion SL-6 crossovers, if was most likely why the speakers just stopped working, as when the woofer and tweeter were out they read the correct ohms, so not blown.
I would need to buy more of the Chip Quik or would the Duratool 63./37 be fine to use? crossover has silver plated wire to the speakers and woofer and terminal. Why i thought of getting more of the Chip Quik, both flow very well. with lovely shinny joints as opposed to the Stannol i used a few years back
I have a roll of Jaycar 60/40 Duratech branded solder from about 20 years ago and I find it pretty good.
Last edited: