low thermal EMF solder

[AnonUser]

Hello - I am looking for a source to supply low thermal EMF solder - can anyone provide pointers ?

In the past a cadmium - lead alloy was used for this, but these are banned (RoHs) ... - the aim is to minimize thermo-couples on a very sensitive 4 layer printed circuit board - 0.1ppm / K on a 10V full scale - so every 100nV we can reduce is taken under consideration (preventive actions)

The assembly factory is located in Belgium.

Thank you for all feedback !

[Dr. Frank]

Just to finish your search quickly:
We habe discussed that on volt-nuts already,  Cd based solder really is not available any more, and nobody found an alternative, either.

Frank

[Vgkid]

http://www.kappalloy.com/tec-solder.php
kapptec solder google it.
I cant find any vendors, most others are hard solders, so that is a sign.

[SArepairman]

how difficult is it to craft this solder? I assume the lower melting point materials are melted first, then the silver metal dissolves in them due to solubility rather then temperature.

Both elements can be bought in high purity fairly cheap, what are the ratios?
Is anyone interested in buying some home made Cd-silver solder?

Sure, I cannot offer you a flux core, but it should still be highly usable. I assume it is only used around references ETC, not the whole PCB?

it could be easily melted under an argon shield aswell if oxide contaminants are a concern, with metal pre-treated with mineral acids to remove oxides.

how hard is this stuff? can it easily be cast? what is the maximum solder diameter that you would consider useable? how much would you be willing to pay by weight?

[SArepairman]

well I am interested in this stuff now too and I want some, I will make it for myself if there are no sources. Cadmium rod is damn cheap on ebay...


for production what do you need, solder paste? I think home made solder rod would be appropriate for through hole, but SMD thin solder would probably require some kind of special extruder machinery, unless you are like my friend and some how manage TSSOP with "pipe solder" and a 10$ soldering iron (while I need a nice iron and 0.7mm or less solder)

I imagine that you are probably using through hole technology if you are worried about thermal EMF, due to strain and whatnot.

But I do wonder if you could melt the alloy then grind it up in a chilled ball mill and put it through mesh sieves and mix it with flux... i bet robrenz would be interested in this ...
or is it produced through atomization and cooling (for spherical particles), not that hard a task for low melting point solids..

do you have any literature about the chemistry of these low emf solders?

i think realistically you need to contact a solder paste manufacturer and ask them if they can do a special batch for you, you could try a kick starter or something like this when you get a quote for low EMF solder production to manage the upfront costs, I'm sure some people might be interested.

i wonder if you can mix up the alloy yourself and then just mail it to some kind of factory to save money, i wonder how hard it is to make an alloy

[Conrad Hoffman]

The old L&N low thermal solder was (according to one reference I found) 70% Cadmimum and 30% Tin. Should be trivial to make, but you probably couldn't legally sell a product using the stuff in most places.

[SArepairman]

rohs compliant unless proven otherwise

[rexxar]

rohs compliant unless proven otherwise

Cadmium is banned by RoHS.

[zapta]

Hello - I am looking for a source to supply low thermal EMF solder - can anyone provide pointers ?

What is a low thermal EMF solder?

[SArepairman]

Hello - I am looking for a source to supply low thermal EMF solder - can anyone provide pointers ?

What is a low thermal EMF solder?

copper-solder connection makes a thermocouple, http://www.eevblog.com/2013/02/03/eevblog-419-thermocouple-tutorial/
some solders make a weaker thermocouple, so you can have instead of 1uV/degree C 0.1uV / degree C or something like this

[acbern]

keithley recommends Ag silver (4%) for use with their low emf connectors with 2182 nanovolt meters. keep in mind a emf voltage can only be generated if there is a thermal gradient in a solder joint pair. this is the first thing to avoid. given the precision achievable today with e.g. the 2182 and its cable this proofs there really is no need for cadmium solder. if you have a emf problem, it is elsewhere in the design.

[Vgkid]

That makes since, as it is probally rohs certified.

[SArepairman]

i wanna punch rohs in the face until all its stupid solder whiskers fall off 

[Andreas]

i wanna punch rohs in the face until all its stupid solder whiskers fall off 

Hello,

I do not think that you really want to have cadmium solder.
During soldering cadmium solder gives very poisonous vapours.
Btw. LT AN86 speaks of 60% Cd + 40% Sn.

With best regards

Andreas

[Conrad Hoffman]

You don't need a gradient across the junction to generate a voltage. If it simply exists above absolute zero, you get the voltage. Now, single joints aren't very popular- there's usually another one involved. If you can keep the pairs of joints at the same temperature, no gradiant, the voltages will cancel out and no problem. I think Jim Williams even suggested that an extra "unnecessary" joint could sometimes be incorporated to provide cancellation when necessary.

[ManateeMafia]

I would like to know if anyone has any documented study or application note that covers testing of different types of solder and measuring the Seebeck coefficients? You can easily find information on paired materials like copper-gold, but I have not seen too much information about the vast array of solders.

I understand the need for paired joints and thermal equilibrium, but having empirical data would give people the best solution based on what is available to them.

[CaptnYellowShirt]

Cadmium solder is to the EE as Loctite is to the ME -- if you have to use it, something is wrong with your design.

[SeanB]

Normally Loctite is used because you either do not have the space to put in a split pin or you do not want to use a deformable use once fastener. Plus it is easy to use in production as you can order fasteners in bulk with it pre applied. No extra steps required.

[mzzj]

keithley recommends Ag silver (4%) for use with their low emf connectors with 2182 nanovolt meters. keep in mind a emf voltage can only be generated if there is a thermal gradient in the solder joint. this is the first thing to avoid. given the precision achievable today with e.g. the 2182 and its cable this proofs there really is no need for cadmium solder. if you have a emf problem, it is elsewhere in the design.

Second that.
Another trick beside obvious advice of keeping solder joints away from thermal gradients is the joint itself. Last thing you want to do is to "bridge" gaps with the solder.
For example wrap the wires together to get good  thermal coupling and solder the joint after that. and pcb mounted component leads should be bend parallel to traces to maximize the contact area.

[SArepairman]

i wanna punch rohs in the face until all its stupid solder whiskers fall off 

Hello,

I do not think that you really want to have cadmium solder.
During soldering cadmium solder gives very poisonous vapours.
Btw. LT AN86 speaks of 60% Cd + 40% Sn.

With best regards

Andreas

oh shit, i did not realize cadmium boiled at 800c, i assume it gives a fair bit more vapor then lead (boils at 1600c).

the vapor pressure at soldering temperatures is enough to make working with it fairly toxic? sounds like it should be done under a fume hood or outside with a fan.. heh glad i caught that
http://www.powerstream.com/z/vapor-press1.png

ooo, kinda bad, 10^-2 torr @ 300c, compare to lead and tin.. which have > 10^-8

certainly don't wanna do that indoors!

whats the working temperature of the solder?

and so long you are not an idiot about it I think it could be done safely. i would feel way more comfortable with cadmium solders (working with or manufacturing) then say, driving a car down the street, or going for a walk around my block.  you are in control of your own safety here. it would be one of the safer parts of my day.

compared to the amount of times I had a elevated heart rate from a bad driver to the amount of times I poisoned myself with heavy metals... hmm.. no contest

the only dangerous part of this would be running the electronics and a part running hot, possibly slowly vaporizing cadmium. this is where the danger is. might be worth while to put some thermal fuses on that circuit board! I would not want it running at too much more then room temperature. This could be considered if you have a thermally stabilized voltage reference soldered down with cadmium solder.

*in all likely hood though, you should probably focus your efforts on staying away from cigarette smoke if you are scared of Cd

[CaptnYellowShirt]

Another trick beside obvious advice of keeping solder joints away from thermal gradients is the joint itself.

I hear this a lot on this blog. There's this idea out there that thermal EMF is generated *at* a solder junction.

To my understanding, thermal EMF is generated along the wires that bridge areas of different temperatures. Its just that because the process is path independent, the thermal EMF can only be realized at junctions:



Here's a sanity check... how can you get a thermal gradient of any significance across a solder junction? And how do thermocouples work? You place the wires between the two baths, not the junctions.

[Andreas]

ooo, kinda bad, 10^-2 torr @ 300c, compare to lead and tin.. which have > 10^-8

The diagram is only valid for pure metals.
With alloys at least the melting point will go down.

With best regards

Andreas

[Andreas]

how can you get a thermal gradient of any significance across a solder junction?

Probably less a problem if you use copper/solder/copper joints. (around 3uV/K)
You will have 2 thermocouples with (slightly) different temperatures.
So they do not cancel out exactly.

The problem arises if you use hermetically tight components.
(Ceramic or metal case e.g. LM399 / LTZ1000)
Soldered on PCB you will get a Kovar/solder/copper joint.
Kovar has 40uV/K against copper.
And if you use a heated reference there will be significant difference
across (+ between different) solder joints of the component.

The clue is: Cd-solder will not help in this case 
The reason is clear. It has only 0.3ppm/K against copper and not against Kovar.
The only way to get out is to do a excellent thermal isolation of the cirquit.
(see LM399 thread with the construction of Mickle).

With best regards

Andreas

[mzzj]

Here's a sanity check... how can you get a thermal gradient of any significance across a solder junction? And how do thermocouples work? You place the wires between the two baths, not the junctions.
[/quote]

I agree that thermal gradients over solder junctions are tiny. Unless you bridge large gaps with solder or something stupid like that.

I did quick test, 2 pieces of 0,75mm2 wire soldered together badly end-to-end and connected to Agilent 34401A:
Holding only one of the wires between my fingers from near the solder junction results ~2uV offset compared to +-0,1uV seen otherwise. 
Depending which side of junction I heat with my finger the readings go positive or negative. 

And yes I agree, Connectors and dissimilar wires(ie kovar) are much bigger problem than the solder you use.

[robrenz]

My experiments on the thermal EMF topic a while back here

[SeanB]

I was able to turn an older ammeter into a thermocouple meter just by heating the one terminal with a soldering iron while repairing a broken wire. When finished soldering the wire back I loked at the front and it was at full scale and slowly coming back down. Basically a copper/constantan thermocouple with a cold junction at room temperature.

[CaptnYellowShirt]

My experiments on the thermal EMF topic a while back here

I had forgotten about that test until you posted it again. Real world data is always the best. Thanks.



So robrenz is seeing EMF's across solder joints in the range of 1's to 10's of nanovolts per deg C.

I've never done the experiments myself, but that's the range I would expect.

From my experience, offsets of 1-100's of microvolts are an issue of the circuit in its entirety -- not just one joint. If you're seeing this type of offset, you have some type of imbalance along the entire path of the circuit -- an odd number of joints or a pair of joints at different temperatures.

Kovar / solder may produce a high off set voltage. But remember, in any circuit you should have at least *2* of these joints which will be opposing one another. The design objective here isnt to minimize the kovar / solder offset voltage. Its to ensure the pair of joints operates at the same temperature.

[David Hess]

The voltage is generated across the length of wire at different temperatures and not the junction itself which explains why extra pure wire is used for runs across large temperature differentials.

[LukeW]

This sounds like a job for... the giant Wikipedia table of solder and solder-like alloys!

http://en.wikipedia.org/wiki/Solder#Solder_alloys

Note Cd70/Sn30 on that table as a low-thermal-EMF solder, or Pb90/Sn10 as a Cd-free alternative.

I haven't checked specifically what the RoHS exemptions list says, but there's probably an exemption for it, because I don't think I've ever heard of a low-thermal-EMF solder which is both Pb and Cd free. I know there's an exemption for high-melting-point Pb solder off the top of my head.

The table says nothing about the thermoelectric properties of Sn95.5/Ag4/Cu0.5

[quarks]

just looked for low EMF solder and as stated in other threads Keithley suggests for 2182 Sn96/Ag4

Does anyone know or where to find ppm/K comparison information for

Pb90/Sn10
Sn96/Ag4
Sn95.5/Ag4/Cu0.5


 

[David Hess]

just looked for low EMF solder and as stated in other threads Keithley suggests for 2182 Sn96/Ag4

You may have just resolved a decades long mystery for me.  I have a spool of Kester solder which is only marked Sn96.  Maybe it is low EMF solder.

[SvanGool]

@quarks

Meanwhile the "volt-nuts" thread on "Thermal EMF" https://www.febo.com/pipermail/volt-nuts/2016-July/004912.html measurements by Andrea Baldoni:

• Copper - Sn96/Ag4 -> 3.33uV/K
• Copper - Sn95.5/Ag3.8/Cu0.7 -> 3.22uV/K

Best "leaded":     Copper - Pb92.5/Sn5/Ag2.5 -> 3.02uV/K
Best "lead free":  Sn95.5/Ag3.8/Cu0.7 -> 3.22uV/K

My personal conclusion would be: avoid soldering and use crimping/clamping of pure copper, possibly with silver or gold plating, without plating use Deoxit to keep it clean from copper oxide.
The best method is, of course, to keep any temperature gradients away from single joints (joints in both wires may compensate each other), by thoroughly isolating them from drafts and other random temperature variations.

[Conrad Hoffman]

From the Department of Useless Trivia- Leeds & Northrup identified the low thermal solder joints in their equipment by painting them green.

[quarks]

@quarks

Meanwhile the "volt-nuts" thread on "Thermal EMF" https://www.febo.com/pipermail/volt-nuts/2016-July/004912.html measurements by Andrea Baldoni:

• Copper - Sn96/Ag4 -> 3.33uV/K
• Copper - Sn95.5/Ag3.8/Cu0.7 -> 3.22uV/K

Best "leaded":     Copper - Pb92.5/Sn5/Ag2.5 -> 3.02uV/K
Best "lead free":  Sn95.5/Ag3.8/Cu0.7 -> 3.22uV/K

My personal conclusion would be: avoid soldering and use crimping/clamping of pure copper, possibly with silver or gold plating, without plating use Deoxit to keep it clean from copper oxide.
The best method is, of course, to keep any temperature gradients away from single joints (joints in both wires may compensate each other), by thoroughly isolating them from drafts and other random temperature variations.

Thanks a lot for the info
I also do prefer crimping whenever possible

[Kosmic]

(sorry for necroposting)


So apparently you can replace cadmium solder by Pb90Sn10. Solders low  in Tin should have the same low EMF properties than cadmium based solders.

Now lead based solders are now hard to find but it's not impossible. I just bought 5 pounds from ebay. I should have enought for a while   

Ref:
https://en.wikipedia.org/wiki/Solder
http://www.analog.com/library/analogDialogue/archives/39-05/Web_Ch4_final.pdf

[cellularmitosis]

(I see no problem with necroposting  keeps the information concentrated)

This may be of interest to this thread (attached).

[RandallMcRee]

(I see no problem with necroposting  keeps the information concentrated)

This may be of interest to this thread (attached).

Agree--necroposting ok, here. Summary would be nice. I think this is it:

The data from this study suggest that the proteinuria
induced by excessive exposure to cadmium
is not reversible even if exposure to the
metal is greatly reduced.

[Magnificent Bastard]

For low thermal EMF's, why not CD weld copper to copper?  A simple capacitor+SCR CD welder is pretty easy to design and build, or you can buy a kit for a more sophisticated model:

http://zeva.com.au/Projects/SpotWelder/
https://www.philpem.me.uk/elec/welder/
http://www.pittnerovi.com/jiri/hobby/electronics/welder/

Kit:  https://www.keenlab.de/index.php/product-category/kspot-welder-kit/

[texaspyro]

For low thermal EMF's, why not CD weld copper to copper?

Welding copper to copper, particularly with a simple welder?  Not gonna reliably happen.  My CD welder is FAR from simple (20,000 amp peak, continuous weld current and pressure monitoring, etc) and copper welds are still crap or non-existant.  Also if the materials being welded are connected to any circuitry the induced spikes will kill it.

Everyone and their mother seems to think you can whack together a reliable CD welder...  nope...

[Kosmic]

(sorry for necroposting)


So apparently you can replace cadmium solder by Pb90Sn10. Solders low  in Tin should have the same low EMF properties than cadmium based solders.

Now lead based solders are now hard to find but it's not impossible. I just bought 5 pounds from ebay. I should have enought for a while   

Ref:
https://en.wikipedia.org/wiki/Solder
http://www.analog.com/library/analogDialogue/archives/39-05/Web_Ch4_final.pdf

So in the end Pb90Sn10 is really a pain to use. Not super liquid and doesn't stick super well.

I don't think I'm going to use it much.

[leighcorrigall]

Herewith another necropost! 

I am in the middle of purchasing a Vishay HZ Series (Model HVA) resistor for a standard and am contemplating how to connect the leads to binding posts. In this pursuit, I was given the following information from a Vishay Applications Engineer who has had 'considerable experience' with hand soldering VHA resistors in custom decade boxes:

'[Given that] Cd-Sn solder is essentially impossible to buy today because of the very toxic nature of cadmium, the recommend solder to use for Sn-Pb leaded VHA resistors is Sn60Pb40. A better and safer solution is to use a razor blade or X-ACTO knife to carefully scrape the solder off of the ends of the resistor leads until there is nothing but bare copper (i.e., no solder coating). Then tightly wind the "no-solder" wire around the test terminal and solder with regular Sn60-Pb40. Thermal EMF (Seebeck effect) is created by heating the junction of dissimilar metals. Eliminating the solder layer between the copper lead and terminal will provide a very low thermal connection.' -- Vishay Applications Engineer with 30+ years experience.

A similar wire-wrapping example can be found with this link (https://youtu.be/a6AIoEj5Bzw?t=1016) from the YouTube channel 'theBreadboard' where he inspects a Vishay-made custom reference box. The reviewer does not allude to low EMF attributes, but he does suggest that the 'quality of the connection' is greatly improved as a result of wrapping the leads around the bus connection.

From SvanGool's previous post, I examined these cadmium-free options to solder precision resistor leads:

Cu conductor to Sn5/Pb92.5/Ag2.5: 3.02 µV/°C, 296 °C melting point (highest melting point)
Cu conductor to Sn95.5/Ag3.8/Cu0.7: 3.22 µV/°C, 219 °C melting point (poor wetting)
Cu conductor to Sn96/Ag4: 3.33 µV/°C, 223 °C melting point (poor wetting)
Cu conductor to Sn60/Pb40: 3.34 µV/°C, 190 °C melting point (best compromise to avoid heating the resistor)

The choice of solder will ultimately depend on your application.  Obviously, there are other properties to consider besides liquidus temperature. 

Ultimately, the reduction of EMF can be achieved by making good contact with the copper conductors and avoiding temperature gradients at dissimilar metal junctions while under test. This rule has been mentioned already but should be emphasized.

Good luck!

[Dr. Frank]

Hello,
low e.m.f. (solder) is required ONLY for voltage measurements!
For reistors, any e.m.f. can be cancelled by using the Offset Compensation method.

Frank