What's new
What's new

Welding extremely small wires

Comatose

Titanium
Joined
Feb 25, 2005
Location
Akron, OH
I need to weld five 36 gauge type T thermocouples at right angles to a length of 28 gauge nickel 200 wire for a test fixture I am building.

Which is a bit nasty to begin with, but here's the real kicker. I have to weld the two thermocouple wires to each other and the carrier 28 gauge wire while remaining stacked perpendicularly. That is, the copper has to weld to the nickel, and then the constantan has to weld to the copper, but not to the nickel wire below it. Or the stack could run the other way around, nickel to constantan to copper.

That part of it is kicking my butt. I have a Unitek HF25 resistance welder, so I have pretty decent control of the weld parameters, but fixturing it to stay neatly stacked is proving... problematic.

Does anyone here do this level of stupid small stuff on a daily basis and either have ideas or just want to do it? I'm thinking laser welding might be a better approach, but I don't have a laser welder.

I only need two of these assemblies ever. I hope.

Thanks
John
 
I don't think laser will help all that much at these sizes. Is there no way you can increase the wire gauge even a few sizes??

Can you replace the middle member in your junction with a piece - strip of foil rather than wire? Simply trim it down post weld should make life a lot lot easier. I know weld - junction size makes little difference to the voltage temp generated hence it should have no effect bar a little extra mass - slow the response speed a tad.
 
No I sadly can't make the wire any bigger, the wire is the device under test, if the thermocouple wires get bigger then they start pulling heat away from the main wire, and the measurement becomes suspect.

Also the wire is electrically live during test, which is why the thermocouple wires can't both be touching the wire, that will induce a voltage into them from the driving current, and then reading an accurate temperature gets VERY dicey.

Using an intermediate foil and trimming it away afterwards seems like it'd be worth a shot. It isn't so much that this particular operation is hard conceptually, it is just the ultra-small geometry is kicking my butt.

I can't make the wire gauge bigger because the transient temperature response of the wire is what I care about. Bigger will react slower, which doesn't tell me anything about the stability or accuracy of the control system driving it.

For lack of a better description I am making a calibration fixture for hot wire anemometer type things.
 
Flattening the wire first seems like a great tip! "Fixture it with fifteen layers of kapton" is the electronics solution to just about everything, no?

Thanks! That gives me hope.
 
Why not just mesure its temp via non contact means? Omron have sensing solutions to do that, yeah they will need some calibration - correction factors figured for emissivity if you need high pre-scion, but may be a lot better option than adding thermocouples to what is already a very low thermal mass object.
 
Why does the weld joining the two thermocouple wires have to be on top of the weld joining the thermocouple to the target? Couldn't you weld the thermocouple wires and trim so one (copper) has a slightly longer lead, and then weld that lead to the target?
 
I'd read somewhere that a third metal in the middle of a thermocouple junction just cancels out and has no effect on the thermocouple voltage. I looked this up years ago when I spotted a researcher making T thermocouple junctions with soft solder and a soldering iron. This would mean you could spot weld your thermocouple wires either side of the nickel wire and it would still be a valid thermocouple junction with the copper-nickel and nickel-constantan junctions cancelling out the nickel component of the thermocouple voltages. You may even be able to use 60-40 electronics solder to attach the thermocouple to the nickel if used sparingly.
 
Contact a resistance weld equipment manufacturer and tell them you want to buy a new machine that can do the weld you are looking to do. Send them 10 or so samples and ask them to specify equipment and fixtures required. They will send back your samples both done well and not so well. Pick out the two best and tell them thanks but you didn't get the contract.
 
Ultrasonic Bonding

I need to weld five 36 gauge type T thermocouples at right angles to a length of 28 gauge nickel 200 wire for a test fixture I am building.

Which is a bit nasty to begin with, but here's the real kicker. I have to weld the two thermocouple wires to each other and the carrier 28 gauge wire while remaining stacked perpendicularly. That is, the copper has to weld to the nickel, and then the constantan has to weld to the copper, but not to the nickel wire below it. Or the stack could run the other way around, nickel to constantan to copper.

That part of it is kicking my butt. I have a Unitek HF25 resistance welder, so I have pretty decent control of the weld parameters, but fixturing it to stay neatly stacked is proving... problematic.

Does anyone here do this level of stupid small stuff on a daily basis and either have ideas or just want to do it? I'm thinking laser welding might be a better approach, but I don't have a laser welder.

I only need two of these assemblies ever. I hope.

Thanks
John
I have had some exposure to Ultrasonic Bonding to bond the wires in Semi-Conductors devices. This may be another way to assemble your components instead of Welding and deserves your investigation.
Roger 01/18/2017
 
Surface temperature measurement? Assuming transient response and minimal effect from external/environmental/convection effects is important? I'd side with SAG 180's comments. It's called an "intrinsic thermocouple". Make a quick search on that to see if I'm misunderstanding you're objective here. But we had very good results with them when measuring surface temperatures on the combustion-side of an external combustion engine's pressure vessel wall to track creep in some nickel alloys. We didn't deal with quite that small of wires, but add some magnification (and maybe some kapton tape) and it shouldn't be a problem to make the junctions with minimal practice. Something like this (okay, this) is what we used:
http://www.dcccorporation.com/onespec.html?&plpver=10&origin=compare&filter=0
Cheap, easy, adjustable power setting for wire gauge and materials being welded. Hopefully this can help. Good luck.

*edit, what I mean here is that you can weld your copper to nickel and within a millimeter or so, also weld constantan to nickel, and the resultant measurement through your bridge will essentially cancel out the effect of the nickel and you'll get a nickel surface temperature measurement between the two weld joints. Much easier to fixture; and from what I've read, been told, and experienced...more accurate for surface temps than stacking weld joints like you'd been asked to do.
 
+1 on Sag180's comments except I'd worry about soft solder 'poisoning' the couples. Why can't the nickel be used as it's own temperature sensor? Nickel RTDs are a standard device.
 
I sometimes join small RTD wire to an extension with the TIG. Pinch the wire between the weld and the sensor to act as a heat sink and ground. The TIG gun, in my case, is in a holder so that it can be positioned more precisely than if I hold it. Most of the time the wires burn back 1/16" forming a ball.
 
Im with the OP, you don't want the wire your measuring in the thermocouple loop, yeah you can do this and get measurements that are just fine but not when that wires carrying power - acting as a heating element. The changing resistance + resistance over the gap between the thermocouple connections and applied voltages will screw with stuff.
 








 
Back
Top