1/4 oz copper

Ughh, Are you sure it's the copper and not the pcb... (1/1000 the thermal conductivity, but maybe 1000 x the area?) Any room to add 'jumpers'.. low value resistors, pieces of phosphor-bronze wire.

George H.

Unfortunately, looks like the standard thicknesses for clad laminate start at 18um.

I don't know enough about every single process for every one of the different board types and makers to say whether that's likely limiting for all available PCB types but it doesn't look like a good start.

I wonder about a serpentine pattern perhaps with the back side "Z-milled" to remove much of the dielectric material.

Stitched wire bonds should be possible. Bonding wire comes in 12.5 to 50um diameter from one datasheet I see online.

A relatively husky 1 mil (25um) diameter wire has an area of 5E-10m^2, which would be equivalent to a 0.5oz trace 1 mil wide, or more like 0.7 mil wide (0.5oz copper equivalent) if you use a gold bond wire.

--Spehro Pefhany

I was thinking of something like a small resistor, 01005 or smaller. But then it is hard to know what will work because the problem isn't just conduction through the trace, but heat sinking into the trace and anything thermally connected to it.

Better I think is to heat the traces. Consider the terminating resistor thing. Or maybe you are trying to minimize the total power. Not much info to go on.

The other idea is to provide more insulation. Maybe the traces aren't so bad by themselves so that insulating around the device will reduce heat flow enough?

I think you might try a few things, or you might want to try to develop a thermal model that will help you analyze the situation to understand it better.

Rick C.

t at 18um.

rent board types and makers to say whether that's likely limiting for all a vailable PCB types but it doesn't look like a good start.

to remove much of the dielectric material.

m diameter from one datasheet I see online.

ich would be equivalent to a 0.5oz trace 1 mil wide, or more like 0.7 mil w ide (0.5oz copper equivalent) if you use a gold bond wire.

I'm starting to think the problem should be addressed in the fluidic device itself. Just as ICs have to be designed with enough thermal conductivity to get rid of heat, it sounds like this device should be designed with enou gh thermal insulation to maintain the heat.

Rick C.

Right, at room temp all conductors are basically the same, electrical and thermal resistances are proportional. (at least in theory... I've thought about measuring, (the Weideman-Franz law.. it's all the electrons.), but haven't done it.)

Three (or more) phosphor bronze wires could hold it in place, and provide power. I guess a lot depends on the number of pieces being made.

George H.

I've calculated for a few metals and the thermal and electrical conductivity track. An exception may be brass, where the numbers I can find imply it has disproportionate thermal resistance. But the available numbers probably can't be trusted... "brass" is a fuzzy term.

Weidman-Franz apparently breaks down at intermediate temperatures such as 100K. Doesn't apply to Win's application, but IIRC when you integrate from room temperature to a few K there are differences.

--Spehro Pefhany

P.S. (maybe) you could consider rigid-flex with the flex layer 1/3 oz. I'm using that in a vacuum application where the (long) flex between rigid sections dominates the non-structural part of the thermal paths.

--Spehro Pefhany

On Feb 15, 2019, Winfield Hill wrote (in article ):

As others have noted, thermal and electrical conductivity are proportional.

But there is a key difference - one can choose the voltage and heater resistance such that very thin copper conductors (with low thermal leakage)

Ethernet.

One can also use nearby jellybean regulators to drop the voltage to whatever is convenient for the valves. Or 400 Hz (or higher) AC and transformers.

Joe Gwinn

An electrical conductor that has low thermal conductivity which is compatible with pattern printing is indium tin oxide on glass. Usually zebra-strip elastomers make the electric contact, but conductive ink would do, too. Not sure how feasible it is for your application, and it's pretty hard to inspect (we're all looking through such conductors when we read LCD displays).

How about additive process?

Some plastic with a half-loop of flex circuit wrapped round it.

Cheers

Phil Hobbs