1/4 oz copper

--Spehro Pefhany

I've used flex with 1/3 ounce. It worked fine.

Cheers

Phil Hobbs

-- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC / Hobbs ElectroOptics Optics, Electro-optics, Photonics, Analog Electronics Briarcliff Manor NY 10510

Not talking flex, not certain if it would work. Is 1/3 the limit, nobody does 1/4 or 1/8 oz? I have seen those settings on CAD software.

I purchased a kitset LED clock-timer it had very thin copper I had to glue the parts into the holes to prevent breakage.

I have not been able to measure the thickness, my calipers give 0.02 mm measured though the solder-mask which would be 1/2-oz but I think that's wrong.

The obvious question... why?

We have microfluidic channels with electrically- operated resistive-bridge valves. The traces carrying current to the valve are draining away the heat, despite using the smallest width, so need less copper thickness..

This sounds very strange. Are these resistive-bridge valves something you make or you buy? If you are buying them, how do they recommend they be connected?

Would it make sense to add some component inline as a electrically conductive bridge that is a thermal insulator? Or is that still too much thermal dissipation?

Rick C.

Yikes, that's the opposite of my usual thermal problem.

Can you wigwag the traces to increase length? How narrow a trace can you do? Some people do 2 mils on 1/2 oz.

Lots of people make custom thin-film stuff, usually on ceramic. They could probably work on polyamide or something.

I imagine wig wag traces would still drain the heat energy just the same way.

Maybe, maybe not. The longer the trace the more resistance, but then it also provides more area to dissipate heat from.

This circuit must use a very tiny amount of power. It seems like the thing itself should have better insulation.

Rick C.

Lintek comes to mind, they use a PVD process for the initial plating on bare FR4 -- much thinner foil, much finer tolerances (including vertical sidewalls on traces). I forget if it's around 1/4oz or how much finer if not.

Have you also considered something less conductive in general? Would carbon paste work? That's available at most fabs.

Tim

Mmmm, I just thought that more copper just provides a bigger sink (especially if the track width / thickness is the same).

Maybe I misunderstood - I thought the problem was that the track was removing heat from the sensor and the aim was to minimise this energy flow.

I'm now scratching my head....

me

t also provides more area to dissipate heat from.

hing itself should have better insulation.

w.

In reality everything is a conductor of heat. Insulators are just poor con ductors of heat. The copper track is a better conductor, but it still has resistance. It doesn't draw the heat out like a vacuum cleaner picks up di rt.

Sure, a thinner track will provide less conduction and more resistance. Bu t in theory a longer track will also do that. But... all copper will tend to spread the heat over a larger area which will couple to the board and th e air better for getting the heat away from the part. So whether making th e track longer will improve operation of the fluidic device depends on how well it couples to the rest of the environment.

Longer traces (zig-zag or not) can perhaps be insulated along with the flui dic device. Do you think that would help with the issue? There would stil l be some heat sinking while the traces warm up, but once they have warmed, they will reduce the heat flow compared to shorter traces.

I think there is just not enough information to suggest real solutions. On e possibility is to provide heating of the trace before it reaches the devi ce. I assume the trace driving the device carries a low voltage and likely not much current. If a higher current can be provided a terminating resis tor can be located near the fluidic device so that it thermally couples to the device through the traces and draws enough current to heat the traces t o prevent heat from leaking out through them. View in a fixed font. Term Res +-------------+ +--------| | o=======| +--+ =| Cntl | Control +-| |---| | Voltage |R | | Fluidic Dev | +-| |---| | o=======| +--+ =| Gnd | +--------| | +-------------+

Rick C.

The pcb will be conducting heat as well as the traces, so it might help to surround the device with a matrix of unplated holes to reduce the pcb conductivity and then to meander the traces around the holes to get extra track length. I have done this with i2c temperature sensors and it does help. Using the thinnest possible pcb helps a bit too. Finally, cover the critical area with a tiny box to inhibit air convection.

The earlier suggestion to use a flexible pcb with very thin copper sounds good and could be combined with holes and meandering tracks.

John

We have to make everything.

Yes, of course, do you know of any such materials?

Conductive glass exists. Have not used it personally but looked at cooking up a batch for a similar task at work. How about a thin film smt resistor ?

Steve

Electrical and thermal conductivity usually track. Most metals are about 150,000 K/w per ohm.

A tight meander close to the heat source might have advantages. That could concentrate the electrical and thermal conductivity into a hot spot close to the heat source. Of course, a thermally insulating substrate will help. Kapton is a horrible thermal conductor.

Wire bonds?

Someone suggested thick-film resistive material. Close to the valve element, their self-heating might make them have effective infinite thermal resistance. Or whatever comes after infinite.

I have asked PCB houses if they will mill out a board from the bottom side, to thin the dielectric below a conductor. They said they could. I wanted to reduce capacitance, but it would add thermal resistance too. That wouldn't be practical if your dims are really tiny.

flex PCBs are kapton

Not all. (Silver on) Polyester and FR4 are also used as flex circuits.