PCBs in UHV

Done a fair amount of UHV work in the past but never put a pc board in a UHV system (Agilent puts circuit boards at the detector end of their little quadrupole mass specs, the MSD series, but they only need 10-7 torr at best and I don't recall any components, just traces for connections). I've read that fiberglass boards can be a real problem in UHV - if there are voids in the epoxy the trapped air can leak out along a glass fiber essentially forever. You can bandaid that with a really big pump :-). Actually I think that the board will be the easiest part - hardly any component you would solder on the board, or the solder, will be UHV compatible. What temperature range do you need the electronics to handle, and what pumping speed and type of pump do you have?

----- Regards, Carl Ijames

I don't have any UHV gear. Hey, I don;t have any techs, either!

No, really, there are some UHV-compatible PCB materials, polyimide and teflon based. I think.

Do they still make them? My customers use turbomolecular pumps, basically fans. A few extreme cases use cryo and ion pumps, too.

John

The system uses turbo pumps. It has existing ghastly electronics that uses plastic DIP ic's and axial passives that are hand point-to-point wired, with solder, on a drilled ceramic slab maybe 2" square. That means they have essentially only radiation cooling, and I calculate the critical (jfet!) opamps run at maybe 120C. If dips and solder joints running at 120C don't outgass enough to trash the system, it must be pretty hardy. Operating ambient is room temp. They obviously don't bake it really hard.

It was designed by chemists. What can I say?

In addition to all the other problems, chip cooling is an issue when there's no air! That's where an AlN substrate would eventually be great. But a multilayer PCB would sure be easier.

John

If you could add a heatpipe or, even better, a fluid-filled cooling loop in the UHV, you could cool the amps and lower noise a bunch.

Cheers, James

John Larkin schrieb:

Hello,

I suspect the plastic DIP ic's and axial passives will outgas too, not only the PCBs.

Bye

Glen Walpert schrieb:

Hello,

if the PCB material may be baked out, but the components mounted on the PCB are not suitable for bake-out, it would not help much.

Bye

That would be the ultimate amp, something cooled a lot. Now I can add a bunch of fun new problems to the mix, like thermal stresses and semiconductor behavior and ceramic caps quitting.

Sounds difficult and expensive. I like that.

John

Another issue is contamination of the UHV chamber. Some of the stuff that outgasses during bake-out of the system or during operation will recondense in cooler parts of the chamber (outbaking is never homogenous, unless you have one of those cool "baking tents") and will slowly keep evaporating from there. Whether or not this is a problem depends on what your customer is doing in the vacuum system. Before introducing long-chained organic material --and what's the fire-retardant stuff they use nowadays? Some bromide compound?-- you need to check back with them.

And I would second other suggestions made in this thread. I'd design the board to be as out-gassable as possible and I'd bake it in a vacuum system (doesn't need to be UHV) at a higher temp than the bakeout temp of the destination system.

robert

I actually _worked_[1] at a place whose product was essentially based on UHV. One day, one of the engineers came into the office bitching that someone had specified a feedthrough or some part for inside the bell jar, and he said, "It was _brass_ " with a demeanor of utter disgust.

Turns out zinc outgasses forever. And contaminates the whole system - you need to take such dramatic heroic measures to get the zinc out that it's cheaper to just toss the whole thing and start from scratch.

Cheers! Rich [1] a little dig on the academicians who have never done an honest day's work in their life.

Does AlN have better heat conductivity than BeO?

Or is it mainly that it's machinable without killing the machinist? ;-)

Thanks, Rich

At the UHV place I once worked (ca. 1976.77), one of their instruments was an X-ray source. It had a copper anode with the -10 KV cathode off to the side, and they were having trouble cooling the damn thing (it was essentially 1/2" copper pipe with the tungsten welded to the end.). I suggested a heat pipe and nobody had a clue what I was talking about. Plus, they had sort of a "what does a lowly tech know about anything?" attitude - the company was run by a PHD. I was tempted to make a heat pipe at home and bring it in to demo it, but this was pre-internet, so I wasn't exactly sure how to build one to demo.

If you built a heat pipe, say, 10" long, with copper tubing, copper wool (or asbestos substitute), and water vapor, would you be able to hold it in the middle, torch it at the end, and burn the hand of someone at the other enc? IOW, would the middle get as hot as the condenser end?

Thanks, Rich

If it's a simple circuit, you might be

Whacky idea: seal the proto in a glass envelope, a "circuit in a bottle"?

That wouldn't be terribly hard to do.

James Arthur

You might also want to look at sorption pumps for roughing. Where I worked (the place is now owned by Perkin-Elmer), they only used the cryopump for the Molecular Beam Epitaxy rig. But, if your ion pumps and getter pumps do their job, you can get a mind-bogglingly clean vacuum.

My association with MBE was working on the control sequencer, which the geriatric PhD in the lab had designed using relay logic. =:-O

Have Fun! Rich

(BTW, before you offer me a job, I should tell you it'd cost a couple thou to relocate me, and I only know enough about UHV to be dangerous. ;-) )

Speaking of PhDs, the company parties sounded like a Phi Beta Kappa convention. ;-D

Cheers! Rich

If you find a decent glass blower, some of their results can be indistinguishable from magic. ;-)

Cheers! Rich

BTW, John:-

A helpful fellow (happens to be a scientist who founded a major Mass Spectrometer mfg. company) told me the standard reference is (I think) O'Hanlon. I should see him again next week if you have any specific questions.

It's not as good but close, just about equal at some temperatures.

Yup.

BeO still has the best ratio of thermal conductivity to dielectric constant, except for diamond.

John

Not so wacky! If you have a university nearby with a vacuum and glass shop. Maybe Kovar pins into a glass envelope?

Brings back memories. Haven't done any glass blowing/vacuum systems since around 1960.

...Jim Thompson

Mass spec company? Which one?

O'Hanlon is on order. Thanks.

I guess the question is just, what's the best way to package a modest (say, 20 parts or so) electronic circuit for UHV.

John