PCB's in liquid nitrogen

ould

nt

rs

Robert, Thanks for the tip on staying with lead based solder. I was also wondering if thinner traces would work better? Letting the copper 'move' with the pcb.

I like the bigger traces, because people may be changing components and the heavier copper will hold up better to resoldering.

George Herold

What are you going to use as an amplifier? It will have to be pretty good to measure Johnson noise with any accuracy.

Some tuned phemt amps can hit noise temps around 40K. Wideband untuned amps usually run around room temp.

It's easier to go hot than cold!

John

Thanks Mikko, I'm definitely concerned about moisture, mostly leaking into the probe when it=92s inside the nitrogen dewar. The probe will have at least one hole in the bottom. (I don=92t want to make a nitrogen bomb!) And will spend part of the time in the vapor above the liquid. The current idea is to try to make a =91soft=92 seal around the top of the nitrogen dewar and control the nitrogen boil off such that no water can leak in. The fall back idea is to use a Plexiglas tube, sealed at the top to the probe and open at the bottom, which extends down into the liquid. (again sealing the probes environment) Either way you=92ll be able to lift the probe up apply heat and let it warm up dry. (and maybe 1/2 of the people will follow our instructions.) I like the plastic bag idea! In the past I=92ve always used probes that were vacuum tight with helium exchange gas.

George Herold

temp

d

ll

e . t e

re

ce

Wow, thanks again Mikko,

At $200 each I can't afford Lake Shore sensors. I'll let you know what I find. I have it on good authority that the regular Si diodes will work fine. I=92ve got a few ideas on how they might be calibrated, and hopefully a two point calibration will be enough for 1% accuracy, which is about all I=92m expecting from the electronics.

We use to use Dale carbon composite resistors as temperature sensors. (But not for sensitive areas, they tended to change a bit with each trip down to 4K)

George Herold

mp

rmally

The amplifiers and filters will all live at room temperature. (Nothing fancy here.) The filter maximum bandwidth maximum is 100kHz. (A state variable filter with 1% C=92s and 0.1% R=92s) The amplifiers bandwidth is up near

1MHz. Except for the first stage where the bandwidth changes with the source impedance. The electronics seems to be working amazingly well. (I always worry that two compensating errors are fooling me.) If you=92re interested I could post my latest shot noise data on Monday.

George Herold

mp

rmally

ce

Price is important, We can always make our own from resistance wire.

George H.

Maybe larger pads will help, but a heavy hand at soldering / desoldering is going to (eventually) lift a pad. A larger trace to a larger pad might allow a "floating" pad to be re-used due to the greater possibility of that trace remaining adhered to the PCB. As far as re-flow goes, have the initial PCB be "HASL" or hot-air solder levelled (tin-lead for low temp work). If you can score any Multi-core Savbit, that would be the ideal alloy to use, as the 2% copper in the solder will not change the MP enough to talk about, but will prevent "eating" (dissolving) away of the copper pads.

...is that what she said??

If there are a few components that will be changed regularly, put several pads in parallel next to each other so if one gets messed up the board isn't a loss. Start at the far end, and if a trace or pad lifts just cut it off and move over to the next set.

----- Regards, Carl Ijames

e:

id

temp

e

thermally

l d m

ill

he

t 10

tance

=A0"The Journey is the reward"

//

speff.com-Hide quoted text -

I think you will find that it changes by about 77/300 between the extremes.

would

HS

rant

gers

s

How about: Solder bus wire to the PCB and then solder the resistor to the bus wire. The soldering cycle will happen to a bus wire and not the PCB.

liquid

temp

are

thermally

commercial

condensed

from

will

the

won=92t

10

resistance

I haven't bought any resistance wire recently. It got to be more expensive than completed resistors for me some time ago, and that was materials cost only.

What "cold" resistor value will you use? And what's the amplifier's noise density?

A good jfet (BF862 maybe) has huge input Z and under a nv/rthz noise. A 1M resistor has Johnson noise of around 130 nv/rthz at room temp. If cable capacitance is kept reasonable, there will be a nice frequency band in which the resistor absolutely dominates.

You can get jfet opamps in the 4 nv noise range, even easier.

John

  quoted text -

The data sheet range of tempco indicates that those resistors are not behaving like a pure elemental metal, or their tempco would be about

5x worse around room temperature. So, I don't think so.

Best regards, Spehro Pefhany

--
"it\'s the network..."                          "The Journey is the reward"
speff@interlog.com             Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog  Info for designers:  http://www.speff.com

e:

id

temp

e

thermally

l d m

ill

he

near

The current plan is to put three different resistors in the probe and switch them at the top. 100k ohm is a nice value and maybe 10k and

1Meg Two feet of cable and the probe have a capacitance of 50 pF or so. The first stage is a single opamp, there's a DIP socket so that different flavors can be tried. Lately I've been using the OPA134 which has a FET input and 8nV/rtHz of noise, (smaller would always be nicer). I do like the 8MHz bandwidth and only 1 or 2 pF of input capacitance. Which Jfet has 4nV? And what=92s the current noise?

George Herold

snipped-for-privacy@gmail.com kirjoitti:

BF862 is pretty good,

. For a low source resistance .

Regards, Mikko

I forgot to mention the venerable AD743 opamp, with

id

le

he

m

=A0I would

ROHS

airant

Rogers

was

Hey Thanks for all the ideas, I also thought I might use a 'turret' post that is 'staked' to the pcb. Keystone makes a bunch of these.

George H.

I'm not quite up to 'rolling my own' front end. I'm sticking with opamp solutions. And then if the students blow it out some how it's easy to put in a new device... The opamp is the fuse ion this case.

George H.