PCB's in liquid nitrogen

Take a look at ADA4817. It's 4 nv and 2.5 fA. Not to mention 1 GHz. You could maybe mount it on one of those commercial adapter boards, and then plug it into a dip socket.

John

d d

or

at

e

Thanks John, Gee, I think I have a few of those that I sampled from analog. I haven't opened the sample box yet. I want to try and use them for a fast photodiode preamp one day (month, year.... It will probably be obsolete by the time I get around to it.) For the current project the supply voltages are an issue. I've got all the electronics running from +/- 15V.

We plan on using the SMD to DIP adapters for trying opamps that don't come in the old DIP package.

Actually the amplifer noise is not all that bad. You can certainly see effects when the additional noise is only 10% of the amplifier noise. Here's a link to the latest shot noise data.

5-09100K.jpg (I hope that works)

The measurment bandwidth was 300 Hz to 100kHz. There is a bit of deviation at high currents. I haven't explored it very much and I hope it that it might be due to 1/f noise in the light bulb.

George Herold

It is curving up. A filament can be surprising... you can send quite decent audio by modulating the current into an ordinary light bulb. So tiny power supply ripples could be doing what we see here. Still, the match is very good.

It's great that you're teaching the brats about Johnson and shot noise. We never studied those in school. I'm just now working on a system where both matter a great deal, and it adds a bunch of constraints. Problems are fun when you have a lot of constraints.

John

John Larkin wrote: : Take a look at ADA4817. It's 4 nv and 2.5 fA. Not to mention 1 GHz.

Ah, one of those new FastFET devices. A bit high 1/f corner although. Thanks for the hint.

Regards, Mikko

and

and

pF or

that

134

s be

ut

s

The power supply running the light bulb is filtered for low noise. (about 4nV/rtHz.) When I was taking the data at the highest currents I noticed that I would get intermittent 'large' spikes in the rectified noise voltage. (The final stage is an analog multiplier, followed by low pass filter) I would wait for the system to recover from these before recording the number...but I thought (later that evening) there may be other smaller 1/f type noise spikes coming from the light bulb. I want to use other light sources also.. But for the moment I claimed success and moved on. I've got to make sure I can measure the noise from a resistor on the end of a cable cooled to

77K. And that's a lot harder than measureing shot noise, which is hugh in comparison.

Yeah, I don't think I understood any of this when I was taking 'real' data as a grad student. I remeber these noise contours in a PAR or SRS preamp manual, showing the noise as a function of source impedance. I was all greek to me.

George

You need a Phil Hobbs "Light-Bulb Noise Canceller"

John

obe and

10k and 50 pF or

so that

OPA134

ways be

input

. .

th

it's

. m

use

ent

...

Well if it is 1/f noise all I have to do is increase the high pass filter set point up to some higher frequency, (3kHz is the current max.) But regardless of which it's always good to have a bit of hair on the data. It gives me something to figure out...and/or maybe some student can figure it out.

George H.

I didn't notice that Mikko, thanks. I worry about the gain peaks out at high frequency. I assume I can tame them with a bit of C.

BTW I was reading about fluxon's today, thanks for the post to some other ( ...) I would have replied there, but didn't want to encourage him.

I was reading a wiki article

that mentioned electric fluxons. I've made niobium/ oxide/niobium Josephson junctions and measured all sorts of weird stuff. But I always thought of it as a magnetic effect. What=92s with electric fluxons?

George H.

I

: I was reading a wiki article

:

: that mentioned electric fluxons. I've made niobium/ oxide/niobium : Josephson junctions and measured all sorts of weird stuff.

A colleague! Pleased to meet you.

: But I : always thought of it as a magnetic effect. What?s with electric : fluxons?

To me it sounds that either the creator of that wiki page has some confused ideas about dualisms involving flux quanta, or maybe 'electric fluxons' are really constructs of some obsucre branch of theoretical physics. You get fluxons in the traditional sense when electrons get a chance to chase their own tails in an environment where they don't scatter and therefore remain phase coherent (e.g. Ahranov-Bohm effect).

I have liked to think electrons and fluxons as two complementary aspects of the same entity. Electrons are localized in (insulating) free space, and dissolve into a 'fluid' in conductors. Fluxons are localized within perfect-conducting material, but become 'fluid' in free space. But this is just nonrigorous idle thinking...

Regards, Mikko

Oh I hope I didn't mislead you. I made these junctions long ago as a grad student in physics. We etched or sanded niobium wire, twisted two pieces together, then slapped them in epoxy (stycast 1266) and dunked them into liquid helium. I made perhaps 10 or 20 of these of which maybe 2 or 3 actually worked nicely.

Ahh, good the article seemed a bit weird.

Hmm, I haven't thought about super conductivity for a long time. I've read that charge and the quantum mechanical phase are 'complementary' variables in QM. (I'm not sure complementary is the right word.) I've never understood that at all! But perhaps fluxons are a way to get at that relationship.

All I can remember is that I didn't really 'get' the physics of Josephson J's until I read the discussion by Feynman in volume (3?) of the "Feynman Lectures". (He makes everything 'seem' so simple.)

So are you making squids or something related?

Say have people made Josephson junctions in HTC super conductors? Just thinking it would be nice to show the effect without having to use liquid helium.

George Herold

(I work here,

if you're interested)

Sure, there are HTc SQUIDs, but they are significantly more noisy than the Low Tc ones. Great for educational purposes.

Ask Mr Squid!

John

Opps my mistake, I've seen that already at the trade shows...

(I must be getting old... How old do I have to be before I can use that excuse?)

George

That's unlikely; that formula (Drude theory for a classical metal) only applies to pure metals (and that is the reason that 3000 ppm/C is a common PTC 'thermistor' value).

The tempco misbehavior that WILL be important, is of carbon resistors (film or composition). Avoid them. Carbon is a 'semimetal' and becomes insulating at low temperatures.

Carbon comp resistors are the poor man's cryo temp sensors. They need to be individually calibrated for decent accuracy. They work in high magnetic fields, where some other sensors (like silicon diodes) don't.

John

snipped-for-privacy@gmail.com wrote: : >   A colleague! Pleased to meet you.

: Oh I hope I didn't mislead you. I made these junctions long ago as a : grad student in physics. We etched or sanded niobium wire, twisted : two pieces together, then slapped them in epoxy (stycast 1266) and : dunked them into liquid helium. I made perhaps 10 or 20 of these of : which maybe 2 or 3 actually worked nicely.

Great, the pioneer's way of making them. Some of the very first SQUIDs, made by Clarke's group in Berkeley, were just solder blobs over oxidized Nb wire.

: So are you making squids or something related?

Yes, we have a full foundry (

) where we make those, among other things.

Regards, Mikko

Hi Mikko, I remember first trying the Clarke slugs. (solder + Niobium) but IIRC they didn't work very well. Whereas the niobium on niobium had 'text book' I-V curves with nice sharp self induced steps.

Wow, lots of cool stuff where you work (micronova) is this a government run facility? Who pays the bills?

George Herold