The Cryogenic P-N Junction

[uwezi]

However, your last paragraphs would imply that - if you only would make the slab of silicon thin enough - you still would expect to see a flow of current and an external voltage over the diode.

The paragraph that starts with "However..." or "Really..." ?

"However..."

[CaptnYellowShirt]

However, your last paragraphs would imply that - if you only would make the slab of silicon thin enough - you still would expect to see a flow of current and an external voltage over the diode. But this is not the case. It would still defy the other basic law of physics, the conservation of energy.

Ah, I see what you mean. I'm going to put that one on in the back of my head and let it simmer for a while.

[CaptnYellowShirt]

I've been walking around my lab for the last 20 mins with my electrometer in hand performing what to an outsider might look like a ghost-hunt.

I've hooked up the various diode configurations I've tried over the past week and seen if location makes a difference. Long story short, I've been plagued by a combination of two problems.

1. The 'silvered' zener diodes (5.6v) are light sensitive. It's not something which is readily notable until you hit it with a *really* bright light. In this case, the application of light becomes notable when using one of those premo 3w LED flashlights up close. But it does look like what I was measuring in the case of the zener tests was photoelectric current -- which would explane why they add in parallel. 

2. A CFL bulb that is positioned over my shoulder on the ceiling has been emitting a pretty horrible RF noise. It's not something I checked in the earlier  tests because it was like 3m away from my work bench. I thought, hey if none of the lights ON my desk or within arm's reach is showing an effect, how could this one 3x the distance away do anything? Well it does. And I think this explains the 'regular' diodes adding in series. It would seem that the pick-up loop formed by the diode ring needs a characteristic diameter to pick up this bulb's RF. One or two diodes doesn't really cut it, but 3 or 4 is big enough.

So taking the Faraday cage advice, I put the electrometer in a steel bucket and put a steel plate over most of the top of that bucket. And wouldn't you know it... no current.

I couldn't seem to produce any voltage or current by way of a thermometric effect. I heated either side of the diode's leads using a soldering iron. The only response I seem to get there is from the pointed tip of the iron "looking at" the lead. It might be that there's is a thermal response, but it would seem to be overwhelmed by (what I'm guessing is) a electrostatic effect.

So It would seem I need to do a few things before moving onto my next stage of cold research: 1. whatever experiment I perform needs to be properly shielded and performed in the dark, and 2) in general I need to think about replacing a few CFL's with incandescents... oh wait... I can't buy them any more.... curse you obama!   

[uwezi]

So It would seem I need to do a few things before moving onto my next stage of cold research: 1. whatever experiment I perform needs to be properly shielded and performed in the dark, and 2) in general I need to think about replacing a few CFL's with incandescents... oh wait... I can't buy them any more.... curse you obama!   

What a detective story! Congratulations!

But it's not your president's fault that incandescent ligth bulbs are so inefficient as they are. Perhaps you should jump the step and go directly to DC powered LEDs with the power supply either far away or powered from batteries.

Colleagues of mine recently had problems with 12 V halogen light bulbs in their measurement setups. The measurements (on wafer-level MOSFETs) would be performed in the dark inside a Faraday cage, but in order to position the sample there were halogen lights installed. Never before had this been a problem, but then my colleagues complained that something was wrong with the measurement setup: for every single transistor the gate would blow after the first measurement. 

Then they consulted me.

By accident and with the involuntary help of another colleague we tracked down the problem to the halogen lights!

In one setup the 240 V/12 V 50 W transformer for the halogen light bulb, including the power cord had been installed inside the Faraday cage  But it was not enough to get the transformer out of the cage - no, even when the transformer (made in the 1980s in Eastern Germany) was connected to a wall outlet on the other side of the room we could consistently blow the MOSFET gates inside the Faraday cage just by switching on or off the transformer - no lamp attached 

It proved to be spikes which through power line and protective earth and through the semiconductor parameter analyzer found their way to the gate of the MOSFETs.

The other setup guided the light through a fiber-optic cable into the Faraday cage - but the problem was similar and most probably the same cause.

Since two months now both setups are equipped with 3 W LED lights, powered with linear regulated constant current sources from old-style transformer wall warts. The transformer is never switched - switching of the light is done by slowly ramping the current down and up.

As far as I know not a single MOSFET gate has been destroyed since... 

[CaptnYellowShirt]

But it's not your president's fault that incandescent ligth bulbs are so inefficient as they are. Perhaps you should jump the step and go directly to DC powered LEDs with the power supply either far away or powered from batteries.

Oh no, it's totally his fault -- that there fox news said so. Basically every problem any american currently faces (including, but not limited to jock itch) is obama's fault. 

[SeanB]

You do know most 3W LED units have a rather horrid SMPS inside them? If you want low noise you rip it out and make a constant current supply using a LM317 and a high enough supply voltage, or use the 3 LED dies in parallel with a resistor and drive from 12V. When the power goes out here ( Yay for load shedding, thanks Eishkom for having wet coal and no other contingency plans as usual SNAFU) the RF noise level drops dramatically, and it is not me radiating, I went hunting as well looking for FRI sources.

[uwezi]

You do know most 3W LED units have a rather horrid SMPS inside them? If you want low noise you rip it out and make a constant current supply using a LM317 and a high enough supply voltage, or use the 3 LED dies in parallel with a resistor and drive from 12V. When the power goes out here ( Yay for load shedding, thanks Eishkom for having wet coal and no other contingency plans as usual SNAFU) the RF noise level drops dramatically, and it is not me radiating, I went hunting as well looking for FRI sources.

I was talking about the "naked" 3 W LED chips - definitely no SMPS inside... 

[uwezi]

absolutely off-topic: this is the circuit I used. RC sets the soft-fading, R_set together with the NPN transistor Q2 determines the LED current. Primitive? Sure, but it works!

[CaptnYellowShirt]

For a the experiments I have in mind, I'm going to need a few containers to store the DUT and the LN2.

To take extended measurements, the components will need to stay cold for a long period of time. The containers will have to be well insulated so the LN2 will stick around for at least 12 hours.

I have two candidate containers -- both constructed out of materials from the thrift store.  The first (on the left) is a coffee carafe which has been slimmed down to fit inside of a 64oz Igloo-brand drink cooler. The carafe's insulation method is a double-walled, fully silvered, quartz vacuum dewar. The drink cooler on the outside is (hopefully) to hold in some of the 78K nitrogen gas that will be enveloping the carafe once it boils off.

The second (on the right), is a double walled stainless steel vacuum thermos -- same kind you'd put your soup in for the day. In a similar way to the carafe, its been slimmed down to fit inside of a slightly smaller drink cooler.

The caps for both devices are similar in nature -- air filled plastic plugs about 2" in depth. The caps for both drink coolers can also be installed. Even if they are just thin single walled plastic, I figure they will help hold in the cold nitrogen gas.

Lastly, to give them both the best fighting chance they can get (I really don't want to be filling these things up constantly), I placed them in a small bucket that also has a cap.

To test them, I put 0.5 L of LN2 in each one. My money's on the coffee carafe outlasting the thermos, but I'll let you know how it goes.

Stay frosty!

[CaptnYellowShirt]

Well, the stainless steel thermos sucks.

It killed the the 1/2 L of LN2 in just under 3 hours. The outside of the insulated drink cooler (red) was cold to the touch and collecting frost when I inspected it at the 2 hour mark -- a sure sign something's wrong. I'm not sure if this is an issue with the vacuum (or lack thereof) of the thermos flask, or if the stainless steel simply conducts that kind of heat. Either way, its right out.

The silvered glass vacuum flask fared much better. It kept the 1/2L of LN2 frosty for 24hours (+). When I checked on it again in 24hrs, there was still around 0.1L in the flask.

I repeated the experiment with a full 1L of LN2 in the vacuum flask and that lasted 48hrs (+) -- holding about 1/4L after the 2 days.


I am considering an improvement of making a deeper insulated stopper/cork for the flask out of styrofoam. Any other suggestions?

[SeanB]

The more styrofoam the better, just leave a small gap for expanding gas to escape. As thick as possible walls, floor and lid, with a long path so heat has to travel a long way into it. Find a foam supplier in your area and get a block about a half meter on each side and cut the hole out of it.

[CaptnYellowShirt]

I have built two more glass vacuum dewars in the same style as the (blue) one in the previous post. One is a similar 1L size, the second is larger measuring in at around 2L of storage space.

I've also found a cheap supply of LN2 (e.g. free) but its a self-serve kind of situation. Which means I need to bring my own transfer hose, etc, etc. Not having properly researched cryogenic hose construction, I pulled a teflon-core braided stainless steel hose out of my junk bin and fitted it up to accept the 1/2" male flare that is standard on the 180L pressurized LN2 dewars.

So I was aware going into this that the teflon wouldn't really be very flexible after the LN2 hit it. But I thought, that's no big deal i'll be able to flex the hose into the correct position, turn the LN2 on, and when the teflon freezes I'll just need to uncouple the flare fitting and walk a stiff 4ft hose back to my truck. BTW, cryogenic transfer hoses are typically made of corrugated stainless steel with a stainless steel outer braid. And there's a reason for this...

As soon as the LN2 hit the hose I made, I started to hear this crackling. No biggy right? Well, yeah... until the crackling became just a cracking. What I was left with was like one of those weeping garden hoses that waters plants over its entire length... more LN2 was getting into my dewar by flowing on the outside of the hose than on the inside.

Back to the drawing board... I actually have a set of corrugated stainless hoses in my junk bin, but they have a weird swagelok type of connector on them that I haven't been able to find an adapter for. Going to pay a visit to the hose guy later today. Wish me luck.

[SeanB]

I know those too well........ Went in one day to fill a LOX unit, and found a little 1m snake busy in the room in the cradle where my hands were. Went and unscrewed the vent line and played a little game of cryo snake, and had a stiff scaled stick after 5 minutes. As I was using cryo gloves when picking it up I dropped it and it shattered. That room was a perfect place to go and get cold, as it had a 1kl cryo tank in it that was always venting a trickle of vapour. Nasty thing to do to the smokers was to quietly "borrow" the pack and go and fill it with LOX, then leave it back without them noticing. Next cigarette would light and burn down to a stompie in about 10 seconds. Even nicer, we were prohibited from polishing our work shoes.... Apparently wax polish and LOX do not mix well.

[CaptnYellowShirt]

As mentioned in the previous post, I had a set of corrugated stainless hoses laying around in my junk pile. But because they had an unknown swagelok style fitting on their ends, I couldn't connect them to anything -- even themselves.

So I decided to make some creative modifications. To connect the two hoses, I turned a coupling from a 3/4" stainless steel bolt that could fit over top of the swagelock threads and TIG'd the entire thing together. For the end of the hose, I turned a 90deg flare fitting out of the remaining stock of 3/4" stainless bolt, and also TIG'd that on.

I went back to the LN2 storage area last night and tried my new hose on it. It worked like a champ.

[CaptnYellowShirt]

Here's an update on the 'home-brew' cryo dewars:

I've settled on a general design which encapsulates a glass coffee carafe inside of an Igloo-brand-style thermos. The empty space is stuffed with fiberglass insulation and a cap/plug is made from stryofoam.

I've made a few versions from whatever was available at my local thrift stores. Its interesting to note that there is a HUGE variation on the quality of these carafes that seems to align well with the county of origin. Anything from China or Taiwan hasn't worked at all -- they've just cracked when pouring in the LN2. Ones from Japan, Germany, or the USA have all worked reasonably well, but inside of those groupings there's still a wide variation on the heat gain rate.

The one pictured below came from a Bunn 2L pump-style coffee carafe -- like you might find in a coffee shop. Its done the best so far with a heat gain of 0.584 watts (mass loss of -10.6 g / hr), which blows most commercial cryo dewars out of the water (the large 30L dewar that I transport the LN2 specs in at -67.2 g / hr).

Variations thereof seem to deal with the depth of the 'neck/mouth' section (better performing ones have deeper necks) and the quality of the silvering. Some have a mirror-like finish all the way around. Others have spots where the silvering is thin or missing. The worst performing one comes from a 1L pour-style carafe meant for home use which specs in at 0.832 watts (-15.1 g/hr).

For an up coming experiment I am planing, I've created a 'cold box' to keep long term variations in temp to a minimum. You can see pictures of this below, also. Its made from 1.5" of stryofoam packed inside of a steel cabinet (~1/16" reinforced sheet metal) so it should also offer good rejection of E&M noise.

Currently, I've added 10 wires from a 30awg(?) 3M ribbon cable to each dewar. Its just slung in the mouth of the dewar and out the top of the outer 'igloo' case. I'm trying to get a rough idea of how much heat several copper kelvin leads will pipe into the dewar. I'll report that when I get some data.

[CaptnYellowShirt]

I've built two new cryo dewars using Bunn's "Airpot" 2L style carafes -- so far I've found these to be the highest quality both in terms of resisting cracks and thermal insulation.

I've also changed the style of the caps on all of the dewars from a simple cylinder of styrofoam to a thick block of styrofoam that's been 'carved' to fit down into the neck of the dewar, around the outside of the neck, and then above it by 1.5". I've also glued a disc of aluminum foil at the bottom of the neck plug to reduce radiative heat transfer. These improvements have reduced the heat gain to around 0.52 watts (~ 10% reduction).

The goal of this work is to place a set of components down into the LN2 for long term study. So the next step is to figure out how to 'wire' the dewar. Obviously I'm going to need something that has a fairly low overall thermal conductivity. The experiment in the last post (the ribbon cable) failed because the cable tended to curl up and pull away from the LN2 (weird). So next I tried a set of 29awg magnet wires. I could probably go thinner, but this is the thinnest I had laying around.

I tried several configurations of the wire ... wrapping it a few times around the neck of the dewar to isolate the outside and inside temperatures, etc etc. But there was very little variation between tests. In the end I decided to simply run the wires straight into the dewar under the sytrofoam cap (no need to waste wire, right?). The tests showed an addition of 0.028 Watts / wire. In other words, 18 wires has about the same thermal conductivity as the entire dewar.

To put that in perspective... these 18 wires have a cross section of 1.16 mm^2, while the surface area of the dewar is something like 76,600 mm^2 !!

2402