Pulse Tube Cryocooler (Part 4) - Valve Controlled
HTML-код
- Опубликовано: 4 фев 2025
- This is the 4th part in my video series on designing/building/testing a pulse-tube cryocooler in an attempt to create low enough temperatures to liquify nitrogen (-196C or 77K). In this video, i'll be investigating an entirely different type of pulse tube that uses two valves - one connected to a compressor and another vented to the atmosphere - to create the pressure oscillations in the pulse tube. This is known as a "Gifford-McMahon" (GM)cycle, and is more commonly used in a refrigeration cycle with a moving displacer, but can also be found in use with pulse tubes.
The GM cycle is less efficient than using a piston (stirling-type) to generate pressure oscillations, but has the advantage that it's cheaper and easier to build, since it can be coupled with a variety of different compressors and matching acoustic impedance is not neccesary. GM-type cycles typically operate at a COP of under 5% of Carnot, whereas advanced stirling-type coolers can approach 30% of Carnot. Despite this, GM cycles are the primary type used in terrestrial applications, such as cryocoolers in labs and production
facilities
where modest quantities (under 10L per day) of LN2 are required. Despite the increased energy consumption, this configuration is typically cheaper to install/maintain than stirling cycle coolers, which are usually reserved for aerospace applications, like being mounted onboard sattelites or missiles to cool optics.
For the valve assembly, i used a pair of solenoid valves activated by MOSFETs which could have their timing controlled by an arduino with a front panel display, allowing me to know the exact timing I dialed in, which was very convenient for repeatability. Real GM cycle coolers use a rotary valve connected to a motor inside the pressurized assembly, which is more reliable than solenoid valves.
The entire assembly runs at around 0.5 hz. Typical commercial units range between 1-2 hz, in contrast to stirling cycle units, which can operate anywhere from 20 hz to 100 hz. However, pressure ratios are typically much higher on a GM unit, ranging from 2-3 with a baseline pressure typically from 10-30 atmospheres, whereas pressure ratios on stirling units are usually in the neighborhood of 1.1-1.3. Most GM units will have a high pressure side around 200-400 psi and a low pressure side from 50-200 psi, depending on the cold head and compressor being used.
The lowest temperature i achieved running on compressed air was -83C, but I didn't consider this a valid result, because I wasn't able to repeat it. The lowest repeatable temperatures were in the low 70's, basically matching the performance of my previous stirling-type pulse tube. Considering that the power input required was close to 1 kW, compared to the previous design's power of ~120W max., it's clear that this setup is far less efficient.
One of the major sources of inefficiency is the fact that my low-pressure side is simply connecting to the atmosphere, rather than being connected back to the compressor in a closed loop. This is because my current compressor isn't hermetically sealed, and not meant to operate with input pressures over 1 atm. With my high side pressure at 100 psi, theoretically, the most efficient low side pressure would be somewhere between 40-60 psi.
In the next part of this series, I'll be generating hydrogen and using it as a working fluid instead of air, with a closed-loop system running off a hermetically sealed fridge compressor. This will dramatically increase internal heat transfer, making the whole system more powerful and efficient, and should bring me much closer to my goal of -196C.
Part I:
• Pulse Tube Cryocooler ...
Part II:
• Pulse Tube Cryocooler ...
Part III:
• Pulse Tube Cryocooler ...
Music Used:
Kevin MacLeod - Lobby Time
Kevin MacLeod - George Street Shuffle
This level of effort indicates that you are not married. Very very dense technical information exchange. Your work is STELLAR. Thanks for your sacrifice..❤❤
I was one of the guys responsible for the 4-valve pulse tube cryocooler at the University of Jena. For GM-type pulse tube cryocoolers you have to avoid a too high pressure ratio. It is not wise to go over 3 to 3.5 because otherwise you get a superheating somewhere in the middle of the pulse tube length, because you cannot exchange about more than 30 to 40 per cent of the gas contained in the pulse tube across the cold and hot heat exchangers.
The heat exchangers are very crucial for the performance of such coolers. So for the heat exchange with the environment (aftercooler, cold hx, hot hx) we have used slittet copper parts manufactured by electro-erosion. For the regenerator we have used stainless steel mesh screens which were slidely over sized in diameter (0.2 mm) to get a dense package with no void passes at the rim of the regenerator tube.
The double-inlet version may be more powerful than the orifice or inertance tube version. But the adjustment of the hot end bypass valve is very tricky. In general it is almost impossible to avoid dc flow in the system which lead to enthalpy flow losses. So the 4-valve version is the most powerful one even in the case that it has comparable dc flow losses.
From the viewpoint of efficiency the stirling-type pulse tube cryocooler is better. The best design is the active reservoir system, which I invented in 2010. In this design you can regain a lot of the expansion work with a second piston / cylinder system at the pulse tube hot end and you can exactly set the phase shift between pressure wave and mass flow, like in the case of a comparable alpha-Stirling cryocooler.
That's really interesting. I read several papers about the benefits of active buffers, but i felt it was too complex and I didn't understand them well enough to attempt to build one and make a video about it.
@@HyperspacePirate It is more or less like in the case of an alpha-Stirling cryocooler. I think, active reservoir is very easy to understand. I use SAGE from Gedeon Associates to calculate my designs.
Active reservoir also runs as an engine. In this case the heater sits between the regenerator pulse tube and the after cooler between pulse tube and active reservoir.
@@HyperspacePirate I love the reference to "The internet is a series of tubes" Ted Stevens.
When Applied Science, AND Nighthawk in light are following along, you know this is good stuff.
This is awesome! There are only two channels that make me enjoy looking at graphs and diagrams - this one and Huygens Optics :-)
consider checking out suckerpinch! pretty funny visualizations, even if i don’t remember any traditional graphs rn
agree
Agree! 😉
Agree
yeah, graphs are awesome :)
hidden gem youtube channel.
Excellent.
Calm down, mr. Burns.
Wood gas powered air compressor next? :D
Yup nice video
this is inspiring stuff, "pulse tube cryo cooler in the basement" is now on my list of things I didn't know I needed
Instead of using acid for H2 production, you may want to try Lye (NaOH) and Al foil; it generates copious amounts of H2 and there is no corrosive acid mist or vapors generated, and the sodium aluminate byproduct is much less corrosive than the lewis acid aluminium salts produced as byproducts from most readily available acids. If you want to dry the hydrogen produced then you can pass it through a drying tube packed with just more fresh NaOH, which you can then feed forward as more reactant for future cycles. If you use HCl for example, you're stuck with having to also neutralize the acid vapors and mists, which requires additional consummation of reagents without the potential for feeding forward the reagents as reactants.
i second this
How hard can electrolysis be? I'm sure he has water in a tap and electricity in an outlet, just put them together and voila, Hydrogen, right?
@@MrRolnicek Separating the oxygen from the hydrogen is the issue
@@nedshead5906 But if you just separate the electrodes and make sure the liquid connects them BELOW the height of the electrodes then you get O2 bubbles on one side and H2 bubbles on the other, right?
Yep.. not hard to separate the two right from the initial process,.. i think though there are some other hydrocarbons that are produced making the gasses not totally pure.. really depending on purity of everything in the electrolyte..
Pure science, with all the numbers. Its a voyage of discovery. Taking energy out of a thing is so much harder than putting energy in, and that's what makes it so interesting. 15 and 6% mesh fill, thats supprising. Nice winding the aly tube , I would be scared of kinking. Exceptionally well presented.
pro tip for small diam tubing bending, tape 1 end shut fill with sand/sugar/salt/fine granular material cap the other end and bend carefully. the interior filling keeps the kinks from happening,
Definitely some of the best project and video quality out here! Vastly underrated Chanel!
This is brilliant, I was not expecting to see a 4 part series on this. I love the process you use to explain the design process, its a lot more transferable
Thanks for creating the best Darth Vader soundeffect since sliced bread. Good job, I love these scientific tinkering videos.
I’m impressed that you managed a 100C delta-T with just air and a low side pressure of 1 bar. Mega-props, can’t wait to see more in this series!
thanks alot for not giving up and continuing the project! cant wait to see it working, looking to make iron nitride magnets so this is definitely a project that will help a ton
I run a bunch of MRI systems and helium recovery systems all which use G-M cryocoolers, both mechanical and pulse-tube. This series is fascinating since I don't get the opportunity to really play with operating parameters of these systems and mostly just run them as-is from the manufacturer. Keep it up!
Pulse Tube Cryocooler (Part 5) - Staring at even more graphs!
This is like the most awesome project ever, turned into a crappy power point by that one guy at work.
Calm your horses 😅
Really loving these videos, great to watch along with your progress.
Best video series!
What a great day! Waking up to part 4. Time to learn.
Now that you mention that it looks like something from a Dr. Soose book, ya it really does. Thanks for the perspective that you did not want
6% density is probably plenty. You should do it as a ratio of mass of the air you are regenerating vs mass of regenerator. Even at elevated pressure, you only have maybe 1 or 2 grams of air moving through that thing. I guess it depends on how many liters of air are passing through each cycle. 115 g of steel wool to 2 grams of air being regenerated... you might even go less to see if the reduced flow restriction is hurting more than a few extra grams is helping. Checking the pressure drop across regenerator vs mass through it per cycle could be enlightening.
Very cool, pun intended. :). I want to get one when they become commercialized.
I’m really looking forward to the next video. Ads a teenager i used to build single and two stage compressor coolers for CPUs. Using propane in stage one and ethene in stage two i got to -108C at 200W+ loads. That was using reciprocating (Danfoss) compressors but I think rotary or screw will be better for you in higher pressure applications.
I also have a 2 stage cascade using R22 and ethylene.
What was your power draw for get 200W at -108
best channel on RUclips fr
Quick tips for hermetic compressor, add an oil separator in the dischargue of the compressor, this is to avoid running the compresor with low oil (if you are using gas that is not compatible with the oil).
Don't forguet to make vaccum before charguing with nitrogen to avoid mixing the oil with the moisture of the ambient (this turn the oil acid with the working temperatures of the discharge of the compressor)
Don't run too high pressure with non condesable gases, I tested a R134A compressor with 100 psi of nitrogen to check the metering deivice (home refrigeration appliance) and only worked a couple of seconds before the piston stall, runs normally after purgin nitrogen and charguer with R134A.
Nice videos by the way, I wish good luck with your next interarion :D
I have run my small R134a compressor with nitrogen for hours without any problem. I have tried many gases like methane nitrogen ethylene with the compressor and never had any problem.
@@Exotic_Chem_Lab interesting, what is the model of your compressor and how much pressure you have with the compressor off?
@@tejonBiker it's standard 1/4 HP Low back pressure R134a compressor from old fridge. I was running it at 18 bar discharge continuous.
@@Exotic_Chem_Lab Thanks for the info, I was testing an 1/6 R134A, I can't understand exactly why the piston stall, but solved after shift to r134A instead of nitrogen
@@tejonBiker It might simply be pulling more amps than the electric motor inside the compressor can do? 1/6 HP is fairly small. Maybe that compressor was already getting worn and had a lot of internal friction, and so wasn't able to handle much additional heat. R134a will carry the poe oil through the compressor and keep it better lubed, but nitrogen probably doesn't pick that oil up. The other guy's compressor just handled that less lubrication better than yours I guess, lol.
-100C is impressive. Could you be running in to issues with CO2 freezing and clogging your heat exchanger? I can't imagine it takes much to reduce it's efficiency. Would be interesting to try this on nitrogen instead of raw air.
I have a quick video on a dental drill turboexpander which may be of interest. I managed 15C drop at 4 bar with no heat exchanger.
I'm curious to see your results if you go forward with that project. I tried to do a brayton cryo cycle with a 3d printed turbine, but got frustrated trying to make my the turbines efficient enough to produce a reasonable amount of cooling. I'd like to see what happens if you connect that thing to a regenerator (or recupterator, or whatever they call it in the case of continuous flow)
@@HyperspacePirate Yes turbine design seemed way beyond my level so I tried to find something already available in the right power range.
I've been having issues loading the turbine output as those small motors I've been using as a generator don't seem to last long at dental drill speeds (and produce more heat than electricity). I'll definitely try a counter-flow heat exchanger once I can run the expander for any significant length of time.
I'm currently experimenting with 2 of these turbines on the same shaft to form the usual compressor-expander set up but haven't had much luck yet, I'll do a video if I get anywhere with it.
@@matthewbeardmore For loading the turbine, how about a diametrically polarized cylindrical magnet running in a hole inside an aluminum or copper block? At those speeds, the eddy currents should be pretty significant, so might be able to provide a decent amount of load. K&J Magnetics and Applied Magnets both sell these, and K&J also has elongated ring magnets (cylinders with central holes) that might make mounting easier. Only trick would be to support and mount them precisely enough for them to spin that fast without the whole assembly flying apart.
Very cool project. Industrial helium compressors typically have a whole system to first introduce and then remove oils from the helium gas to lubricate the compressor.
Right now that seems to be the biggest challenge with re-purposing a hermetic compressor from a fridge. I got a coalescing oil separator specifically made for that application, so hopefully it'll do the trick. Separation of the oil is relatively straightforward, but I'll have to do a little engineering to meter the oil return flowrate into the low pressure side of the compressor
Great video, it must be hard keeping track of results and having so many variables you can tweak.
Dude you are awesome, explaining all math involved, love it
Can’t wait till part 5. Excellent video. 👍👍
Honestly I am thoroughly enjoying the consistency of your uploads, and how you thoroughly explain your fault finding process and how you go about problem solving.
I can't say I ever even knew about Stirling coolers before your videos but I am hooked, keep it up brother. 👌
This project has evolved so far. Keep at it!
i would try a small layer of lead balls at the cold end of the regenerator, there are also some other materials u can try, but i think they are not easy to get. I think that s the best backyard PT i ever saw :).
This is by far the most approachable way to get super low temps at home. I have almost everything I need to build this today. ~ -100c is good enough for so many projects. I also have a larger compressor with an auto drain. 😀
I look forward to additional experiments on this.
All that for a chunk of ice; Doc Brown would be proud :)
Absolutely love this series, excited to see what you come up with next!
Don’t nuke Portland!!
Just kidding.
Boy, what a channel. This is exactly the kind of stuff I enjoy seeing (and doing as a hobby!)
Please keep it up, champ.
i absolutely love how nerdy this shit is. i subbed immediately after i saw part 1 ;)
You deserve way more credit great work
im watching your video at 3am :)
really nice diy cryo cooler
Thanks for so the ruff drawings of the puzzled guy that so adequately represents me! 😅
This is so freakin' interesting! I wish I could have lived next door to the mad scientist 🤣. Would have loved to be a part of the build process and testing.
I'm in awe at how much time you must have put into this project. Its all very interesting... thx
I have been following your journey since the second video, after seeing these I am so glad that I was among the last people that got a stirling cryocooler.
If you want to know how to maybe get one dm me I may be able to help.
Absolutly love this series and looking forward to Part 5!
This experiment is so dang cool! Excellent work both with the experiment and the video!
Always excited to see an update on this project. Keep it up!!
This is by far the most interesting youtube series I'm following atm.
If i just had the space i would build and experiment along with the series, but at least i will be entertained while waiting. :)
Can't wait for part 5!
hopefully I can learn from you how to build a hydrogen storage system (compression and cooling is needed)
Great project. You grabbed me from the start. I have always wanted to make my own liquid Nitrogen. Eagerly watching to see if you crack this.
Really nice project, and so impressive taking into account that you're building it from scratch. About using hydrogen as a cooling gas - you have to remember that it will be leaking out of the system quite substantially because of its super small molecular size, and also might be hazardous when combined with electrical valves. Helium, despite limitations caused by price, might be a better choice when taking into account your safety. Also, after longer usage, hydrogen might intercalate metal parts of the system reducing their lifespan.
My plan is to use hydrogen for prototyping since i expect to lose a lot to leakage, but it would be very easy for me to replenish with electrolysis / chemical generation. If i get a cryocooler working close to LN2 temperatures with that, i'll switch to helium.
Looks very promising, can’t wait to see more
I love this channel. I have a suggestion for achieving higher packing densities for the heat exchangers. Use a steel pipe and steel rod with a OD slightly smaller than the pipe ID. Pack the steel pipe with small quantities of the material and compress in a large vice or hydraulic press. It may be easier to achieve higher higher packing densities with stacked thinner pucks of compressed material vs one large unit. You will need to press fit the pucks into the copper tube; using a longer length and making a swag to assist with installing the pucks. Cut off the swagged portion once packed. I don't think compressing the filler material in the pipe directly will work as it will probably swell under pressure.
Omg I can't wait for the next video..!!!!
Woot! I have been waiting for this video. !
Such a cool project ! Can't wait for more.
I wonder what comes first:
100k subscribers
-100K temperature
🥶🥳
100k subs, -100K isn't possible. Even with -100c i think the subs will come first
I presume they meant 100k subs or 100K temp (no minus)
@@V8Power5300 :ø)
Isn't -100K 'just' a drop in temperature, from whatever ambient starting point?
But absolutely - yes, -100K is like going faster than light. It is currently not possible - but who knows if it will be (or was possible) ;ø)
Anyways, so much respect for doing such amazing projects - you really deserve it, and much more!
@@V8Power5300 Wow.. I just got smarter..
Negative temperatures can only exist in a system where there are a limited number of energy states (see below). As the temperature is increased on such a system, particles move into higher and higher energy states, and as the temperature increases, the number of particles in the lower energy states and in the higher energy states approaches equality. (This is a consequence of the definition of temperature in statistical mechanics for systems with limited states.) By injecting energy into these systems in the right fashion, it is possible to create a system in which there are more particles in the higher energy states than in the lower ones. The system can then be characterised as having a negative temperature.
A substance with a negative temperature is not colder than absolute zero, but rather it is hotter than infinite temperature. As Kittel and Kroemer (p. 462) put it,
The temperature scale from cold to hot runs:
:+0 K (−273.15 °C), …, +100 K (−173.15 °C), …, +300 K (+26.85 °C), …, +1000 K (+726.85 °C), …, +∞ K (+∞ °C), −∞ K (−∞ °C), …, −1000 K (−1273.15 °C), …, −300 K (−573.15 °C), …, −100 K (−373.15 °C), …, −0 K (−273.15 °C).
The corresponding inverse temperature scale, for the quantity β = 1/kT (where k is the Boltzmann constant), runs continuously from low energy to high as +∞, …, 0, …, −∞. Because it avoids the abrupt jump from +∞ to −∞, β is considered more natural than T. Although a system can have multiple negative temperature regions and thus have −∞ to +∞ discontinuities.
In many familiar physical systems, temperature is associated to the kinetic energy of atoms. Since there is no upper bound on the momentum of an atom, there is no upper bound to the number of energy states available when more energy is added, and therefore no way to get to a negative temperature. However, in statistical mechanics, temperature can correspond to other degrees of freedom than just kinetic energy (see below).
It is on the Wikipedia page on Negative temperature
274k subs and physics breaks!
love your tenacity!
Very nice content! Tank you from France
Keep up the good work
Man i really love this series! Great stuff!!!
Looks like a bit of a frankenstein by the end but appreciate that you following up every little idea and possible tweak. Reasonable results already, hopefully in the next few videos you can get some serious cold going, good luck :)
very interesting. thank you for the details, and for explaining the inefficiencies
Thank you for understanding and explaining
This is a cool series. I need a cryo cooler cold enough to liquefy hydrogen for my spaceship
Based on 2 decades of work inside cascade chambers professionally, ditch the type K and go with the equally inexpensive type-T (blue) thermocouple which works wonderfully at -200c and beyond. Type K is great for high temp use melting metals, and use down to 0c, but at deep cold, type K gives flaky output jumps on digital meters due to its small voltage change when cold, whereas type T is very consistent and stable at -100c and below.
Excellent videos as always
12:15 "i swear to god it is a functional device and not JUST a device i copied form a dr seuss book" key word, just
I'm sad you didn't get this to work.
Your newest build is amazing, but it's beyond my skill level.
This pulse system seems to be much easier to build, just harder to tune.
Great video!
-Sir! Our equations have 100 different parameters!
- OK, let's painstakingly sweep every single one! I have charts to fill!
Love your work. Please keep it up!
really nice work thanks a lot gor so much information
Knowledge is like having superpowers.
Pretty cool! Maybe recycle the air into some Plastic bags to keep it dry
To run higher pressures in the pulse tube and reduce the thermal conductivity you could use a plastic tube inside of a metal pipe to make it alot stronger. You probably won't get the whole tube insulated from the inside but close enough to reduce all that surface area.
Top notch content as always. Quite inspiring! Good luck with pending experimentation =)
Now I’m not a chemistry expert but I worked at a liquid nitrogen plant and how they do it is they use a air compressor that squeezes and release helium using a pneumatic piston. One side has a heat sink and the other side gets like very cold so they put a condenser coil on it and runs pure nitrogen gas through it cooling it to a liquid.
If you go with a closed loop setup as you described at the end of the video, you will need an oil separator which is, uh, more work:(
I never klicked so fast after seeing the thumbnail.
I freaking live for these series. I bet you're there design wise, just need helium.
Good job.
I like your funny words, magic man.
For the Joule-Thomson process a 4500psi compressor isn't that difficult to make. If you look around online for how the home use compressors for air rifles are made it's dead simple. Crank, linear rail as a crosshead, and two polished rods (Round linear rail from Amazon) as pistons sliding into a cylinder headed with a check valve, a simple port at the bottom of stroke for inlet, and a couple O-rings to seal. Feed it with 100psi or so air from a regular compressor and they can bump the pressure up to 4500psi pretty easy. Just depends on how much flow you want.
The cyclonic steam ENGINE ? Are they GONE? I liked the low temperature ORC for power generation. You know with a scroll expander. ❤❤
@HyperspacePirate
you can try to pre-cool the air with a copper tube and peltier modules stuck to the outside of the tube. and also reduce the resistance of the heat sink in the same way.
Thanks again ! Great content !
You can decrease the valves closing time, replacing the frewheeling diodes by resistors. The value is valve current multiply by desired clipping voltage.
I subscribed, but I really wanted to press the other button!
hi there! Great project. Congratulations on your persistence! One question I need to ask you about your graphic at 16:49 (the GM assembly with helium pump). The helium compressors I've seen have been 3 phase brutes! Can this be done with more modest power requirements?
Could you take the cooled discharge air and use it to cool the intake air thus getting back some efficiency?
GM compressors are pretty much entirely built out of refrigeration components inside, and even standard fridge compressors can withstand a total system working pressure of up to around 30-40 bar, provided the pressure difference across the inlet and outlet are no more than ~8-10 bar.
The only major difference is that a very, very good oil separation *and cooling* system is *mandatory* after the compressor, before the aftercooler, or efficiency drops several orders of magnitude.
And definitely don't use hydrogen, just stick to balloon grade helium.
I'm hoping you succeed with this, because until I saw your project, I thought the only approachable way to build something to liquify air would be the regenerative joule thompson / hampson linde type thing. Although, a vortex tube operated with restricted outlets to maintain the right pressure ratio may be an option instead of pure throttling, and of course there is precooling to help.
This is sick but You know what’s also cool, a submarine kayak pls bro I’m begging you
Can you increase the efficiency of each specific pulse with different shaped outlets from the valve? Like a trumpet bell off of the valve to create a low pressure zone at the valve itself just after shutting? Like old timey carburetor trumpets, that create a high pressure wave when the intake valve closes, and timed to the length of the bell just as that high pressure wave goes pop and creates a low pressure zone at the base of the trumpet with an inrush of air the intake opens reducing parasitic friction on the air flow? In this instance it could possibly cause a little bit of "breathing" through the regenerator mesh instead of 1 way travel. Might require way to much volume of an airbox to be practical.... But hey your way better at the math then i am.
I don't know if you've seen my comments, but I really think you should try scrubbing the CO2 out of your working fluid.
The amount of energy required to condense the CO2 might be at least partly responsible for the long time it takes your system to reach steady state, however the biggest issue is the solid CO2 building up on and insulating your cold heat exchanger surfaces, while also restricting the free space that the air has to pass through. (perhaps significantly enough to choke the flow at the regenerator's outlet?)
Insulating the coldest sections of the loop against the outside air, mostly to prevent water from consensing onto it.
This would also exclude the variable heat loading that might be affecting the consistency and accuracy of your inter-day data, as even just the normal intra-day dewpoint variation can add up to a 2x increase in the amount of heat being pumped by the system.
The CO2 is a problem, but I think the leftover water is still a bigger issue. At a dew point of -40C, the vapor pressure of water is ~26 pa, making it 260 ppm by volume, whereas the ambient co2 is around 350 ppm.
However, heat of fusion of Co2 is ~9 kJ/mol whereas the water has to condense and then freeze so ~41 kJ/mol to condense then ~6 kJ/mol to freeze, so quite a bit more energy consumption overall from the trace moisture
have you considered precooling with the output of the system to increase efficiency over the course of runtime as opposed to trying to start off with maximum efficiency?
Very cool 🤙
I wonder if you could use the two warm parts above the cold end, and apply that heat to the walls one of those vortex mosture removers.
Have you tried adiabatic demagnetisation? That might help you get a bigger ∆T