Peltier enviromental chamber builds

[rhb]

 I found several threads mentioning this, but nothing seemed to be a finished project.  Does anyone know of a documented build that came out well?  The objective is to use one to cycle over a wide temperature range under CNC control and collect TC data from various devices.  I'd like to look at the accuracy of functional approximations.

It seems pretty straightforward, but there are always those little details that get in the way. I have in mind something 4" x 4" x 2" made of 1/2" aluminum plate with a Peltier device and a CPU heatsink and fan. It would be nice not to have to write the software.

[cellularmitosis]

This is what I use:

https://www.eevblog.com/forum/projects/mini-tempco-characterization-rig/

I haven't actually built a full H-bridge for it yet, so I don't yet have the ability to reverse current direction and switch between cooling and heating during the same test run (I have to manually swap the leads).

[CalMachine]

It's not a finished project, including a write up with details on the components used... But, here are a few pictures of my finished TEC box that I use for testing  reference boards in.  Hopefully it might give you some ideas!


https://xdevs.com/cm/TECBOX/

[joeqsmith]

I used a packing box for shipping meat.   I've used it to cool below 0C.    I just run it from the bench supply and hook a thermistor to my meter and run it with Labview.   
 
https://www.eevblog.com/forum/testgear/hear-kitty-kitty-kitty-nope-not-that-kind-of-cat/2225/

[rhb]

Very nice.  Thanks to all.   Because I happen to have some thick aluminum plate, I thought I'd apply a bit of brute force.  Thick plate on 5 sides and a thinner plate on the access side.  Four inches of insulation on all sides except for the Peltier device.

The links should keep me busy all evening. :-)

[texaspyro]

I recently added support for some environmental sensors to Lady Heather.  The code is currently set up to read/plot/log humidity, pressure, and two temperatures (well actually 3 temps... an ambient temperature of the device supports it).  I have it working with the dogratian.com USB-xx devices and the lookingforsoultions.com LFX10xx devices.   They work pretty well, but are not really optimal... sooo... I am designing a (hopefully) better-for-metrology devices.

It will have a BME280 ambient temp/humidity/pressure sensor.  Also a 4 channel ADS1115 16 bit ADC and a REF200 dual 100uA current source for driving thermistors and RTD1000 sensors.  ADC readings (around 200 samples) will be averaged and output every second giving around 20 bit effective resolution.  The BME280 and ADS1115 will be on those Chinese breakout boards plugged into sockets... cheaper than buying the chips.  The BME280 board can be extended on a cable to get it closer to where you want to monitor.  The ADS1115 can be easily changed if (when?) you fry it.

The board will also have a 15 kHz, 16 bit PWM channel driving an analog filter/output and a pair of 50A MOSFETS (one enabled when heat is called for, the other when cooling is called for)... it's not a full H-bridge but the board could drive one.

Heather has a temperature control PID built in that should be able to work with it.

[doktor pyta]

texaspyro
sounds very interesting.

[rhb]

sigh...  I'm afraid I shall be accused of overengineering.   This has morphed into a temperature controlled chamber for my 4" x 5" x 7" shielded test chamber.  Rather more ambitious than my initial notion.

At present I'm considering two  Peltier devices between a couple of pieces of 1" 6061 bar.  Something of the order of a 5" x 6" x 1" bar attached to the top of the chamber and a 3" x 6" x 1" bar attached to two  heatsinks and fans with the Peltier devices between the two.  The upper bar  cut back to allow insulating the bars from each other from each other a little.   Something of the order of 1/2" plate for the top, 3/8" for two sides and 1/4" for the bottom and the ends. All contacting metal faces to be hand scraped flat  to 0.0005" before assembly with silicone grease.

Before I start building, I plan to calculate heat fluxes for the faces to the outside and from top to sides and sides to bottom.  The goal being to make all the interior faces as equal as possible.  Which is part of the reason for the varying wall thicknesses.  The other being I don't have that much heavy Al plate.

For insulation multiple layers of 1"  foil face foam with a narrow (1/8") air gap between foil faces using PVC pipe standoffs and then a a thin plywood outer skin.

I had kept for a number of years several of the meat shipping containers others have used for exactly the same purpose, but in a fit of decluttering tossed them.

I'm going to take a crack at using DesignCAD 3D for this.  If that works out I'll post the drawings.

[joeqsmith]

The added insulation inside the meat packing box was a plus but If I were to do it again, I would make a 2-stage system with higher power TECs.  The camera has been nice to have but every bit of heat hurts the cooling.   It takes too long to get to 0C even with small loads.   

I looked at a buying a small ice maker to steal the cooling system from.     

Vid starts at the meat pack box w/camera.  Not great but functional.
ttps://youtu.be/bNzd5wo6FWU?t=1773

[rhb]

The added insulation inside the meat packing box was a plus but If I were to do it again, I would make a 2-stage system with higher power TECs.  The camera has been nice to have but every bit of heat hurts the cooling.   It takes too long to get to 0C even with small loads.   

Do you know of any references for multistage TEC systems?  It never occurred to me to do that, but the heat flux is much larger for small temperature differences.

For what I would *like* to have, the thermal load is *rather* large.  I think I better calculate the heat transfer assuming perfect insulation.  I may be trying for something which is impractical and need to rethink the requirements.

Many thanks for the heads up.

[ch_scr]

I have completed my build today, it is based around
an Hammond 1590Y (9x9x3.8cm) enclosure
a 4x4cm 12706 TEC Module and
an old CPU heatsink, choosen for its massive Al base plate.
4 M5x20 Nylon screws clamp everything together. Insulation is 10mm Aerogel, used about 20x30cm of the stuff. Gloves and a vacuum on hairline trigger recommended. Easy to cut with a sharp kinfe, can be drilled. Tools without rubber overmolding preferred, the dust loves to stick to it. Dust wipes away with some wet paper towels.
At 12.5V i reached -12°C inside (after about 15 minutes). Outside around the lid felt cool to the touch, so if i were to do it again, i would double up the insulation on top just to have a longer sealing area. The double insulation on the side where the wires come out, with the wires making an u-turn between the layers seems to work pretty good to stop thermal flow (wires don't feel cold).

Imgur Album with build Pictures: https://imgur.com/a/gfpLU76

Now to build some control electronics...

[cellularmitosis]

I have completed my build today, it is based around
an Hammond 1590Y (9x9x3.8cm) enclosure
a 4x4cm 12706 TEC Module and
an old CPU heatsink, choosen for its massive Al base plate.
4 M5x20 Nylon screws clamp everything together. Insulation is 10mm Aerogel, used about 20x30cm of the stuff. Gloves and a vacuum on hairline trigger recommended. Easy to cut with a sharp kinfe, can be drilled. Tools without rubber overmolding preferred, the dust loves to stick to it. Dust wipes away with some wet paper towels.
At 12.5V i reached -12°C inside (after about 15 minutes). Outside around the lid felt cool to the touch, so if i were to do it again, i would double up the insulation on top just to have a longer sealing area. The double insulation on the side where the wires come out, with the wires making an u-turn between the layers seems to work pretty good to stop thermal flow (wires don't feel cold).

Imgur Album with build Pictures: https://imgur.com/a/gfpLU76

Now to build some control electronics...

Nice build!

Where oh where can I get my hands on some aerogel?

[TiN]

I missed the point of using aerogel? TEC in photo have alumina plates which are dielectric already.

[ch_scr]

Just throw "Aerogel" in the ebay search. I was also quite suprised, last time i looked they only had the blue-smoke looking pieces for $$$, now there are these affordable insulation mats. When cutting you not only end up with fibers and dust, but also with what looks like teeny tiny chunks of the blue smoke stuff. I suppose the dust is the same, just smaller. Feels quite strange on skin, but washes off.

I used aerogel because of its supposedly super high insulation capabilitys, really mostly wanted to try it. Also its fireproof and does not take moisture. With my wires being transformer type magnet wire i should be able to go to 80-90° without problems, styrofoam would like that not so much i suppose, at least in the long term.

[TiN]

You may want to consult with your TEC module datasheet. Unlike it is expensive speziol type, you would be looking to replace that TEC very soon, if you go 80-90 °C. I wouldn't go over 60°C on standard modules, to allow for some local hotspot margins. 

[ch_scr]

Thanks for the Hint! Datasheet says 138° max operating temp, but with a few € ebay one, who knows. Will keep it low for now, replacing that sure would be a hassle.

Edit: After looking at a few affordable peltier elements from reputable distributors, their datasheets indeed report 70-90°C as maximum working temperature; the only ones stating 138°C are the real cheap China ones.

Edit#2: The Part number for the High Temperature ones is e.g. TEC1-12706T200 ( add T200 for 200° apparently). Costs about 13$.

[TiN]

I think I have 4 or 5 dead ones already, so I paid my price for learning  .

[rhb]

I'm planning a 2-3 layer stack.  I've ordered 10 of the cheapies via Amazon.  Plan is two 2x2 layers with 1/4" Al plate between them.  From what has been described I should add  a layer of the 200 C units if I want to cover the full industrial temperature range.

Naturally, I'm asking for trouble by building a large enclosure. I may build a small one for a test before starting the big one. It would be a good use for the 2 extra cheapies and if it won't work with just one 200 C unit, there's no point in buying 4 more.

[texaspyro]

What is the proper way to implement a Peltier that is supposed to both heat and cool?   Normally you put the heatsink/fan on the hot side, outside of the box.   But if you reverse the peltier polarity the fan is now on the cold side.

[rhb]

I've not done this yet, so I  could be wildly wrong, but I think the hot side and the cold side are determined by the polarity of the applied voltage.   Whether you heat or cool, you have to transfer heat to or from the other side.

[cellularmitosis]

What is the proper way to implement a Peltier that is supposed to both heat and cool?   Normally you put the heatsink/fan on the hot side, outside of the box.   But if you reverse the peltier polarity the fan is now on the cold side.

I think typically the non-fan side is an entire enclosure, so there is effectively a large heat sink on that side already.  You could also mount a fan on the inside, but likely you'll be trying to avoid air drafts and such.  Depends on the application I suppose 

[texaspyro]

If the cold side is in/on the enclosure there is a limited amount of air/moisture to condense.  Once you swap the polarity to heat-mode,  the fan will be pumping lots and lots of nice juicy air onto the cold side... doesn't seem like that is a good thing...

I'm thinking about using the Peltier to cool and resistive heaters to warm.

[rhb]

I received  10 TEC1-12706s.  I plan to use 8 of these for the large chamber along with 4 of the high temperature units.  So I'm building a small trial system using the 2 extra low temp units.

I'm using 4" drawn steel surface mount electrical boxes for the trial chamber  The TEC will be mounted on the lid so it can swap between the two.  The TECs are in series with a 1/8" aluminum plate separator to equalize thermal transfer.  The bottom plate is a 3.5" square piece of 1/4" Al plate.  Both plates were hand lapped using silicon carbide (aka wet or dry) sandpaper on a surface plate using the traditional figure 8 method.  Were I able to find my spotting paste I'd have scraped them, but this should do.   Dow Corning 340 silicone grease on all surfaces.  I also lapped the bottom of the CPU cooler and the TECs.

I'm still debating how tightly I want to seal this chamber.  The big chamber will be a complete seal with provision for injecting dry nitrogen gas to eliminate humidity and resulting frost which impairs the thermal efficiency.

Getting everything lined up properly has proven challenging.  So tomorrow I plan to make some tooling to allow me to get everything aligned correctly.  With the tight tolerances, stick slip of the grease has been problematic doing it with a small dowel.

If anyone has suggestions about TEC controllers for this I'd be very interested.  Some that people have referenced are $73 which is rather more pricey than I care for.  Moreover, the series configuration suggests I should control each device independently.  So I'm leaning towards an Arduino or STM32F103 "blue pill".

FWIW I own a mill, but I'm doing all the work at the bench with files although I did rough out the pieces with a bandsaw.  I'm holding under 0.002" on the edges and probably 0.0005" on flatness.  The squareness is gratuitous accuracy but I felt like doing it.     I figure I need a little practice and this is easy.

[cellularmitosis]

Looking good!   

I'll be very interested to see how the stacked peltiers work out -- I haven't tried that yet.

Just to get yourself started (before buying / making a controller), you might just wire all three in series and hook them up to a bench supply in constant current mode.  Maybe start with an amp and go from there.

I do remember something about there being a point of diminishing returns with regard to current through each peltier, and in a stacked configuration I'd imaging the ideal current will be different for each layer, but that's a concern down the road when you are trying to tweak that last 5% out of your system.

[Andreas]

I'll be very interested to see how the stacked peltiers work out -- I haven't tried that yet.

on commercial solutions the size of the colder stack is much smaller (around 4:1 area)
since you have to remove the heat of one peltier. (60 W heat against 10-15W cooling capability).
And you have to take care that you have not too much leakage from the heat sink to the cold side.

regards

Andreas