DIY Large thermal chamber for metrology T&M testing

[Vtile]

HoHoHold your horses! Is that R13-value in imperial pseudo units ... Not the real units which Yankees refer sometimes as RSI-value.

[dietert1]

One difficulty i found when using multiple TECs was mounting them with good thermal contact. Usually i glue them in between, but was wondering about thermal expansion/contraction during usage. Once i made the mistake to glue two TECs side by side in between one heatsink and one oven.
If you use grease, it may work for some time, but grease also has a tendency to become glue after many temperature cycles. Watch out with multiple TEC devices!

Regards, Dieter

[Anders Petersson]

Dont forget to allow for the significant loss of available pumping capacity at higher delta T.

If you meant my calculations, I did pick the Qc from the "datasheet" based on 20 K delta-T. As you say, the theoretical pumping capacity goes to zero somewhere around delta-T of 60 K no matter the power applied.

HoHoHold your horses! Is that R13-value in imperial pseudo units ... Not the real units which Yankees refer sometimes as RSI-value.

I don't know what they mean by R13 so I picked an assumed W/mK value.

One difficulty i found when using multiple TECs was mounting them with good thermal contact.

TiN's picture of the heat sink assembly shows the water blocks are individually pressed in place by screws, and I assume the TECs are pressed at the same time. That should solve the problem of allowing expansion...?

[mzzj]

I highly recommend playing around with Kryotherm peltier calculator software https://kryothermtec.com/kryotherm-software.html
It allows you to "build" your cold box and cooling system.

But even without starting the calculator I can tell you right away that 900W heat load with peltiers would be madness.

To cool 900W heat load to 0 degree C would take some serious amount of hardware:  30x Frost-74 peltier modules parallel, 30x LARGE cpu heatpipe coolers on hot side and 30x smaller cpu coolers on cold side. 
"Efficiency" would be actually pretty decent thanks to several parallel modules run at lower than rated voltage. You would need only 1.3kW power supply for the peltier modules. 

[TiN]

Madness it is. I don't want to deal with phase-change refrigeration in this project. And watercooling loop capacity for peltier is indeed fit for 1500W.

Also 0C at 900W maybe not possible, but that's rather corner case. I'd be happy with +16C at that power load (to allow testing 18-50C range for commercial T&M gear) and use colder only for special cases (when load is much less).
Specs are not written in stone, but rather a best target to get. Temperature uniformity and stability in practical 16-40C range is most important for me here.

Anders Petersson
Calculation is fun, but I don't think it align with reality here much. Right now chamber at +17 degrees (ambient is 23.5) and K2510 reported power for TEC is 11W.
When chamber temperature was at +38 degrees (same ambient) K2510 used about 18-20W. So it is at least twice less than your calculated 38W heat loss.

Also I don't need very large temperature difference and fairly confident that coolant water could be kept around +30..35°C max. So that gives delta-T about -40°C at maximum (Tset 0C from +30 coolant temp + ThetaJ) or +30°C (Tset +60 from +30 coolant + ThetaJ). Some good TEC modules are rated for 120-150W at Delta-T 40°C, so its not completely out of the question.

[Rerouter]

For a wide range of thermal enviroments. The way I have seen it done in the past was TEC's mounted to a refrigeration loop to make it easier to work within the delta-t requirements. As it is a cooling dominated system. Turning off the TEC-s with any signifcant heat load will make the container heat up. As for cooling. The refrigderant loop can help keep the delta t to a minimum for the highest possible heat load. While having the faster response time and precision than a compressor cycling.

[Kleinstein]

The calculated power was the thermal power to be delivered by the TEC. The power reported by the K2510  should be the electrical power going in. Especially at low temperature difference the efficiency of the TECs is not that bad, so that the thermal power can be higher than the electrical.

Modern good TEC modules seem to be at around 60-70 K maximum difference and thus halve there nominal power at 30-35 K difference. With a reduced current (because of less than perfect heat sinks) the maximum dT would be reduced somewhat. So 120 W cooling at 40 K differnece would be a really large TEC module (like 300 W nominal power).

If one needs high cooling power it would really help to lower the cooling water temperature. If it gets hard to keep it cool, it would help to reduce the current use, as this improves efficiency, at not to extreme temperature differences. Best efficiency at 40 K dT would be at about 2/3 the nominal current. With relative poor cooling of the hot side, the highest power would also be at less than the nominal current, as this would cause less heating of the hot side.
 

[TiN]

Yes, in the final heat exchanger I do plan to use large 340W TECs (most likely 6 or 8 chips), when we get there.

Here's latest chart (with just 40W TEC).



We can see it chocked up at around +13 °C point. K2510 reports 22W used now, and temperature drops a little, but I think at this point it reached the capacity.

[Anders Petersson]

Calculation is fun, but I don't think it align with reality here much. Right now chamber at +17 degrees (ambient is 23.5) and K2510 reported power for TEC is 11W.
When chamber temperature was at +38 degrees (same ambient) K2510 used about 18-20W. So it is at least twice less than your calculated 38W heat loss.

The power when cooling is in line with my calculation, given that the delta-T of 23.5-17 = 6.5 K is 37% of the 20 K I calculated at (37% of 38W is 14W).
For the heating scenario of +38 C (delta-T of +15 K), the heat loss should be 15/20=75% of the 20 K value, so 28 W. Perhaps I missed something? The table could be isolating. Or perhaps there's a T difference between the inner side of the wall and the location of the thermometer.

Anyway, I hope I don't sound negative. I'm just trying to help out predicting what performance you can expect. No doubt it's an interesting project, useful regardless of the limits of operation.

[dietert1]

In your design, the inner chamber serves as an air guide for the vent, is that correct? So it can be a thin material.

Recently i also thought about making a temperature chamber for our Fluke 8502A DVMs (references without oven!). I wanted something without fans, so i ordered some large aluminum sheets to put the instruments on. On the back of the chamber i planned to mount a large heatsink we already have, like 40 x 80 cm and then the TEC elements between the aluminum sheets and the heatsink (with some mounting parts). This way i intended to have a stack of instruments where i can adjust the temperature of each instrument to a nominal 23 °C. It's a different application - not meant for temperature sweeps but for ambient temperature control.
Of course the whole procedure starts with mounting temperature sensors into the old instruments. Maybe then the TC can be determined and corrections applied without building the chamber..

Regards, Dieter

[Anders Petersson]

In your design, the inner chamber serves as an air guide for the vent, is that correct? So it can be a thin material.

Yes, the inner chamber makes the air flow in a single direction over the DUT before the air is (unfortunately) warmed by the fan and returns via the outer chamber, on three sides. The outer aluminium plates just distribute the cooling to the 24pcs of 9x9cm heat sinks that fill the outer chamber. Any air parcel has to pass along a 9cm fin of 4 heatsinks on the return path through the outer chamber so it's a lot of air-to-heatsink contact.
The inner chamber can be thin (plastics), except the weight of all heatsinks makes it sag a bit.

Recently i also thought about making a temperature chamber for our Fluke 8502A DVMs (references without oven!). I wanted something without fans, so i ordered some large aluminum sheets to put the instruments on. On the back of the chamber i planned to mount a large heatsink we already have, like 40 x 80 cm and then the TEC elements between the aluminum sheets and the heatsink (with some mounting parts). This way i intended to have a stack of instruments where i can adjust the temperature of each instrument to a nominal 23 °C. It's a different application - not meant for temperature sweeps but for ambient temperature control.

Yes, sounds reasonable up to some unknown level of instrument self-heating. Maybe you can attach heatsinks to the inner walls and ceiling, like the 9x9cm ones I used. The attached pictures show how I pressed the heatsinks to the aluminium sheets. There are probably better ways. Thermal glue would be much much easier.

[leighcorrigall]

Received Edwards E2M8 vacuum pump for future pressure chamber experiments. I expected it to be somewhat smaller, but its quite a hefty 24 kg monster.
Gotta learn a bit about vacuum gear and buy more accessory like oil mist filter, KF25 flange adapters, pressure valve ports, solenoid valves, etc.. But that's a story for another time, not a current priority for thermal chamber work.

Hi TiN,

You may want to consider a diaphragm roughing pump to avoid oil problems.

Regards.

[MK]

Also. for the TEC mounting bolts you must use stainless steel for its low thermal conductivity, and make the bolts as long as possible to reduce the heat leak back through them. Also belleville washers help you to control the clamping force so that you dont crush them.
And if mounting multiple TEC's to a heatsink lap them to the same thickness, or one might be crushed before the others are clamped.

[TiN]

Meanwhile I have received TEC modules (two units , each with 3 x noname TEC-12706 68W chips, with premounted cheapie aluminum waterblocks, some 8mm tubing.
I've cut some channels in styrofoam sheet for tubes and cut holes for waterblocks. Then this is mounted and glued onto inner rear lid (which is 101.6mm thick).
Tubes exit thru the lid on the back. This is rather a test drive concept and most likely will be refined and redone once I have idea how chamber works with 360W TEC combined power.



I also moved box to the lab, and had to modify and raise shelf on my engineering bench. I do love my 4545 extrusion, very configurable at any time. 



Rest of the parts (1500L/h pump, fans for radiator (SanAce 120 100CFM x 4 pcs), fittings and insulation for tubes) should arrive in next few days. Then we could do a wet run with PSU and Arroyo 5310 as controller.



I've also glued another 50.8mm thick sheet of foam board on top. Some more insulation will be added on the sides as well later.



I also need to think thru on how to run cables out of the chamber.

[TiN]

Received water pump today.

[leighcorrigall]

My own attempt uses a complex design with an inner and an outer chamber for max temperature uniformity. I've only done initial testing yet, with mixed results.

Now that others have shared their design, allow me to share my own 'work-in-progress'. Shodan and TiN, thank you for this Peltier-inspired idea.

Components:
- A Fisher Scientific desiccator cabinet with two shelves. For those who do not know, these air-tight enclosures are used to reduce humidity. The lower shelving area will accommodate some of the hardware and it has not yet been insulated.
- Modular aluminum foil sheathed insulation blocks to fit inside the cabinet and accommodate various sized setups. Insulation was sourced from the abundance of packaging left outside my apartment. Styrofoam and cardboard are cut to size and then carefully wrapped in aluminum before adding a protective Scotch tape layer. An extremely messy process, but worth it. The blocks are assembled by compression -- no double-sided tape or glue to hold them to the cabinet.
- A NI cDAQ-9174 with a NI 9216 temperature module will be implemented for monitoring the control volume and DUT. 100 Ω RDT probes can be found on eBay for 25 USD/unit, including shipping. Luckily, I just happened to remember that this module was lying around.
- Two Cooler Master MasterLiquid LC240E (55 USD/unit) coupled between a Peltier device (TEC1-12706, 6.5 USD/unit). The water heat pump concept is excellent because the required through-holes for the coolant loops are small and do not compromise the insulation much. The active pumping of the water allows for heat to be extracted to the external radiator efficiently. Coolant loops can be insulated to prevent thermal leakage with off-the-shelf pumping supplies. An advantage of two water coolant loops is that the Peltier device can be located outside the main testing chamber.

This environmental chamber was the natural course of action when faced with measuring TCRs and VCRs of high-precision resistors for a project that I am working on with Garrett M. I think I have approximately 33 cm x 33 cm x 40 cm to work with. It is not exactly 'large' by comparison, but it will fit most things that I am interested in testing. My favourite part of this idea is that I can partition sections to optimize thermal response time with a temperature controller.

Enjoy!

[IanJ]

Hi all,

Here's mine, it's quite small and is used to test hand held sized products.
Made from the same foam board type insulation, or KingSpan as we normally call it here.
I didn't use any glue, just plenty of metalized adhesive tape as you can see.
There's a wee removable door on the front.

I used two peltier coolers from Ebay, and since the pic it has a switch on the top now to control either Off, Cooling or Heating (reverse peltier) mode.

There's no control circuit as such to set the temperature, never got around to that yet. I just stick a temp probe through the side to record the internal temp.

Ian.

[Grandchuck]

Here is my feeble attempt.  The polystyrene foam box is 300mm x 200mm x 200mm.  The temperature drops from an ambient of 25 C to 13 C.  That seems feeble  .  Ian, it looks as if my Peltier module is similar to the ones in your cooler.  What kind of performance would you expect from a single module and a similar volume?

[TiN]

Alright, small update. Setup is live now.

2 liters of water filled into cooling loop, fans installed, wiring mounted and fixed.



First test, 200W cooling with fixed CC (two power supplies, HP 6652A and HP6653A for rescue). This dropped temperature in chamber below +6°C.
Measurement is taken by calibrated Fluke 1529 Chub-E4 thermometer + Omega RTDCAP-100 Pt100 sensor.

Next I've connected actual PID temperature controller, Arroyo 5310 (100W capable, 10A/12V unit) to 3 peltier elements (1 module) while second module with remaining 3 TECs set to constant 5W cooling by HP power supply (to avoid heatleak from outside). Arroyo TECsource controller has own temperature sensor, which is YSI 44006 precision thermistor in this case.



Usual RPi3 + Python app with xDevs.com teckit open-source app was used to run a temperature sweep:



Settings for the run are below:

Code: [Select]

[testset]
mode                = none     ; Execute resistance delta measurement if delta3 or delta4 (H1-2182A), dsb_v for DSB-nV
all_test_disable    = false      ; Run bogus data if true
sv_start            = 17.400     ; Chamber start temperature
sv_end              = 17.400     ; Chamber end temperature
peak_temp           = 35.000     ; Top soak temperature
delay_start         = 0          ; Delay before any operation start, seconds
slope               = 8          ; Hours, Time for slope (symmetric positive/negative) ramp
time_start          = 2          ; Hours, Initial hold temperature time, before positive slope starts
time_dwell          = 1          ; Hours, Dwell temperature duration time at peak-start/2 temperatures
time_hold           = 8          ; Hours, Hold temperature duration time once reached peak_temp
time_end            = 2          ; Hours, Final temperature duration once rampdown finished
slope_shape         = lymex_step ; Advanced shape type, lymex_step = soak time_start in middle of the ramps

This will take about 30 hours to perform full sweep, from +17.4°C to +35.0 °C and back.

To make it more fun, I've put small low-power DC Voltage reference device inside the chamber and connected it's output to usual 3458A's and 34420A.
Measurement is taken by calibrated Fluke 1529 Chub-E4 thermometer + Omega RTDCAP-100 Pt100 sensor. You can see a sensor with blue tape at the body of the DUT.



This reference is MNIPI N4-100 zener-based device, capable generating +10.000, +1.000 and +1.018 VDC. It is manufactured in 2006 in Belarus, and cost only $1700 USD MSRP so rather budget device (comparing to usual Fluke 732 stuff that is around $10k). N4-100's annual drift specified at unimpressive 4 ppm/year, despite being designed around Linear LTZ1000A super-zener chip.

It has option for custom 10 cell NiMH 24V battery pack but natively powered by mains, rated for 220VAC 50Hz. I don't have battery for it, so it is powered from my Chroma 61604 programmable AC source.

Overall view during benchmark:



Live data (clickable for interactive plot, updated every 5 minutes)

[MegaVolt]

Are there any plans to look inside the MNIPI N4-100? Maybe there are other data about him? Long-term aging?

[MegaVolt]

You will be surprised, N4-100 based on LTZ1000 with classic Soviet semi-precision 25ppm/C resistors.

[MegaVolt]

No problem here "MNIPI black magic" makes are N4-100 extremely stable.

Any details? New topic?

[bsw_m]

The PCB has a universal footprint for 2 types of resistors: S2-29V or metalfoil RFP-2 for LTZ1000.
Dividers for 10V and 1V uses DN-xxx type dividers (metalfoil RFP-2 resitors) or NRMP dividers  (microwire).

[bsw_m]

S2-29V

Yup... 25 ppm/C without long term stability.

Please read the datasheet )))
https://eandc.ru/pdf/rezistor/s2-29v.pdf

[serg-el]

I don't want to offend anyone, but take a closer look at the data tables.  ± 300 are indicated for the worst in the temperature range -60 ... + 20.  For the best, no more than ± 5 is guaranteed in the range of +20 ... + 70.
I won't say anything about long-term stability.  I did not take such measurements.