Environmental chamber build idea

[derZockerMP]

Hello, I want to build some Environmental chamber to test some semiconductor parts for there Temco. Also Multimeters like Keithley 2000 sould easily fit into it. But not bigger to keep the volume to cool and heat low. The temperature range sould be around 0°C to auround 40°C, so a normla TEC model can do the job. Are there any projects out there that fit in my idea?

[Echo88]

https://xdevs.com/article/megatec/ comes to mind.
A repurposed refrigerator should also be capable to do the job, when using the existing inner cooling loop together with TEC powered liquid chiller.
I havent yet realized the refrigerator idea myself, too many other projects have priority. 

[coppice]

Yep, an old fridge, a fan heater, and a little effort over temperature controlling them can produce a great result. If you want lower temperatures an old freezer and a fan heater will take you from -20C to 50C nicely. If you want higher temperatures you really need something that can take the heat. Fridges and freezer usually don't. Even though many freezers have a thin metal liner, the insulation behind that typically doesn't like higher temperatures.

[Terry Bites]

You probalbly don't want your nice and very expensive Keithley covered in condensate in the chiller.
Cooking it dosen't help either.
With sufficient fan assisted heatsinking a TEC can get you down below -20 and above +40

[joeqsmith]

Shown using a PC CPU heatsink with peltier to cool some ICs below 0C.   I wouldn't want to put my test equipment into the chamber, unless they were the device I was wanting to test.
https://www.eevblog.com/forum/projects/waveforms-in-a-74ls04-ring-oscillator/msg1033429/#msg1033429

Normally, I use a discarded meat packing box with added insulation and stacked peltiers.   
https://www.eevblog.com/forum/projects/poor-mans-thermal-chamber-meat-packing-box-ii/msg2454591/#msg2454591

***
Using the PC heatsink setup to temperature test an RF filter..
https://www.eevblog.com/forum/beginners/high-frequency-passive-low-pass-filter-with-cheap-components/msg4618780/#msg4618780

[derZockerMP]

Thanks for the answers. Are there any suitable PID controllers available for this application? Maybe an H-Bridge to control the TEC for cooling and heating?

[ch_scr]

AFAIK the "TCB-NE" (there are other models with different power) is the cheapest "off the shelf" Peltier-controller option. Can be found on AliEx. This might be the OEM.

[ejeffrey]

You probalbly don't want your nice and very expensive Keithley covered in condensate in the chiller.

This and also TEC coolers have *very* limited cooling capacity so you don't really want to burden them with any heat load you can avoid.  Keep your instruments out of the chamber whenever possible.

[derZockerMP]

Thanks, is this controller able to find the PID constants itself?

[ch_scr]

Sadly, not. You can set P,I,D and "update interval". When the latter is a lot faster than your physical system response time, no value for PID will result in stability. Since it's all controlled by a serial interface anyway, the "auto-tune" program could live on the pc side...

[tooki]

Hello, I want to build some Environmental chamber to test some semiconductor parts for there Temco. Also Multimeters like Keithley 2000 sould easily fit into it. But not bigger to keep the volume to cool and heat low. The temperature range sould be around 0°C to auround 40°C, so a normla TEC model can do the job. Are there any projects out there that fit in my idea?

There is a reason commercial environmental chambers don’t use TECs: TECs are spectacularly inefficient. Like staggeringly bad. They only ever make sense where other factors absolutely demand a TEC because nothing else will work. But environmental chambers need powerful cooling and heating in order to give the tight temperature control and quite steep temperature curves needed, and TECs’ characteristics are a poor match.

Technology Connections (which if you don’t know it, is one of the very finest tech nerd YouTube channels of all time) recently made a video about this exact topic, and why regular refrigeration is a better choice in nearly all situations.

https://youtu.be/CnMRePtHMZY?si=pPIsjkKXlebisMxU

[derZockerMP]

I think you're right that TECs are quite inefficient. However, a DIY implementation of regular refrigeration seems very difficult to me. If you have an easy method or a low-cost kit, it would be very helpful.

[ch_scr]

For a very small desktop chamber (Euro-pcb size 100x160x50mm) or a fixture for single components, I think TEC is a sensible solution.
Once we're talking "multimeter fits inside"-size, it gets into "compressor refrigeration" territory, at least in principle.
Refrigeration is a lot harder to DIY, leaving the "comfortable field of electronics"; to control a stable temperature one might need an electronic expansion valve and/or variable speed compressor (and a reversing valve to switch heating/cooling to boot).
IMHO it get's quite complex quite fast, and to DIY a chamber with reasonable effort, even taking the TEC efficiency hit, that still might be the more attractive/reasonable solution.
One could dream up a "hybrid" approach, using a commercial refrigeration water chiller e.g. like is used for desktop CO2 laser (KH 6000 water chiller ~230€)
and use a it to cool a TEC in turn, that then would only have to "move" a small thermal difference.
But the TEC might allow to "smooth out the bumps" of the refrigerator cycling, without complex refrigerator modifications.
But then, for thermal sweeps, the water temp would have to move as well, or maybe some kind of thermostatic mixing valve is enough?

[dietert1]

In 2021 we made a 19" chamber with a large Dr. Neumann peltier cooler and a Cotek AE-800-30 as programmable supply. A polarity switch with four mosfets and a STM32 nucleo module was added to implement the PID. A pretty easy project.
More details and images are here:
https://www.eevblog.com/forum/metrology/diy-large-thermal-chamber-for-metrology-tm-testing/msg3559937/#msg3559937.
The thread is a bit strange as a former contributor deleted all his posts.

Regards, Dieter

[derZockerMP]

Ok, thanks.
Are they any recommandations to find the PID values quickly. Never worked with something like that.

[dietert1]

The PID controller isn't a big hurdle. When i turn on the system the first time with some "guesstimated" loop gains there will be initial overshoot and then some oscillation. One may need several attempts until one gets damped oscillation and can determine oscillation period and damping factor. Then there are rules how to optimize PID gains for high bandwidth or for low noise etc. This can be found at Wikipedia and in many other places.

Regards, Dieter

[ejeffrey]

Once we're talking "multimeter fits inside"-size, it gets into "compressor refrigeration" territory, at least in principle.
Refrigeration is a lot harder to DIY, leaving the "comfortable field of electronics";

It's as much about the active heat load as the size.  This is made worse that many TEC builds are trying to be "small" but still have usable interior volume so they skimp on insulation.

A DIY refrigeration system is probably not something most electronics hobbyists are going to have the skills or tools to do.  But a fridge conversion isn't too hard.  Starting with either a mini fridge or a full size one, you can add a digital temperature controller, fans for air circulation, and heaters rather than a reversing switch, and feedthoighs for wiring.  You will be stuck with on-off control and the associated tempesture oscillations, but with a good controller and some thermal mass you can reduce that somewhat.

I built one as a fermentation chamber / kegerator.  I used the door light as the heater, but you can also use "lizard heaters".  Basically just install SSRs inline with the compressor and the door light, set the internal thermostat to "coldest" so it never shuts off and you are good to go.  Of course having 5 gallons of liquid helps buffer the temperature.

[fourfathom]

True, the Peltier TECs are inefficient, and it's very hard to get a rapid / extreme temperature change, but sometimes you can live with that.  Where I've worked I've used smaller liquid CO2-chilled chambers (electrical heat), and larger walk-in chambers with massive refrigeration systems.  These are quite impressive and useful for thermal-shock testing.

But for my hobby projects I've found the TEC "Reptile Incubator" units
as shown in Dave's video

to be extremely useful for evaluating oscillators (LC, XO, TCXO, OCXO, etc.) and other testing where rapid temperature changes are not needed.  I modified the electronics, keeping the power supply and monitoring, display, etc, but replacing the TEC driver with my own design so I could put the chamber under remote software control.  One disadvantage of the stock chamber is that the thermal setpoint has about a 1 deg C hysteresis, so I was able to improve the setpoint stability as well as run defined thermal patterns.  The stock chamber also has the internal fan cycling along with the TEC, and the air-flow changes affected the DUT temperature -- I now have the fan run continuously.   The useful temperature range of the chamber (with minimal power-dissipation DUT) was about +5 deg C to +60 deg C.

[derZockerMP]

Here are some photos of the build. It's still a work in progress. I'm using four TEC1 Chinese Peltier coolers and getting around 10 degrees with a 25-degree room temperature. That's good enough to test my voltage source build in the 18-28-degree range for tempco.

[coppice]

to be extremely useful for evaluating oscillators (LC, XO, TCXO, OCXO, etc.) and other testing where rapid temperature changes are not needed.

Huh? How are you going to test that the oscillator can maintain the necessary internal temperature control or measurement/compensation when the temperature changes rapidly? I've had real problems with both OCXOs and TCXOs taking time to get the frequency back on track after a thermal shock.

[fourfathom]

to be extremely useful for evaluating oscillators (LC, XO, TCXO, OCXO, etc.) and other testing where rapid temperature changes are not needed.

Huh? How are you going to test that the oscillator can maintain the necessary internal temperature control or measurement/compensation when the temperature changes rapidly? I've had real problems with both OCXOs and TCXOs taking time to get the frequency back on track after a thermal shock.

Yes, there are good reasons to test an oscillator response to rapid temperature changes, and I can do that (in my garage) by moving the DUT into the refrigerator or freezer (both close to my workbenches), or into my warmed-up reflow oven.  Or, even by just opening the door of my heated or cooled incubator chamber.  But most of the parameters I'm interested in can be studied with extremely slow temperature changes.  In fact, those have been the most interesting ones (oscillator retrace being one, behavior of cheap TCXOs is another).

I'm doing this at home -- I'm retired and this is both a hobby and a very small business.  When I did design and development in well-funded companies we used the necessary equipment.

[derZockerMP]

I recently bought that TEC controller, and it works like a charm with the appropriate TCBView software. But when I try to connect via Python or a serial software, I can’t send commands to it. However, I still receive messages from the controller. Somebody an idea there could be the problem?
Thanks

[ch_scr]

Might be that the commands have to be terminated by CR/NL "special chars"? In my Python scripts, after opening the port

Code: [Select]

tcb = serial.Serial('/dev/ttyUSB0', 9600, timeout=1)I first deactivate the automatic reporting

Code: [Select]

tcb.write('SC\r\n'.encode())then read out what comes just in the next line

Code: [Select]

tcb.read_until('\r\n').decode('ascii')then set temp (& read response, but do nothing with it again)

Code: [Select]

tcb.write('ST 25\r\n'.encode())
tcb.read_until('\r\n').decode('ascii')in the closing code I activate automatic reporting back with

Code: [Select]

tcb.write('SS\r\n'.encode())

[derZockerMP]

Hello, thanks for sharing. There do you have the commands form? I have only the manual that is hostet from eevblog. Can you share the full scrpit to me? But only if you agree. Thanks

[derZockerMP]

Works fine now, thanks