Thermal enclosure for measurement stuff

[Echo88]

Im considering the use of a 19" network cabinet (+ PID-temperature-controlling of the interior and added insulation) for my voltnutting measurement setup: E34465A, reference box, ADC/DAC-box and son on, since my room-temperature isnt really stable and me fingering around the setup isnt really helpful for temperature-stabilizing of the components. Also all of the measurement can be done via GPIB from the outside. Does anybody here use the same approach and can give helpful hints or is this idea not worth following?

Maybe theres also a better solution, for example a small glassdoor-refrigerator with an already excellent insulation were the interior is heated to, say, 25°C.   

[EmmanuelFaure]

Maybe theres also a better solution, for example a small glassdoor-refrigerator with an already excellent insulation were the interior is heated to, say, 25°C.   

Absolutely.

Click :

[joeqsmith]

I just use a meat shipping box, with another box inside with some more foam.   Resistor, fan and RTD controlled by the PC.  I have never had a good way to cool.   

[Kleinstein]

Unless you have a high power device inside there is limited value in having active cooling. When going below room temperature the relative humidity will go up - so unless you have a very dry environment this can cause trouble with condensation. So you would also need humidity control (some kind of absorber at least). Already going above some 60% RH can cause surface leakage to increase. So the slightly higher and thus dryer conditions inside the instruments also have an positive side.

For temperatures slightly a above room temperature it's more like a heater and a temperature controlled fan at the outside to bypass insulation.

The wine cooler still could be good as it has the window and insulation.

[Echo88]

I recently aquired a Acopino BC-48 wine cooler, at 21 °C room temp it reaches 11°C after about 2h and stays at 11°C when no heat load is in the cooler. The attached curve shows the 34465A as a heat load inside the box, while the cooling function is deactivated -> temp stabilizes after a few hours at 31°C. After that i activated the cooling and the temp got down to 16,5°C with the 34465A acting as a heat load.

I will add a PID-controller and maybe a beefier peltier module, so that the box temperature will always stay at 25°C (room temp is seldom over 21°C, so the peltier acts most of the time as a heater) even in summer, when the indoor temp can reach 30°C.

[zlymex]

I recently aquired a Acopino BC-48 wine cooler, at 21 °C room temp it reaches 11°C after about 2h and stays at 11°C when no heat load is in the cooler. The attached curve shows the 34465A as a heat load inside the box, while the cooling function is deactivated -> temp stabilizes after a few hours at 31°C. After that i activated the cooling and the temp got down to 16,5°C with the 34465A acting as a heat load.

I will add a PID-controller and maybe a beefier peltier module, so that the box temperature will always stay at 25°C (room temp is seldom over 21°C, so the peltier acts most of the time as a heater) even in summer, when the indoor temp can reach 30°C.

Very impressive. How do you plan to handle the cross over from heat to cool for the peltier?
I also bought some peltier units and haven't start the thermal enclosure yet. I may use bridge to reverse the polarity or use dual supply.

[Andreas]

Hello,

it´s easier to put a heater foil on the bottom of the Cooler Cabinet.

http://www.reichelt.de/Heizfolien/THF-100200/3/index.html?ACTION=3&GROUPID=7777&ARTICLE=108464&OFFSET=16&
I am doing 50 Hz PWM modulation since most instruments filter this out with some NPLC.

Peltiers do not like a reversal from hot to cold without a 5 minute dead time.

Has the BC-48 any possibility for a temperature setpoint from a PC?

With best regards

Andreas

[Echo88]

The BC-48 only has a Min/Max-Switch. Thats why i will use a generic PID-controller. Since the 34465A will stay in the cooler, its now just necessary to cool the box down to 25°C, instead of needing a seamless cooling/heating-peltier-controller.

[Andreas]

This could be a real problem with a variety of products that use these Peltier devices, and may explain some reliability problems some of these products have.  I would like to learn about this phenomenon.  Could you please post some links to information about this?

Thanks,
Ken

Hello Ken,

I found this hint in one of the manuals (unfortunately not in all) of peltier cooler boxes which can be switched from cold to warm.
And I have also observed that the cooling power is only around halved if you dont start from temperature equilibrum of both sides.

One reason is that the Peltier also acts as generator if you have a cold and a hot side. (Seebeck effect).
You can see the indicator LEDs glowing for some time if you switch off the power.
If you reverse the power in this state there will be some excessive current flow and thus some kind of thermal runaway.

The BC-48 only has a Min/Max-Switch. Thats why i will use a generic PID-controller. Since the 34465A will stay in the cooler, its now just necessary to cool the box down to 25°C, instead of needing a seamless cooling/heating-peltier-controller.

I hope you planned a DC-source for the PID controller.
Efficiency goes down with any ripple current.

With best regards

Andreas

[pelule]

I am using an old fridge (get it used for free). "Heating" is a 60W bulp. A fan is genrerating const air circulation.
I first had a look to Daves peltier fridge video, but refused to use it, because of the limited cooling power and slow change of the temperature.
The fridge would able to go below 0°C easily, but one need to be careful with condensation.
I am using a 3-point regulator (hot-of-warm, dT at 0.1° C) as the fridge isolation is very good.

[chuckb]

Here's an idea that I will someday get around to building and testing.

To stabilize the electronics temperature in modern DVMs just regulate the air temperature coming into the DVM via the fan. If your room normally has a 23 deg C maximum temperature then build a small box to heat the fan air to 24 +-0.1 deg C before it enters the DVM. The electronics inside the DVM will only experience 24 deg C air. Your DVM will suddenly think it has move to a cal lab!

There is very little heat flow of critical parts (Vref, resistor, ADC) via conduction to the enclosure in modern high end equipment. These components only interact with fan air. This is one reason the readings shift when the fan air velocity changes as the filter gets restricted.
 
This would also help stabilize DVM operation when it's installed in large racks of other equipment. Alternately you could just run a hose to duct cool air from outside the rack to the DVM fan.

Hope this sparks a useful idea.

[zlymex]

Hello,

it´s easier to put a heater foil on the bottom of the Cooler Cabinet.

http://www.reichelt.de/Heizfolien/THF-100200/3/index.html?ACTION=3&GROUPID=7777&ARTICLE=108464&OFFSET=16&
I am doing 50 Hz PWM modulation since most instruments filter this out with some NPLC.

Peltiers do not like a reversal from hot to cold without a 5 minute dead time.

Has the BC-48 any possibility for a temperature setpoint from a PC?

With best regards

Andreas

Hello,
I didn't mean to revere the Peltiers in short period of time, especially not in PWM rate.
What I need may be two fold. One is to set the temperature of the enclosure at 23 degree C, but the room temperature may be lower or higher than that, thus requires either a heat or cool, without my intervene of the hardware.
The other requirement is the gradual temperature ramp, say from 15 degree C to 30 degree C, which will cross the room temperature for most of the case, and the auto switch over from cooling to heating will happen at some point.
In either case, I can use PWM method to drive the Peltier device, only either from negative voltage to zero(in the case of cooling) or from positive voltage to zero(in the case of heating). There will floow "reversal from hot to cold without a 5 minute dead time".

[guenthert]

Here's an idea that I will someday get around to building and testing.

To stabilize the electronics temperature in modern DVMs just regulate the air temperature coming into the DVM via the fan. If your room normally has a 23 deg C maximum temperature then build a small box to heat the fan air to 24 +-0.1 deg C before it enters the DVM. The electronics inside the DVM will only experience 24 deg C air. Your DVM will suddenly think it has move to a cal lab!

There is very little heat flow of critical parts (Vref, resistor, ADC) via conduction to the enclosure in modern high end equipment. These components only interact with fan air. This is one reason the readings shift when the fan air velocity changes as the filter gets restricted.
 
This would also help stabilize DVM operation when it's installed in large racks of other equipment. Alternately you could just run a hose to duct cool air from outside the rack to the DVM fan.

Hope this sparks a useful idea.

While there is a temperature coefficient in the measurements by a meter, it often is quite small, e.g. for the HP34401 it's 6ppm/K in the 10VDC range *outside* the comfort zone of 18 to 28 degree C (inside, the TC is not stated, presumably temperature related error is then part of the range error).  Well, that's the stated error -- with any luck, your unit behaves even better.  The TC of the device under test is generally much larger.

The fellows who put their DVM in a temperature chamber did so only to use them as a heat-load counteracting the cooling element.

[montemcguire]

{...snip...}

Peltiers do not like a reversal from hot to cold without a 5 minute dead time.

{...snip...}

This could be a real problem with a variety of products that use these Peltier devices, and may explain some reliability problems some of these products have.  I would like to learn about this phenomenon.  Could you please post some links to information about this?

Thanks,
Ken

I looked into this and found this from Ferro Tec's website.

https://thermal.ferrotec.com/technology/thermoelectric-reference-guide/thermalref10/

While there are some annoying things you can do to a TEC, they can actually be quite reliable. Large temperature swings seem to be a problem, and from what Ferro Tec states, the problems with these large swings are mechanical stress related. So, I'm not sure that there's any "5 minute rule" for all TECs, but perhaps specific limitations for certain mechanical and mounting designs, as well as specific operating temperatures and temperature changes.

TECs get used to stabilize semiconductor lasers for telecommunication quite regularly, under closed loop control, and seem to be pretty reliable.

[Cerebus]

While there are some annoying things you can do to a TEC, they can actually be quite reliable. Large temperature swings seem to be a problem, and from what Ferro Tec states, the problems with these large swings are mechanical stress related. So, I'm not sure that there's any "5 minute rule" for all TECs, but perhaps specific limitations for certain mechanical and mounting designs, as well as specific operating temperatures and temperature changes.

I'd have thought that if there is a limit, it would be more in the nature of a rate limited change of temperature across the TEC - so 'x' degrees C per second where x might be quite small. If that is the case, and one is controlling the temperature with something that has a proportional element to the control, then switching from heating to cooling and vice versa is likely to only produce a fairly small slope around the set point.

[montemcguire]

Exactly - which is why a manual for a heater-cooler appliance which only has a hot or cold switch would add the 5 minute warning.

Still, TECs are mechanically a little delicate - the Bismuth Telluride pellets are soldered to a ceramic substrate with some probably brittle, low melting point solder, and while they seem to have reasonably high compression strength, their shear strength is low. Shear stress would result from heat and expansion coefficient differences between the pellets and the substrate they're soldered to.

From what I can find, the thermal coefficient of expansion of the bismuth telluride pellets is directional, ranging from 13 to 22 ppm per degree C, whereas alumina, the typical substrate, is 7.2 ppm per degree. Not a huge difference, but with large temperature swings, large stresses could be created. Ferro Tec also said that modules with a large number of pellets will be less reliable, since these expansion coefficient problems are magnified with a large array of pellets. An aluminum nitride substrate has an even lower thermal coefficient of expansion, and would be worse.

So, a handful of TEC modules, each using a relatively small number of elements, soldered to small Alumina substrates, mounted in an arrangement that protects them from shear forces, and driven by a fairly slowly changing DC voltage, will probably be pretty reliable. This seems to accurately describe the small TECs that are used to drive laser diodes for optical communications, and those seem to be pretty reliable.