Temperature stabilised enclosures Heating and or Cooling

[timb]

When using a peltier element for control, there is no need to do a lot of insulation. The TEC will give quite some thermal coupling from the outside to the inside. It still makes sense to have the rest of the case with a thermal resistance that is lower than the TEC internal one, but no need to go very much below that. Some insulation is good, but no need to overdo it.

As much of the heat-flow in will also be through the TEC, it really makes sense to measure the temperature on both sides of the TEC and not just the inside.

He was considering using heating pads, cylindrical elements or power resistors for heating instead of a TEC. In which case good insulation can really help improve set point stability.

I have a hot wire cutter and blue foam from building R/C wings so a custom foam box is easy. Depron in 2,3 and 6mm is worth a look for circuit board insulation as it cuts clean with a sharp knife and is a more closed cell than white styrofoam.

I have seen EPP or similar being used on some of the threads here which isn't a good insulator but it is certainly the most abuse proof and mechanically shock proof to keeps boards away from mechanical shock.

I’ve also made custom insulated enclosures by taking a larger plastic project box and filling it with that expanding spray foam stuff, then placing a smaller metal project box inside (using small shims to elevate it). Once the expanding foam dries, it can be easily cut and sanded.

[beanflying]

No problem at all hijacking the thread into a more general discussion on ideas and solutions. I partly started the thread as the site search only yielded some odd posts.

I will drill and tap out a test enclosure in the next few days for sensors and power in. Also make up a few dummy thermal blocks from foam covered sheet for mock boards.

[beanflying]

Minor change of plan for my internal needs. Still looking at methods of making a portable stable Reference/s as well.

Took delivery of this 'former' 6 bottle wine cooler today. Less than $60 delivered. Waeco branded but available around the world it seems under the part number MF-6W. It runs a removable (clip in) Laptop type power supply and there is already a 12 input plug so swapping out this one for either a linear or 12V battery based supply is really easy.

Typical Peltier cooler/heater module on the back and the smarts above the door in the front. 5-20 degrees C rated and driven by a Samsung S3C9454B 8 bit micro. https://www.insidegadgets.com/wp-content/uploads/2014/09/3F9454BZZSK94.pdf Temperature Hysterisis seems nearly 2 degrees with the stock controller and fitted sensor and driven via an on off relay like most of these types.

Even though it supports PWM and a bunch of other nice features I plan to junk the controller board rather than learning about Samsung's programming/assembler etc. The Plan is an Arduino based controller, 4 wire RTD, Peltier Cooling and non Peltier warming pads or elements. Locally it will spend 75%+ of it's time needing warming to a set point of 25 to maybe 30 degrees (haven't decided yet) so as Peltier efficiencies SUCK heating by other means makes sense but using the Peltier for cooling. Generally sticking with 12Vdc Linear control methods to minimize the noise to the DUT's.

This isn't meant to become an environmental test chamber much more of a stable reference enclosure be it Volts, Temps or Resistance. Internal Airflow is very low but internal protection from that will be done and at this stage laying it on it's side in the rack at the top (peltier hot end outside the rack), 34970's below it, Mini ITX PC Rack mount, UPS and spare batteries at the bottom.

Any thoughts are welcome

[cellularmitosis]

In my stash of bits I have a Willhi Peltier controller...

http://willhi.com/product_detail.php?product_id=115&lang=en

I picked up one of cheap ebay knock-offs of that controller and was really disappointed with the results.  Every time the relay switch on, I could see the temperature readout jump by several degrees (instantly).  I suspect the ground current of the relay is influencing the NTC reading (i.e. they didn't pay attention to their ground current paths).

Hopefully the Willhi unit behaves better!

https://www.ebay.com/itm/DC-12V-F-Fahrenheit-Temperature-Controller-Thermostat-Control-1-Relay-NTC-Sensor/141777090187?epid=595237246&hash=item210292b68b:g:lOgAAOSwQPlV-i0Y

[beanflying]

If anyone buys a second hand peltier cooler full strip and clean inside and out. Major fluff magnet!

Temporary 2k series NTC Bodge gets the stock controller to 25 while reading 20 

[cellularmitosis]

So, if you were to roll your own h-bridge to control a TEC using PWM, and you were to use an LC filter to protect the TEC, do you need two filters, as in the diagram below?

Initially I thought you could get away with just one filter (in series with the TEC), but wouldn't it only act as a low-pass filter in one direction?

[cellularmitosis]

Some additional googling indicates that yes, two LC filters are required.

http://www.ti.com/lit/an/slua202a/slua202a.pdf

[beanflying]

Still thinking about the basics with long term power consumption being fairly important. Even for cooling to drop the power consumption I could run a scavenge loop from my couple of commercial fridges and freezers in the shack using water/coolant and pump and remove the peltier from use all together. For reasons of trans portability and independence I will still likely use it as the number of sustained 25 degree weather we get is minimal.

Heating will almost certainly be as I put a few posts back with Linear controlled 12Vdc elements. Providing the noise is manageable the gear under it should help limit the heating needs. Managing the thermal drift of the 34970's is another issue 

A couple of internal temp sensors and an ambient external to help set the PID settings and minimise over and undershoot at this stage.

[beanflying]

Clearly the designer of this was deluded as to the needs of Beer Drinkers 6 pack minimum required. 30-36+C in the shack today.

Waeco holding 25 +-1 and a bit with ease 

[cdev]

Flat foam core board is extremely easy to cut neatly  to any size you want, it can also be layered, and it also is styrofoam so it insulates.

But its not fire safe. Styrofoam is very flammable.

[cellularmitosis]

My little peltier-based chamber for measuring resistor tempco's worked out fairly well: https://www.eevblog.com/forum/projects/mini-tempco-characterization-rig/

I'm currently working on a resistive (heating only) version of a similar box, which is intended solely for keeping components or a small board at exactly 25C.

It consists of 12 10R resistors affixed to the bottom of a Hammond 1590BB using Arctic Silver thermal epoxy, insulated with 1/4 inch "craft foam".

I hooked it up to a 12V supply (actually, at 95mA, to simulate 12V - 0.6V for a 2N3904, so about 1.1W) and let it run unregulated overnight, and it looks like it maxes out around 28 or 29C (in 72F ambient), so it should be able to regulate at 25C.

[cellularmitosis]

Here's a useful trick from an old thread to apply both amplification and offset to a linear temperature sensor (e.g. the MCP9701A), using just an op amp and three resistors.  Thanks again, T3sl4co1l

Here, I have 20C to 30C mapped to 0 to 5V, which interfaces well with an Arduino.

I've included the LTSpice circuit with this post, so that you can dial in the values for your temperature range of interest.  R1 and R2 control the amplification, and R3 controls the offset.

To demonstrate how to modify the circuit, I've also included a screenshot of values which map 20C to 50C to the 0 to 5V range.

[cellularmitosis]

The Arduino's 3.3V rail is a bit less noisy than the 5V rail.  Here are values for a 3.3V version.

[TiN]

snip...
It consists of 12 10R resistors affixed to the bottom of a Hammond 1590BB using Arctic Silver thermal epoxy, insulated with 1/4 inch "craft foam".
snip..

I'm sorry, but it just looks like you don't value your time and going bit wrong direction, gluing resistors not designed for heatsinking to the box. 
I'd go proper TO-220 packaged ones (example), or at least heatsinked cement ones (e.g.). Yes, yes, it's 20$, but at least you not guessing will it work or not... 

[cellularmitosis]

snip...
It consists of 12 10R resistors affixed to the bottom of a Hammond 1590BB using Arctic Silver thermal epoxy, insulated with 1/4 inch "craft foam".
snip..

I'm sorry, but it just looks like you don't value your time and going bit wrong direction, gluing resistors not designed for heatsinking to the box. 
I'd go proper TO-220 packaged ones (example), or at least heatsinked cement ones (e.g.). Yes, yes, it's 20$, but at least you not guessing will it work or not... 

Actually, I’m waiting on my polyimide heaters to arrive from China. I have three weeks to kill, so I thought I’d give this a shot in the mean time.

All I need to do is maintain my current source shunt resistors within 0.1C, which is 100x smaller than the 10C ramp I’ll use to test the tempco of resistors.  This should be adequate.

[Kleinstein]

The alternative to using resistors as a heat is to use transistors not only for control, but also as the heater.  So no more loss at an external transistor and no PMW that might produce EMI problems.

[Magnificent Bastard]

The alternative to using resistors as a heat is to use transistors not only for control, but also as the heater.  So no more loss at an external transistor and no PMW that might produce EMI problems.

Also, perhaps series high-wattage Zener diodes as well as the heat from the control transistor.  This would change the square-law power term (with only a resistor heater) into a more linear one-- resulting in better temperature control.

[CatalinaWOW]

Clearly this is an electronics forum, with many comments on circuit configuration, loop stability and the like, and only one comment on actual thermal modeling of the environment, plus Kleinstein's comment on heat distribution.  Admittedly thermal modelling software is every bit as complicated as EDA software, and generally just as pricey so most of us have little access to it, but that isn't an excuse for ignoring key parts of the problem.

There needs to be attention paid to the fundamentals of the problem.  Which elements (if any) need to be isothermal?  Which need to be stable over time, but not necessarily isothermal?  What are heat paths to and from those elements?  What are the heat capacities and thermal conductivity of those pieces?  What are the overall control loops, including those elements?  Note that making heater power much greater than heat flow out makes control inherently non-linear and complicates stabilization.  This is one advantage of TEC as you can get better balance to these flows.

For a reference oven, where there is much less time pressure to reach the operating temperature, no variation in the thermal characteristics of the controlled item, and little or no need to vary that temperature the inherent reliability of a mechanical solution (insulation, nested boxes etc) combined with a really simple outer control loop will be the best choice.  Very different case than an environmental test oven.

[Kleinstein]

The alternative to using resistors as a heat is to use transistors not only for control, but also as the heater.  So no more loss at an external transistor and no PMW that might produce EMI problems.

Also, perhaps series high-wattage Zener diodes as well as the heat from the control transistor.  This would change the square-law power term (with only a resistor heater) into a more linear one-- resulting in better temperature control.

One can still get a linear control with a resistor, if PWM is used. However with voltage or current control the resistor is square law, while a transistor at a constant voltage and with current control (e.g. with a small emitter resistor) is about linear, which makes control easier.
For power distribution one could still use transistors in parallel or just normal diodes in series.  It is also possible to use a 2 nd transistor in series (with a base voltage at about 50% or the supply) - which could also act as an emergency off.

A reference heater only needs to be constant and might thus be a rather simple linear analog  controller. For a test chamber one might want faster control and a programmed ramp - thus more like a digital control. This could compensate for a nonlinear heater (a peltier element is also nonlinear, when used at relatively high currents).

[Andreas]

I'm currently working on a resistive (heating only) version of a similar box, which is intended solely for keeping components or a small board at exactly 25C.

Hello,

I don´t know your room temperature in summer.
But I have changed the reference temperature to 30 deg C to have some headroom even with somewhat higher room temperatures.


with best regards

Andreas

[cellularmitosis]

Hello,

I don´t know your room temperature in summer.
But I have changed the reference temperature to 30 deg C to have some headroom even with somewhat higher room temperatures.


with best regards

Andreas

I am a spoiled American.  My room never gets above 72F (22.2C).  I am curious though how thin of a margin I can pull off and still maintain regulation!

[bertik]

Have a look at my Arduino project of a while ago: http://hololaser.kwaoo.me/electronics/Arduino_TEC/Arduino-TEC.html

[cellularmitosis]

Have a look at my Arduino project of a while ago: http://hololaser.kwaoo.me/electronics/Arduino_TEC/Arduino-TEC.html

wow, stable within one millikelvin!

[Pipelie]

Have a look at this one, they have very nice software run on windows.

https://item.taobao.com/item.htm?spm=a230r.1.14.18.6985ef74GxyII7&id=535913852209&ns=1&abbucket=17#detail

[TiN]

I have no problems for 5 mK stability over long time with my crude setups and fancy Keithley 2510. Thermal sensors are either Honeywell HEL-705 RTD or YSI 44006.
Setup with ILX 5910B and YSI 44007 shows simialr performance too, with much smaller cost then 2510.