Welcome to the E2E Community Discussion Forum!

I need to maintain a temperature within .2 degrees C in a 10x12 in. enclosure for a biometric scanning prototype. Any recommendations?

---

"We are the dream makers and the dreamers of dreams..."

I think the only way to even approach this type of uniformity is to have an engineering analysis done using thermal imaging and a standard sized catalog heater.  Another more theoretical analysis could be done using finite element analysis.  Once it is determined which areas need more power due to edge losses or other heatsink features, a custom "profiled" heating element could be designed.  A profiled heater can have many zones some of which would have a higher watt density than others in order to account for losses and maintain a very high degree of temperature uniformity across the surface.  Almost any type of heater can be profiled, so there is some choice available as far as the best type for your application.  Silicone rubber or polyester insulation heaters are usually most cost effective in that size range. 

---

"A witty saying proves nothing."
-Voltaire

Another approach you can take to get the uniformity is to create a heater with multiple independent elements.  By placing a sensor or thermocouple on your heating surface in the area that each one of your elements are and then controlling each element separately based on the sensor feedback, you can adjust each area until you get the uniformity you need.  The downfall with this approach is you need multiple controllers.  Sometimes this works very good for prototyping though if don't want to do finite element analysis.

---

'Experience is a hard teacher because she gives the test first, the lesson afterward.' --Vernon Law

You don't state your desired operating temperature but since your application is biometric I assume that it is not typically much over ambient. This is in your favor for maintaining .2 degrees C. However, all of this advantage can be wasted by an improperly designed device. You must first design to maximize the transfer of heat from the element to the chamber or surface being heated. Use of high thermal conductivity material such as aluminum in the device or as a heat spreader for the element will assist in achieving this. Secondly, you must isolate the  device form external thermal influence through insulation to eliminate heat loss, shielding from air currents, etc. The more isolation the better.

With consideration of thermal stability as part of the original design you will minimize the time to achieving the .2 deg C uniformity. However, you will stll need to rely on profiling and/or multiple elements and sensors plus some trial and error iterations to get there.

Based on jthompson's original post, it seems that the goal is to maintain the air temperature in the enclosure within .2 C of the setpoint.  There does not appear to be any need for a "profiled" or zone heater for temperature uniformity, if this is actually the case.  The requirement appears to be uniform and accurate air temperature control in the enclosure.  To do this, some source of air circulation, like a fan, will be needed to make the air temperature uniform.  Depending on the environment that this device will be used in, a thermoelectric module may be a better choice than just a heater.  The thermoelectric module would allow heating and cooling.  Obviously, the main heart of this system, using a heater or a thermoelectric module, will be some sort of controller to maintain this highly accurate temperature.  A controller with PI, PID, or Fuzzy Logic capabilities will most likely be required.

Also a possible solution but jthompson's message listed 2 dimensions for the "enclosure" which I see as leading to the premise of a surface temperature requirement. Assuming an air temperature requirement is still an assumption but it may be the right one.  In light of this discussion, the OP should fill in some of the blanks like air vs. surface temp, target temp, ambient environment conditions, etc..

A recirculating fan may keep the average temperature in the .2C range but I would expect a hot spot near the heat source (the existence of which is also an assumption).

---

Not everything that counts can be counted, and not everything that can be counted counts. - Albert Einstein (trad.)

Star_E

Sorry for not providing the detailed info.  My aim is to keep the boad level electronics and scanning device withing a .2°C range for 28°C, which is where the scanning device has optimal operation.  The whole enclosure is actually 10" x 12" x 5" deep, but the heatsink is a 10" x 12" Al plate.  The elctronics and scanning device are mounted on the heatsink, so I need to transfer through the heatsink and maintain the temp on the electronics.  The enclose will be sealed, so wind and other elements will not affect the heat transfer.  The ambient conditions will never be below 16 or 17°C and not above 38°C.

 

My thinking is to use some standard heaters I have on hand and test to see how close I can maintain a temp across the electronics and scan device.  I will probably hook a couple variacs up to try and profile the heat transfer, and then look to get a profiled prototype.  Is there any benefit to doing FEA?  Budget is an issue.

---

"We are the dream makers and the dreamers of dreams..."

Maintaining .2C is tough. Consider the job the heater is doing. It's replacing heat loss. If your heat loss is low, you'll never get control that tight. The typical way to maintain .1C control is to run a chilling unit, and use a heater with PID contol to maintain temp. Yes, you waste power with a heater fighting a chiller, but you get tight control.

My idea is to bore a water channel through the heat sink, and circulate silicon oil through it from a seperate chiller. The chiller is a Peltier based unit. The Peltier switches between heat and cool. These chillers are comercially available for those that raise salt water fish. About $200.00 to $300.00. If you can't bore, then add bored blocks to the sides of the heatsink. Aluminum is an excellent conductor of heat. Use heat transfer compound between the blocks and heatsink.

Your idea with the variac sounds good. With a controller, you'll probably have problems if the control sensor is too far away from the heater. If so, mount the sensor closer to the element, or directly on it. Set the temperature of the heater to give the desired temperature at the electronics. If your heat loss (load) is constant, this method should work. Do the electronics generate heat when active?
Adding fans to increase heat loss may help, but changes in ambient temperature will affect cooling and control.

Keith

Consider using one or two thermoelectrics as the heat regulating device. As a previous respondee indicated, having a balance between heating and cooling is probably the best way to achieve precision temperature control. This heating/cooling balance is how precision temperature bath manufacturers maintain temperature. By sandwiching the thermoelectrics with a large heat sink, creating a large thermal mass, temperature swings can be avoided.

You will obviously need some kind of cooling if the ambient can get above your control temp!  We maintain temperatures at these tolerances using TEC modules as both heaters and coolers.  Maxim makes an IC, MAX1978ETM, I think, that will drive the TEC in both heating and cooling modes and they claim control to within less than 0.1 C.  The actual load will determine how big the TEC must be.  You can also just run the TEC cold and buck the cold with a Minco type heater using PID control to get to the temperature you want.  Sandwich the TEC between your heatsink and an appropriate fan to get rid of the excess heat (or cold) generated by the TEC. 

Your ambient range requires that you have both heating and cooling.  As suggested earlier, a peltier device or water flow through the heatsink can be used.
Maintaining a constant temperature at the heatsink is a problem if there is a time lag betweent the heater/cooler and the heatsink.  I did this for a device many years ago and found that by having the thermal sensor and heating element very close to each other on the heatsink, I was able to maintain very tight control.  If there is a time lag between the heating element and the temp sensor, there will be overshoot and undershoot of the temp, and maybe  even oscillation if the time lag causes a 180 phase shift.

In your case, I would suggest a peltier device mounted directly on your heatsink and let the control sensor be mounted as closely to the peltier contact point as possible, even under it. 
Alternately you could start with external water at a controlled temp of about 25C and put a heater element into your heatsink next to the sensor, but that requires some experimental temperature balancing across the heatsink and may defeat your budget.  You can do some thermal design with a CAD program, but having an accurate model may be more time consuming than experimental evaluation.
If your box is fairly well insulated, a small circulating fan may be advantageous to keeping internal gradients to a minimum.

Note that  you will have a gradient along the length of the heatsink to the point fartherest from the controlled area, but if the heatsink is heavy, that will be minimized.  You want most of your heat transfer to be through the peltier device or water cooled point, not the entire surface of the heatsink.  That way, you have control of the thermal gradient.
Finally, Good Luck!

Great answer! And the only answer that will work.
A PID controlled TEC with IC.
A TC or RTD are not nearly accurate enough to control within 0.1 deg C.
A limited budget this will not comply.