Announcement

Collapse
No announcement yet.

OT? How to Size a Heating Element for a Small Electronic Enclosure

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • OT? How to Size a Heating Element for a Small Electronic Enclosure

    I am building a Voltage reference with a circuit board that uses the AD584 chip. It is pretty accurate just at room temperature, but I am toying with the idea of adding a heater and thermostat to bring it to a constant temperature for even better stability. We are talking about a box that will be around 2" x 2" x 1.5" or so, probably made of plastic. If I do the temperature control I will probably add some insulation to minimize the heat loss. Since the temperature in my shop and house is around 68 to 70 degrees F year round, I would shoot for around 75 degrees F or just a bit higher.

    But I have not had any experience with controlling the temperature in such a device and am wondering if anyone has a way to at least ball park the amount of heat (Watts) required for such a task. From a good starting point I can just experiment with the values and components until it works like I want. So far I have an unregulated 18 VDC and a regulated 12 VDC to work with but other Voltages could be designed in.
    Paul A.
    SE Texas

    And if you look REAL close at an analog signal,
    You will find that it has discrete steps.

  • #2
    im ball parking less than 1w.

    Comment


    • #3
      Paul,
      I have never done something that small, but I have seen it done very simply. In a 70's vintage Motorola two way commercial radio, Moto simply put a carbon resistor next to the crystal to maintain frequency stability to FCC regulations. I do not remember the resistance, and do not have my manuals any more, but the resistor was a 1/2 watt carbon and it was connected to the incoming 12 volt dc supply. The resistor never got "HOT" just slightly warm. Do the math and give it a try. The absolute heat is not the issue, just stability.
      I think I would make a good estimate and then put the resistor in contact with and over the chip with heatsink compound to make a firm thermal connection. Your idea about insulation will also help. Good luck!!
      Robin

      Happily working on my second million Gave up on the first

      Comment


      • #4
        Even a half watt resistor and temperature sensor sitting in close contact on the chip and then given a block of foam or wad of cotton or fiberglass insulation over top would do the job. Include a blob of thermal grease to fill in any air gaps. Send the output of the sensor to a chip that drives heating power through the resistor and it'll all work out just ducky.

        If you are using it at room temp though will the usual seasonal temp differences matter?



        Chilliwack BC, Canada

        Comment


        • #5
          Watts = U factor x area x temp difference. A very small number indeed. Electronics inside the box may overheat without added heat source.

          Comment


          • #6
            Originally posted by BCRider View Post
            Even a half watt resistor and temperature sensor sitting in close contact on the chip and then given a block of foam or wad of cotton or fiberglass insulation over top would do the job. Include a blob of thermal grease to fill in any air gaps. Send the output of the sensor to a chip that drives heating power through the resistor and it'll all work out just ducky.

            If you are using it at room temp though will the usual seasonal temp differences matter?


            What level of accuracy do you really need? If keeping the reference chip at constant temperature is actually needed, then you'll need a variable output for the heating element. In which case, BCRider's solution gets my vote. For a few pennies, you can populate a 1 or 2 watt resistor next to the chip along with an RTD. Use the output from the RTD to drive the resistor through a power amplifier. You can either do this using entirely analog components or you can make a digital control loop, which is theoretically easier to tweak PID coefficients.

            I do this kind of thing at work on a regular basis. We have extremely high gain amplifiers (PMTs with 10^8 gain) that are temperature sensitive. To maximize accuracy of our detectors, we keep the internal temperature constant within 0.1*C over an ambient temperature range of -30C to 55C. For a large enclosure (>1000 square inches), I need about 12 watts to keep the temperature 10C over room temperature, and this is with about 1" of insulation on all surfaces, but I need 120 watts to warm up from -30C in a reasonable amount of time... So it's not just how much you need to keep steady state, but how much you need to reach steady state without waiting for "ever".

            How to calculate? Well deltap has the general formula. The hard part is figuring out the "U factor". You have to factor in the thermal resistance of each material, the thermal resistance associated with each interface, and then make some guesses / assumptions regarding heat transfer to the surrounding environment. This can change dramatically depending on whether or not there is solar loading, air speed across the surface (is your shop / work bench drafty?), etc.

            I usually run a back-of-the-envelope calculation and then simulate everything using some $$$ modeling software. But for a one-off box like this, I'm betting a 1-2 watt resistor in close proximity to the chip is more than sufficient. You can then drive it with a power amplifier (if you have an analog feedback) or PWM (if you have digital control) to maintain a constant temperature at the reference chip. Your control electronics and heater driver circuit need to be thermally isolated from the reference chip and heater, of course.

            Comment


            • #7
              My estimate would be about 1 watt. Try a 330 ohm 2 watt resistor connected to your 18 VDC and put it in your box with a temperature probe. See what maximum temperature is reached, above ambient. Figure what your minimum and maximum ambient will be, and ascertain that the resistor will heat the IC to a temperature above maximum ambient, and use a simple thermostat circuit to hold that temperature. The AD584K is rated 15 PPM/oC and the J version is 30 PPM/oC. If your ambient is always between 68 and 70 F you really should not need any temperature compensation, although other circuitry in your enclosure may raise the temperature above ambient by an unknown amount.

              Here are some ideas for a crystal oven heater circuit, which is basically what you are asking for:

              http://www.masterscommunications.com...s/ch/ch25.html

              https://ka7oei.blogspot.com/2018/01/...that-uses.html
              http://pauleschoen.com/pix/PM08_P76_P54.png
              Paul , P S Technology, Inc. and MrTibbs
              USA Maryland 21030

              Comment


              • #8
                So this made me think of the crystal ovens that were not uncommon in old raido gear. I suspect I may have a few still lurking around here someplace. But in searching for an example, here is a guy who has done something already that is almost exactly what you are trying to do. Might be a good jumping off point for your project:

                http://romanblack.com/xoven.htm



                late edit: another version here, about halfway down the page:



                https://www.qsl.net/d/dl4yhf//dcf77_osc/
                Last edited by alanganes; 05-03-2021, 07:24 PM.

                Comment


                • #9
                  Some interesting suggestions and ideas here. That RomanBlack circuit seems dead simple and probably works well enough for my purpose.

                  The general consensus here seems to tend towards controlling just the temperature of the device itself (my circuit's AD584 chip) as opposed to the entire, small box it is enclosed in as I was thinking. It is strange that I was thinking that way as I have worked with OCXOs before. Or perhaps not as those OCXOs were always in a metal can that was about the same volume that I imagine for this circuit.

                  It seems that around 1 Watt or less may be the power level needed for this. That gives me a starting point. I think my first stab at it may be to re-package the RomanBlack circuit around the AD584 chip.

                  Thanks to all for the info. and suggestions.

                  Now back to my income tax.
                  Paul A.
                  SE Texas

                  And if you look REAL close at an analog signal,
                  You will find that it has discrete steps.

                  Comment


                  • #10
                    For another dirt simple idea you could use a PTC Thermistor as a small heating element for the chip. The advantage is that the device is inherently self regulating as the resistance goes up as the heat goes up. You would need to check out what is available and experiment a bit.
                    Robin

                    Happily working on my second million Gave up on the first

                    Comment


                    • #11
                      Someone over on eevblog has done the same thing" https://www.eevblog.com/forum/projec...oltage-source/

                      Comment


                      • #12
                        I don't really like the RomanBlack circuit. It is certainly dead simple, but perhaps too much so. You could use an LM324 quad op-amp, which can drive 40 mA per channel (although supply is limited to 50 mA). So it could provide 900 mW from an 18V supply, and all you would need to add is a thermistor or an LM134/334 temperature sense IC.

                        Being a Microchip advocate, I might also suggest something like a PIC12F1501, which has an internal temperature sense module as well as other peripherals. It would be possible to use PWM to drive a power MOSFET directly as a heating element, or to drive a resistor for that purpose. If you wanted to get fancy, you could use output ports to drive LEDs showing the temperature - perhaps red and green for above and below setpoint, and both on for temperature within, say, 1o F of desired temperature. And the DAC could be used for an analog output for temperature. Of course those would be unnecessary "bells and whistles", but might be fun as a project. At least for me it would be...
                        http://pauleschoen.com/pix/PM08_P76_P54.png
                        Paul , P S Technology, Inc. and MrTibbs
                        USA Maryland 21030

                        Comment


                        • #13
                          Originally posted by PStechPaul View Post
                          It would be possible to use PWM to drive a power MOSFET directly as a heating element, or to drive a resistor for that purpose. )
                          That was my thought as well but it occurred to me that whatever small gains in precision / accuracy are gained by keeping the reference at constant temperature could be offset by the introduction of noise from the PWM signal. That's not to say it can't be done, but careful layout and maybe some filtering would seem to be in order, even at relatively small currents associated with a 1 watt load.

                          Comment


                          • #14
                            Using a PIC would allow displaying an indication that the device has stabilized at the set temperature and also displaying that temperature. And probably tons of code already available. I am going to search.



                            Originally posted by PStechPaul View Post
                            I don't really like the RomanBlack circuit. It is certainly dead simple, but perhaps too much so. You could use an LM324 quad op-amp, which can drive 40 mA per channel (although supply is limited to 50 mA). So it could provide 900 mW from an 18V supply, and all you would need to add is a thermistor or an LM134/334 temperature sense IC.

                            Being a Microchip advocate, I might also suggest something like a PIC12F1501, which has an internal temperature sense module as well as other peripherals. It would be possible to use PWM to drive a power MOSFET directly as a heating element, or to drive a resistor for that purpose. If you wanted to get fancy, you could use output ports to drive LEDs showing the temperature - perhaps red and green for above and below setpoint, and both on for temperature within, say, 1o F of desired temperature. And the DAC could be used for an analog output for temperature. Of course those would be unnecessary "bells and whistles", but might be fun as a project. At least for me it would be...
                            Paul A.
                            SE Texas

                            And if you look REAL close at an analog signal,
                            You will find that it has discrete steps.

                            Comment


                            • #15
                              Ach..... "digital", PWM, etc....... Plain old Analog is perfect for this. Use the temp sensor, and a reference voltage, an opamp, and a booster transistor, and you can have the whole thing with no frigging digital noise at all!
                              2801 3147 6749 8779 4900 4900 4900

                              Keep eye on ball.
                              Hashim Khan


                              It's just a box of rain, I don't know who put it there.

                              Comment

                              Working...
                              X