Designing for very hot environments

I’m contemplating a P1 design that will be operating basically forever at 80 mhz in a very hot environment with all 8 cores well utilized. The question is do I go with the SMD version or the DIP? There are no tight constraints on board dimensions so either package would work. I can also attach a heatsink or use the metal case, but the enclosure will be tightly sealed so no fans. Which package would work best for heat rejection? I’m leaning toward DIP but I’d like to hear from someone who has maybe already walked this road.
«1

Comments

  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 22,442
    edited 2019-09-12 - 18:07:56
    Consider this: dimensionally, a DIP has a thicker blanket of plastic around the chip than does an SMD (about twice as thick, actually), and plastic isn't a great conductor of heat.

    Here's a document that might be helpful:

    https://www.ti.com/lit/an/spra953c/spra953c.pdf

    -Phil
    “Perfection is achieved not when there is nothing more to add, but when there is nothing left to take away. -Antoine de Saint-Exupery
  • That TI whitepaper was a bit of an eye-opener. I appreciate you posting the link.

    Based on that read it looks like I’ll be using the SMD version and mount the board upside down with a heat spreading copper plane below (above, really). That would seem to make the best of a marginal thermal environment.
  • To answer your question on "a P1 design that will be operating basically forever at 80 mhz in a very hot environment with all 8 cores well utilized"

    What exactly is the "very hot temperature" that are you anticipating? What is the size of the PCB? How large is the internal size of the enclosure? Are both the internal temperature and external enclosure temperatures going to be the same? ie. A static condition and/or a heat soak condition?

    BTW - The TI report is very good and a great starting point. Follow it closely!!!
  • @PropGuy2 The environment is in an industrial facility. The enclosure is an explosion proof doghouse that is bolted directly to a 40 hp 3ph/480 volt motor. My gizmo has to live in the doghouse. It gets to 100c pretty routinely. Part of what is driving the high temperature is the nitrogen purge gas temp. The facility produces nitrogen as a byproduct of another process at about 600 degrees. After cooling it, they use it internally as an inert purge for basically everything. They produce volcanic quantities of the stuff and most of it just gets vented as waste.

    The problem comes when there is a high demand for nitrogen somewhere downstream. This causes high flow rates thru the heat exchangers causing the purge nitrogen to get quite hot. And wet. The wetness gets mostly handled by an inertial separator, but the heat from the motor and the purge gas combines to make it challenging. The outside of the doghouse runs about 60c, but my gizmo has to be completely contained in its own enclosure within the doghouse, so there wont be much of a temperature gradient internally.
  • Ow! Those are truly brutal operating conditions! Since the Prop is only rated to 25C over your presumed ambient, I'd be tempted to put a Peltier cell between the chip and the box your system is enclosed in. Also, make sure that all the other components on your board are rated for the temps they'll be experiencing.

    -Phil
    “Perfection is achieved not when there is nothing more to add, but when there is nothing left to take away. -Antoine de Saint-Exupery
  • @Phil Pilgrim (PhiPi) Did I mention the motor is driving a hydrogen compressor? Just a “little” one to run a bit of H2 to a couple of catalytic flares they use to burn-off emissions from other processes.

    Its like a chain. Every process depends on other processes to make it run. If the nitrogen stops, the purge is lost to the entire site. Not good. If the hydrogen stops, the catalytic flares stop which means those processes that need venting have to stop. So their on-site nitrogen farm keeps a few zillion cubic feet of nitrogen in cryo storage as a back-up so they can do an orderly shutdown. They also keep reserve natural gas (as LNG) in storage as a temporary hydrogen replacement for the same reason. Its all just amazing to me. It is totally outside of my usual comfort zone.
  • Crikey!!

    That is indeed brutal!!
    Windows 10 Home:
    Propeller Tool 1.3.2, BASIC Stamp Editor 2.5.4, Javelin Stamp IDE x.x.x, Arduino IDE, PST
    Hardware:
    BASIC Stamp Board of Education, Propeller Board of Education, Propeller Project Board, Propeller Mini

    Me, a programmer in English class, asking for grammar help: “What is the syntax for this statement?”
  • JRoark,

    What are you doing that you need 80 MHz?

    Normally parts get de-rated, or run at lower rates, for higher temperatures; that is unless you want to be at the mercy of "Uncle Murphy". :P
  • Surface mount to save space and heatsink right in top of it maybe with a tiny fan on it ala R Pi. Like Phil's peltier device, and maybe enclose the cold side and the electronics inside an aerogel insulated box if space permits. If all cores are "well utilized", maybe a little 8 pin mpu for controlling the peltier device.
    Ordnung ist das halbe Leben
    I gave up on that half long ago.........
  • I would structure the software so that it didn't load the P1 too much anyway. But what about the regulators themselves? They will be getting quite warm and at 100'C ambient they will be going into thermal shutdown. Another good reason to only lightly load them through more efficient software.

    Tachyon Forth - compact, fast, forthwright and interactive
    useforthlogo-s.png
    --->CLICK THE LOGO for more links<---
    P2 +++++ TAQOZ INTRO & LINKS +++++ P2 SHORTFORM DATASHEET
    P1 +++++ Latest Tachyon includes EASYFILE +++++ Tachyon Forth News Blog
    Brisbane, Australia
  • Would the QFN have the best thermal conductivity thanks to its thermal pad?

    P1 can function at 190C. https://youtube.com/watch?v=EjkXokgcBZw

    James https://github.com/SaucySoliton/

    Invention is the Science of Laziness
  • Some of what is happening in the box is the subject of an NDA, but stuff that isnt includes several temperatures, several gauge pressures, baro pressure, tachometer, a hydrogen sensor (ya think? Lol) AC voltage and current on three phases, two different DC voltages, one pulse width monitor, while at the same time managing a fiber comm link for galvanic isolation and explosion prevention reasons. Coupled with the rest of it, it will be a fairly busy little box.
  • I have done one or two projects like this. It is usually more than just the microprocessor! I am thinking every single component - Resistor, capacitor, logic ICs, voltage regulator and certainly any and all sensors, need to be carefully spec out. Then again an explosion proof Doghouse and being next to a 3-Phase / 480 volt (compressor ?) motor - is an engineering challenge in itself. Add to that exceptional life cycle / possible component failure / redundant controls - and who knows what else? My gut feeling is that this is something that requires a whole team of engineering specialists. Has anyone previously / actually built this controller? Maybe you can reverse engineer it and save yourself a lot of pain.
  • @PropGuy2 I'm actually being brought in as a bit of an Instigator by the senior engineer (a neighbor). He mentioned the problem to me "over the fence" one day, and one thing led to another...

    There is no penalty for failure. This is an experiment. My stuff is totally passive and has no control of the machine. Control gets handled by the vetted, megabuck process controllers. I just get to watch for some things that nobody "upstairs" wants to admit are happening. The entire point of this exercise is to cause some institutional panic in the "it's fine / it'll cost too much / take too long / you can't do that / we don't have enough data to justify that" crowd. Basically the engineer knows that it can be done and wants to show them what a small effort it took. Once that happens, they'll bring in the big boys and my stuff will go into the bin. But it's going to take many months of data gathering to get there.
  • Do you have space in the box?

    I’m thinking a smaller internal box to house the electronics with insulation between the two boxes. That way your electronics will probably remain cool even with the outside heat.
    My Prop boards: P8XBlade2, RamBlade, CpuBlade, TriBlade
    Prop OS (also see Sphinx, PropDos, PropCmd, Spinix)
    Website: www.clusos.com
    Prop Tools (Index) , Emulators (Index) , ZiCog (Z80)
  • JRoark - That information is good to know.
    "There is no penalty for failure. This is an experiment. My stuff is totally passive and has no control of the machine"
    Still, careful selection of components and giving everything a wide operational margin for temperature and electrical noise, will ensure some degree of success. A look-see at automotive and aircraft instrumentation is probably a good start IMHO.
  • Cluso99 wrote: »
    Do you have space in the box?

    I’m thinking a smaller internal box to house the electronics with insulation between the two boxes. That way your electronics will probably remain cool even with the outside heat.

    Surely equilibrium will eventually be reached and the electronics will be the same temperature as the environment?
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 22,442
    edited 2019-09-13 - 17:41:18
    Surely equilibrium will eventually be reached and the electronics will be the same temperature as the environment?
    Of course. In fact, with the insulation, and the waste heat being created from the electrical energy inserted into the insulated box with no way to get out, the inside of the box might become warmer than the interior of the larger box.

    -Phil
    “Perfection is achieved not when there is nothing more to add, but when there is nothing left to take away. -Antoine de Saint-Exupery
  • kwinnkwinn Posts: 8,329
    edited 2019-09-13 - 22:27:11
    Using a die cast box as the heat sink for the propeller board and then mounting that inside the doghouse (in the coolest area if there is one) to transfer the heat to the exterior surface of the doghouse might work. Have done that a few times in the past successfully, but not quite to those temperatures.
    In science there is no authority. There is only experiment.
    Life is unpredictable. Eat dessert first.
  • Provided the heat generated by the other parts inside the box is minimal, I would not expect the area inside the box to get too hot as the prop doesn't generate much heat. Regulators are the worst culprits for heat generation. This can of course be checked out quite easily. Then a peltzer cooler could be added if necessary.
    There is some great insulation available which would prevent the inner box from heating due to the outside area.
    My Prop boards: P8XBlade2, RamBlade, CpuBlade, TriBlade
    Prop OS (also see Sphinx, PropDos, PropCmd, Spinix)
    Website: www.clusos.com
    Prop Tools (Index) , Emulators (Index) , ZiCog (Z80)
  • jmgjmg Posts: 13,928
    Would the QFN have the best thermal conductivity thanks to its thermal pad?

    Yes, In theory, but that package is EOL.
    JRoark wrote: »
    It gets to 100c pretty routinely.

    If that is momentary enough, you could add layered ‘Russian Doll’ thermal zones and some thermal inertia to the inner most zone, so you get a thermal LPF effect.

    As mentioned, get good caps, and a top rated Automotive EEPROM, and regulator is going to need some care around their thermal shutdown points.

    Multilayer PCBs are quite cheap, and a good way to spread all the heat about, and so lower the peak temperatures. Throw as much copper at it as you can.

    What is the 3v3 derived from ? - A switching regulator may be better than LDO, and companies like Infineon have Automotive SMPS parts.
  • The problem isn't an internal heat build up. Genetix has the right answer, just underclock the prop1 so it will run calmly at 150 °C. It'll handle a high ambient better than other components like capacitors.
    "We suspect that ALMA will allow us to observe this rare form of CO in many other discs.
    By doing that, we can more accurately measure their mass, and determine whether
    scientists have systematically been underestimating how much matter they contain."
  • Build in a temperature sensor or two to the enclosure so the prop1 can record peak temperatures it has endured.
    "We suspect that ALMA will allow us to observe this rare form of CO in many other discs.
    By doing that, we can more accurately measure their mass, and determine whether
    scientists have systematically been underestimating how much matter they contain."
  • Cluso99 wrote: »
    There is some great insulation available which would prevent the inner box from heating due to the outside area.

    Not 'prevent' but 'slow down'.

    As it sounds like the electronics will be subjected to these high temperatures for long periods then they will eventually reach the same temperature as their surroundings.

    Basic thermodynamics. Heat flows from hot to cold.

  • Basic thermodynamics. Heat flows from hot to cold.

    Yeah, forget insulation, forget cooling, let the prop1 run at ambient. It isn't beyond what the prop1 can handle.

    The power supply might need some attention though.

    "We suspect that ALMA will allow us to observe this rare form of CO in many other discs.
    By doing that, we can more accurately measure their mass, and determine whether
    scientists have systematically been underestimating how much matter they contain."
  • Closing the loop here, at least for a bit, in the hope this helps someone else at some future date.

    It turns out we completely missed a very basic trick. There are two nitrogen supplies. A hot, variable temperature low pressure supply that runs about 50 PSI (max), and a high pressure supply that stays between 150 and 200 PSI. Probably 95% of the facility purge systems use the low pressure side, and this is what we were looking at using too. But that high pressure port is ideal for our uses. Why? Because when you drop a compressed gas from high to low pressure, it expands... AND ABSORBS HEAT.

    Duh. (Smacks head)

    The magic solution was to put a needle valve downstream of a flowmeter on the high pressure line. With the flowmeter at 5-7 CFM, the worst case outlet temperature after the oriface is about 140F. Most of the time it is about 20 degrees cooler. A short neoprene hose brings the cold, low-pressure gas right into the case with the Prop. Problem solved. And because the moisture has been largely removed from the stream by the site compressor/aftercooler, the nitrogen is pretty dry.

    So we are trying this approach for a bit to see how it works in the field.

    Amazing what you can miss when you are fixated on a problem!

  • jmgjmg Posts: 13,928
    0d03c235d2babe33c3e5231b805e98a1.jpg
  • Btw: at this point I have a box-stock FLIP module installed as a sacrificial lamb for testing. All it does is spool a couple of temperature readings every 15 seconds to a memory card. The unused cores are spinning their wheels doing various mathematical calisthenics to make some heat.

    If this test setup runs without errors, Phase 2 begins with a custom board and some real monitoring.
  • @jmg EXACTLY!
  • Great new! Thanks for the update.
    My Prop boards: P8XBlade2, RamBlade, CpuBlade, TriBlade
    Prop OS (also see Sphinx, PropDos, PropCmd, Spinix)
    Website: www.clusos.com
    Prop Tools (Index) , Emulators (Index) , ZiCog (Z80)
Sign In or Register to comment.