• AVR Freaks

Hot!core temperature of the processor?

Page: 1234 > Showing page 1 of 4
Author
edisonc
Senior Member
  • Total Posts : 155
  • Reward points : 0
  • Joined: 2008/11/10 21:11:09
  • Location: 0
  • Status: offline
2014/09/26 13:51:39 (permalink)
0

core temperature of the processor?

I want to find out max core temperature Tj of the microcontroller we are using, pic18F67J60, but I can not find it anywhere.
The datasheet says that chip will work in range of -40 to +85. But it locks up around 55C ambient.  I did thermal calculation based on the package, it shows that this particular part if locked up at 55C, its max Tj is only 101C.  Would that be too low for this industrial grade part?
 
Is is kind of disappointment for this chip not able to work in the full -40 to +85C range.  Does anyone know how much heatsink I should add to make it work in 60C or 70C?  I like passive heatsink not active fan if that could help cooling off this chip.
post edited by edisonc - 2014/09/26 14:52:12
#1

75 Replies Related Threads

    NKurzman
    A Guy on the Net
    • Total Posts : 18410
    • Reward points : 0
    • Joined: 2008/01/16 19:33:48
    • Location: 0
    • Status: online
    Re: core temperature of the processor? 2014/09/26 17:49:25 (permalink)
    +1 (1)
    -40 to +85 That is the DIE Not Ambient.
    What Clock rate?  The High the Hotter.
    Are your Sourcing or sinking a lot of current?  That builds up heat in the Die.
    What Package?  The degrees C per watt is in the data sheet.  Some packages are very Poor.
    #2
    edisonc
    Senior Member
    • Total Posts : 155
    • Reward points : 0
    • Joined: 2008/11/10 21:11:09
    • Location: 0
    • Status: offline
    Re: core temperature of the processor? 2014/09/26 23:24:18 (permalink)
    0
    Really?  The operational temperature is only +85 for the core?  That is below commercial grade.  Normally commercial grade semiconductor works up to 70C, industrial works above 85C military grade work above 125C, but those are ambient temperature.  This chip datasheet says industrial grade, but locks up at 55C.
    No, it does not have much load at all, 3mA to drive another gate on couple TTL output only.  There is only one speed for this chip to operate, since Ethernet PHY requires 40MHz can not change.
     
    The chip with Ethernet on takes 0.7W power, and the heat-power ration is 76, so that it will require like 53C from junction to ambient. If 85C is ambient temperature, it can only operate up to 32C, that is below human body temperature.  That can not be right.
    #3
    NKurzman
    A Guy on the Net
    • Total Posts : 18410
    • Reward points : 0
    • Joined: 2008/01/16 19:33:48
    • Location: 0
    • Status: online
    Re: core temperature of the processor? 2014/09/27 12:03:47 (permalink)
    +1 (1)
    I will assume all you decoupling and Core Caps are correct
    chapter 28.2 21.6ma at 3.3V and 41.6667 MHz = 0.07128 watts
    214ma for the Ethernet so that  0.77748 watts
    You can add up the peripherals but that is low
     
    you are allowed 1 W (chapter 28.0)
    You are allowed 250ma max so there is not a lot of drive left ( are you sure you are not grounding and output pin.)
    The Max Die Temp is 100C so the limit is 40C ( assuming your 76 C/W) worst case
    so 55C is not honorable.  But not what you need.
    The Ethernet is some Power Hog.
    Not sure about TQFN Heat sinks find one that brings the degrees C/Watt to where you need
    #4
    edisonc
    Senior Member
    • Total Posts : 155
    • Reward points : 0
    • Joined: 2008/11/10 21:11:09
    • Location: 0
    • Status: offline
    Re: core temperature of the processor? 2014/09/27 15:05:57 (permalink)
    0
    basically 0.7W x 76 C/W = 53.2C
    Max Ta = Tj -53.2C.
    Now question is what is max Tj, and if I add heatsink, would C/W ration reduce, and by how much.
    With global warming, any processor can only work 50C Ta is pretty much useless, can not be called commercial grade, I think.  We added a heatsink, so that it locks up at 55C, but that is still way lower than we expected. 
     
    My question is how big the heatsink required to get it runs above 60C and really desirable to operate at 70C Ta?
    #5
    NKurzman
    A Guy on the Net
    • Total Posts : 18410
    • Reward points : 0
    • Joined: 2008/01/16 19:33:48
    • Location: 0
    • Status: online
    Re: core temperature of the processor? 2014/09/27 15:58:59 (permalink)
    0
    the key is the 76C/W  that is constant to the surface of the chip.  the attachment to the chip matters. thermal paste or epoxy is needed.  If you find a heat sink made for the chip it may have the number that it improves by.
    other wise you need to measure the chip before and after the heat sink is add to get the new number.
    so if the chip surface drops 10C with the sink, then die does the same.  I assume ambient it the inside of the box not the outside.  if the box is thermally conductive you can sink to that.
    The QFNs are actually be thermally due to the center mounting pad.
     
    Note the on and transmitting numbers.  If you do not transmit alot your avg. power is lower
    #6
    edisonc
    Senior Member
    • Total Posts : 155
    • Reward points : 0
    • Joined: 2008/11/10 21:11:09
    • Location: 0
    • Status: offline
    Re: core temperature of the processor? 2014/09/27 16:07:14 (permalink)
    0
    76 C/W is constant for chip without any cooling.  A lot of cooling is done through pins, not through chip surface.  I think Microchip should provide information for how to calculate when adding certain cooling area through heatsink.  When this chip datasheet says -40C to +85C, I do expect the chip working in the whole range, or tell me how to make it work in that whole range.  We are in a bad spot after manufactured 5000 boards, suddenly found it does not work within expected temperature range.
     
    We tried to get straight answer for how to make it work, by passive cooling heatsink or active fan, but we got no reply from Microchip.
    #7
    NKurzman
    A Guy on the Net
    • Total Posts : 18410
    • Reward points : 0
    • Joined: 2008/01/16 19:33:48
    • Location: 0
    • Status: online
    Re: core temperature of the processor? 2014/09/27 17:02:50 (permalink)
    0
    It does not list the location but in the U.S. you would contact your F.A.E.  or the Distributor.
     
    You can put in a ticket to get Help from Microchip.
     
    "When this chip datasheet says -40C to +85C, I do expect the chip working in the whole range"
    I am mostly a firmware guy so I am not sure if that is a reasonable expectation or not.
    It is Toasty for a PIC18.
    #8
    edisonc
    Senior Member
    • Total Posts : 155
    • Reward points : 0
    • Joined: 2008/11/10 21:11:09
    • Location: 0
    • Status: offline
    Re: core temperature of the processor? 2014/09/27 22:07:56 (permalink)
    0
    We put in a ticket to Microchip support for over a week, got no real answer.  The person said he was expecting to get answer from another person in Microchip. But five days got nothing back.  He could not even tell me 85C is core temperature or case temperature or ambient temperature. But he insisted if Microchip said +85C, the chip will be able to work up to 85C.  I showed him the thermal calculation for ambient 54C based on 100C core temperature, and ask if there is another formula to use for add heatsink. He could not answer.
     
    If a datasheet saying operation temperature range is so and so, I would expect that is the ambient temperature it can operates with either free air cooling, or with some specified method of cooling.  Check Wikipedia for IC temperature range, it cited another chip maker industrial spec to work in ambient temperature up to 95C.  If I knew this thing only working up to 55C, I would not recommend selected this chip from first place. 
    #9
    tmr_suresh
    Senior Member
    • Total Posts : 105
    • Reward points : 0
    • Joined: 2012/09/13 04:00:07
    • Location: 0
    • Status: offline
    Re: core temperature of the processor? 2014/09/28 00:29:43 (permalink)
    0 (1)
    usually as product development engineers when we require technical help. but  most of the time FAE are unqualified to help us. it has happened to me many times in india . once  I had contacted international rectifier FAE to solve a issue in PFC controller IR1150. even I have sent a prototype pcb to Bangalore to get it work properly .
     
    this is the case mostly with all big companies.
     
    experienced forum members  usually help better (free ) than FAE (who are paid to help).
    #10
    RISC
    Super Member
    • Total Posts : 5392
    • Reward points : 0
    • Status: offline
    Re: core temperature of the processor? 2014/09/28 02:33:42 (permalink)
    +1 (1)
    Hi,
    The temperature range indicated in the PICs datasheet is always the ambient temperature.
    If the datasheet indicates -40 to +85C operating temperature, this is garanteed operation.
    If you device locks up there are 2 possibilities :
    1/ a design issue in your system
    2/ device issue ( probability here is about 10ppm as this is the average defect rate of microcontrollers )
     
    So in general issues are much more likely to be searched in the above category 1/, especially if you find several devices which behave the same.
     
    55C ambient temperature is absolutely not an issue. PIC do never need heatsinks. Their package can withstand the max garanteed temperature.
     
    Can you please show you schematic diagram ?
    Are you using a 3.3V power supply ?
    Did you pay attention not to connect any pin of the PIC to an external voltage ABOVE 3.3V.
     
    The PIC18FxxJ60 tend to heat a little because the ethernet PHY is on board but nothing extreme.
     
    Regards
    #11
    Mysil
    Super Member
    • Total Posts : 3667
    • Reward points : 0
    • Joined: 2012/07/01 04:19:50
    • Location: Norway
    • Status: offline
    Re: core temperature of the processor? 2014/09/28 04:32:47 (permalink)
    +3 (2)
    Hi,
    Seems to me that microcontrollers and similar are usually low power devices that usually get by without much attention to temprature and heatsinking.
     
    In this chip, it seem the CPU itself use 20 mA or less,
    but the Ethernet module add between 130 and 214 mA of supply current, causing some 700 mW of heat.
    Then  the chip is packaged in a plastic package without even a thermal pad.
     
    Microchip datasheet really specify absolute maximum ambient temprature = 100 C
    and maximum total power dissipation = 1 W,
    but I think any common sense would aim to keep the chip temprature below 100 C or even below 85 C 
    and provide heatsinking and/or cooling to achieve that.
     
    A chip packaging company lists the thermal resistance of TQFP 100 pin package 14x14mm as 43 C/W, 
    and 32 pin 7x7 mm package as 69.3 C/W in still air on a single layer PCB.
     
    Regards,
       Mysil
    #12
    edisonc
    Senior Member
    • Total Posts : 155
    • Reward points : 0
    • Joined: 2008/11/10 21:11:09
    • Location: 0
    • Status: offline
    Re: core temperature of the processor? 2014/09/28 16:20:41 (permalink)
    0
    RISC
    Hi,
    The temperature range indicated in the PICs datasheet is always the ambient temperature.
    If the datasheet indicates -40 to +85C operating temperature, this is garanteed operation.
    If you device locks up there are 2 possibilities :
    1/ a design issue in your system
    2/ device issue ( probability here is about 10ppm as this is the average defect rate of microcontrollers )
     
    So in general issues are much more likely to be searched in the above category 1/, especially if you find several devices which behave the same.
     
    55C ambient temperature is absolutely not an issue. PIC do never need heatsinks. Their package can withstand the max garanteed temperature.
     
    Can you please show you schematic diagram ?
    Are you using a 3.3V power supply ?
    Did you pay attention not to connect any pin of the PIC to an external voltage ABOVE 3.3V.
     
    The PIC18FxxJ60 tend to heat a little because the ethernet PHY is on board but nothing extreme.
     
    Regards




    Sorry, I can not show you schematics. It is our company's product that we made 5000 of them. We only tested ourselves to 45C, but customer who running on 55C told us during hot day, it locked up, when weather cooled off in the evening, it started working again.  The first thing quit is the Ethernet, which both MAC and PHY is in this processor.  The chip only works on 3.3V, through a dedicate regulator chip and verified.
    #13
    edisonc
    Senior Member
    • Total Posts : 155
    • Reward points : 0
    • Joined: 2008/11/10 21:11:09
    • Location: 0
    • Status: offline
    Re: core temperature of the processor? 2014/09/28 16:25:33 (permalink)
    0
    Mysil
    Hi,
    Seems to me that microcontrollers and similar are usually low power devices that usually get by without much attention to temprature and heatsinking.
     
    In this chip, it seem the CPU itself use 20 mA or less,
    but the Ethernet module add between 130 and 214 mA of supply current, causing some 700 mW of heat.
    Then  the chip is packaged in a plastic package without even a thermal pad.
     
    Microchip datasheet really specify absolute maximum ambient temprature = 100 C
    and maximum total power dissipation = 1 W,
    but I think any common sense would aim to keep the chip temprature below 100 C or even below 85 C 
    and provide heatsinking and/or cooling to achieve that.
     
    A chip packaging company lists the thermal resistance of TQFP 100 pin package 14x14mm as 43 C/W, 
    and 32 pin 7x7 mm package as 69.3 C/W in still air on a single layer PCB.
     
    Regards,
       Mysil


    Base on the power it consumes and thermal resistance, if core temperature is 100C, this chip can only work on 54C.  We have asked Microchip to provide us information on what is the max core temperature, and how to calculate thermal resistance if we added heatsink. But we got no answer.
     
    One thing we did got from Microchip was a QA test report, in which, Microchip stated 55C as derated temperature during their thermal testing.  What derated means part of the sub system must be shutdown.  But that was never mentioned in the datasheet.
     
    post edited by edisonc - 2014/09/28 16:27:02
    #14
    NKurzman
    A Guy on the Net
    • Total Posts : 18410
    • Reward points : 0
    • Joined: 2008/01/16 19:33:48
    • Location: 0
    • Status: online
    Re: core temperature of the processor? 2014/09/29 09:23:05 (permalink)
    0 (1)
    Which package do you have?
     
    A  Heat sink will help
    http://www.newark.com/fischer-elektronik/ick-smd-a-8-sa/heat-sink-dip-ssop-tssop/dp/34M6428
    a U section glued to the top may be enough.
    #15
    edisonc
    Senior Member
    • Total Posts : 155
    • Reward points : 0
    • Joined: 2008/11/10 21:11:09
    • Location: 0
    • Status: offline
    Re: core temperature of the processor? 2014/09/29 10:14:47 (permalink)
    0
    There is only one package for this chip, no choice.
    We had already installed a little heatsink on top the surface.  I suspect pins working better than chip surface, although the CPU surface heatsink is hot to hold with 35C ambient temperature already.  I need to have a calculation formular for the area size of the heatsink, or if active air, how much, to let this chip work in the specified working temperature range. 
    #16
    Mysil
    Super Member
    • Total Posts : 3667
    • Reward points : 0
    • Joined: 2012/07/01 04:19:50
    • Location: Norway
    • Status: offline
    Re: core temperature of the processor? 2014/09/29 13:34:51 (permalink)
    0
    Hi,
    If the heatsink is noticeably hotter than the ambient air when the chip is running,
    then a bigger heatsink will help more.
    You may calculate thermal conduction and resistance analog to a network of resistors in parallel and series,
    and the formula you need is essentially the same as Ohm's law.
     
    The package has 3 ways to conduct heat:
       Top surface into any heatsink you mount there,
       Connecting pins into the circuit board, and
       Bottom surface into circuit board and all the thermal via's you design there to the other side of the board and  heatsink you mount there.
     
    These 3 thermal paths act in parallel, and each path consists of at least 2 thermal resistances in series.
       Junction to package top surface in series with heatsink bottom surface to ambient air,
       Junction to connecting pins, in series with circuit board to ambient air,
       Junction to bottom surface, in series with thermal via's and heatsink on underside to air.
     
    Thermal resistance for any heatsink you buy should be in datasheet for the part.
    Thermal resistance for junction to package surface for 10x10 mm TQFP, I do not know, 
    but somewhat optimistic assuming 20 C/W for entire package,
    you may need 20 C/W or 10 C/W heatsink both top and bottom to achieve 100 C junction temprature with 70 C air temprature.
     
    Regards,
       Mysil
     
     
     
      
     
    #17
    edisonc
    Senior Member
    • Total Posts : 155
    • Reward points : 0
    • Joined: 2008/11/10 21:11:09
    • Location: 0
    • Status: offline
    Re: core temperature of the processor? 2014/09/29 13:48:12 (permalink)
    0 (1)
    I am facing angry customers that saying this +85C range is a plain lie. I need exact solution to help my customers.  There is no way for me to tell customers to add heatsink on the bottom side of the PCB, since four layer PCB was not designed for conduct much heat through the fiberglass material.
     
    At the 40C+, the top of the chip is too hot to keep finger on it. But I don't know how to determine if heatsink on that side alone is enough or not.
     
    We had made 5000 of this board, and plan to make another 5000, now, we must stop and solve this problem first.  Personally, I think Microchip must redesign the package, so that heat can be transferred to heatsink much better than current design.  I wonder if they had any QA on this part to run it in ambient temperature up to 60C or 70C, which is commercial grade, or 85C to 95C, that is industrial grade.  For an industrial grade part can not even work in 50C range, that is embarrassing.  
    #18
    NKurzman
    A Guy on the Net
    • Total Posts : 18410
    • Reward points : 0
    • Joined: 2008/01/16 19:33:48
    • Location: 0
    • Status: online
    Re: core temperature of the processor? 2014/09/29 15:26:24 (permalink)
    +2 (2)
    You can try to convince them that they should redesign the Package.
    I am not sure to what.  That is the Limit of the Plastic package.
    Is the 85C a lie? about as much as half the "information" on the first page of any data sheet.
    Is there a pin compatible PIC that supports the 125C range?
     
    Put a thermocouple on the chip measure it.  Add a Heat sink, Measure the chip again. That will tell you how far you need to go.
    The weight of the sink and your shock and vibe requirements will tell if you need to glue down the chip.
    If the PCB layer under the chip has a lot of copper you can add thermal grease or epoxy.
     
    Is the PCB in ambient environment, or a box. 70C may be an issue from the PCB heating alone.
     
    If you are in the U.S. find your F.A.E.
    I do not know what the support in the Rest of the world is.
    #19
    Mysil
    Super Member
    • Total Posts : 3667
    • Reward points : 0
    • Joined: 2012/07/01 04:19:50
    • Location: Norway
    • Status: offline
    Re: core temperature of the processor? 2014/09/29 18:40:13 (permalink)
    +1 (1)
    Hi,
    My guess, he is in India or some other tropical country.
     
    In my opinion, this device need a TQFP with exposed thermal pad.
    Even then, if it is to work in 85 C air temprature with junction temprature below 100 C,
    it will need either forced air cooling (500 LFPM),
    or a heatsink in thermal contact with the pad and air circulation.
     
       Mysil
    #20
    Page: 1234 > Showing page 1 of 4
    Jump to:
    © 2020 APG vNext Commercial Version 4.5