Should MOSFET get warm?

Discussion in 'General Electronics Chat' started by Flug540, Oct 24, 2013.

  1. Flug540

    Thread Starter Member

    Mar 24, 2012
    60
    1
    I've built a PWM circuit using LM2902 opamps driving a 12V fan with irf8734pbf and stps1545d as a flyback diode. I ran the circuit for a few minutes with PWM advancing from 0% to roughly 40% duty cycle and both mosfet and the diode went up in temperature to about 40 deg celcius. Neither of them have any cooling at all. My clamp current meter measured 6A at the peak. Not having much experience in this I wonder if it's ok for the mosfet and diode to heat up like that or should they stay room temp?
     
  2. THE_RB

    AAC Fanatic!

    Feb 11, 2008
    5,435
    1,305
    40'C is only a fraction above human body temperature, so that's not really a heat problem. :)

    If you are concerned about heat rise, you should specify heat as "rise of X degrees ABOVE ambient temperature". That spec will directly match the amount of power dissipated by the components (ie; double the power, double the heat rise).

    Components in modern appliances will often run at 50'C or 60'C total temp, which is maybe 20'C to 25'C heat rise (assuming the ambient temp inside the enclosure is 30-35'C).
     
    anhnha and PackratKing like this.
  3. Flug540

    Thread Starter Member

    Mar 24, 2012
    60
    1
    Thanks, that's a good way to look at it :)
     
  4. tindel

    Active Member

    Sep 16, 2012
    568
    193
    Warm is okay - hot is not... Most FETs shouldn't exceed about 100degC - look at the FET spec. Reliability is also affected as temperature goes up... that's something to keep in mind. A good rule of thumb is to not exceed 110degC on any junction.
     
    anhnha likes this.
  5. #12

    Expert

    Nov 30, 2010
    16,261
    6,770
    Every part that has current flowing through it gets warmer, even copper wire. (That's why house wiring has a temperature molded into the outer insulation.) The thing that matters is, "how much". Regardless of what seems apparent, your transistor and diode do have some cooling method. If they didn't, they would keep getting hotter until they failed. The choices are radiation, convection, and conduction. Parts get along mostly by conducting their heat to the circuit board, but radiation and convection are usually happening too. Electronic components have their temperature rating listed on their data sheets. Your job is to make sure the parts do not get hotter than their rating.
     
    anhnha likes this.
  6. ScottWang

    Moderator

    Aug 23, 2012
    4,853
    767
    Talking about Temperature that is talking about the money and space and technology, the better design is under 40℃(104℉), if you can't reach that goal and then you should trying to keep the continuing temperature raising under 60℃(140℉) in one hour, it didn't means that when the temperature over 60℃(140℉) the parts will damaged.

    when the 7805 worked for the Z80 or 8085 learning system, the temperature is really high, and you can't touch it, because it could burning your fingers, but the 7805 still could survive for many years, but it will burn out someday, because the temperatur is too hot.

    If the space is enough and you have more money, then you could using the better parts, the bigger heat sink, even using the fan to reducing the heat.

    Sometimes we have to make a choice between the heat and money and space, how to get a better choice, to find out the balance point is then key.
     
    anhnha likes this.
  7. TorontoBob

    New Member

    Oct 5, 2013
    15
    4
    40C is fine, but is that under all load conditions? As The_RB said, what matters is the temperature rise over the ambient (ie. room) temperature. We don't know if your room temperature is 38C or 0C (hey I am in Toronto and it is about 7C outside). The cooler it runs the longer it will last, (be sure to keep any hot parts away from any electrolytics), but the old-timer rule-of-thumb used to be that if you could hold onto a heatsink and keep your finger on it, you probably didn't have any heat problems. (Just don't use this finger test with anything running voltages high enough to shock you!)
     
    anhnha likes this.
  8. Flug540

    Thread Starter Member

    Mar 24, 2012
    60
    1
    Thanks guys for the input! Somehow I did not get the new post notifications...

    Talking about different load conditions... my contraption worked for almost two days :) Toward the end of the second day I noticed that the fan does not come on anymore, turned out the wire that was soldered to the MOSFET source got loose, I suspect because the MOSFET got hot enough to melt the solder :eek:

    As far as the rating goes, I believe the MOSFET should be up to the task. From what I gathered the reason for overheating could be the time that it takes for the MOSFET to fully switch its state? Which makes me suspect that the waveform generated by my LM2902 comparator isn't good enough. BTW, the circuit is currently running at around 300Hz and the comparator is powered by 8V so the switching waveform is roughly between 0V and 7.5V. As an option I'm considering using a gate driver IC instead of directly driving by the opamp, but I wonder if my schematic is getting more complex then necessary.
     
  9. #12

    Expert

    Nov 30, 2010
    16,261
    6,770
    What schematic? I am having difficulty finding it.
     
  10. Flug540

    Thread Starter Member

    Mar 24, 2012
    60
    1
    Here it is, it doesn't have the current pot values (I believe I use 100k for Level pots), and the cap I believe is now 22n, but the rest of the schematic is correct. The idea here was to be able to pick any range of voltage (Vtemp input) between 0V and 3V, for example 600mV to 900mV, subtract 600mV from it using -Level pot to re-bias the curve, then amplify it to get the necessary slope and then offset to get proper intersection with the triangular wave using +Level. PWM output is directly connected to the MOSFET through a 33ohm resistor.

    EDIT: a couple more corrections: as I mentioned, I'm using LM2902 and U6b is powered by 8V.
     
    • PWM.JPG
      PWM.JPG
      File size:
      98.4 KB
      Views:
      69
    Last edited: Oct 28, 2013
  11. ian field

    Distinguished Member

    Oct 27, 2012
    4,413
    782
    Personally I'd aim for about 80c - the cooler it runs, the longer it lasts.

    Having said that; back in my TV repair days, once I was called out by a little old lady to fix a Ferguson 'portable' B&W set, the chassis was various board arranged like a box around the back of the CRT. The PSU board at the top being upside down - a TO5 transistor had got hot enough to unsolder itself, so when the little old lady thumped her telly, the transistor dropped to the bottom of the case. When I tested the transistor it was all OK, so I refitted it - once the rectifier reservoir electrolytic that caused it all had been replaced, the set worked perfectly.

    Some semiconductors are rated for max junction temp of 200c - its any plastics used in the construction that are more at risk.
     
  12. #12

    Expert

    Nov 30, 2010
    16,261
    6,770
    That is a very convoluted circuit to do a variable offset PWM! I bet a good designer could do it in less than 10 opamps.
    By the way, the output amp is labeled as 13.8 volt supply, not 8 volts as mentioned in post #8 and #10

    This is a very achievable goal. The frequency range is child's play. The MOSFET is entirely capable of doing this well. The opamp is capable, too. Something stinks, but it isn't the parts list. Do you have an oscilloscope?
     
  13. Flug540

    Thread Starter Member

    Mar 24, 2012
    60
    1
    Thanks a lot, honest and straight feedback is exactly what I'm looking for :) I'm just learning all this stuff as I go.

    In post #10 I added an edit that the posted schematic does not reflect the latest change I did to switch the output amp to 8V.

    I do have a scope and I checked the output wave and it looked OK. Although at the time I wasn't focusing on its "squareness", it looked square enough to me at the time. I'll check again and see what it looks like under magnification.

    Overall, do you think I need a gate driver or the opamp should be able to drive the MOSFET just fine?
     
  14. #12

    Expert

    Nov 30, 2010
    16,261
    6,770
    Post a drawing of how you have the mosfet connected. One of the newbie mistakes is to connect the motor from the source to ground instead of in the drain side.

    I don't think you need a gate driver.
     
  15. Flug540

    Thread Starter Member

    Mar 24, 2012
    60
    1
    The first schematic is the one that de-soldered itself. The second one is how I have it running today in an attempt to have a temporary solution.

    EDIT: please ignore markings on the diode and the MOSFETS.
     
  16. #12

    Expert

    Nov 30, 2010
    16,261
    6,770
    Those are both valid.
    I expect we can get this thing running on one mosfet.
    I expect that is what you want.

    Measure the rise time and the fall time on the gate of the mosfet.
    Try to find some way to get a tenth of an ohm or less into the source to ground circuit and look at the current the motor is using by looking at the voltage across the resistor.
    Yes, I know nobody has a .02 ohm resistor in their junk box.
    How about .3 ohms or less in the 13.8V line to the motor and scope the wave there?
    I have some .18 ohm, 2W resistors leftover from emitter matching in a power supply.

    I have to go load a truck before sunset, which is about 90 minutes from now.
     
    Last edited: Oct 28, 2013
  17. Flug540

    Thread Starter Member

    Mar 24, 2012
    60
    1
    Yes, one MOSFET is my goal. I'll get the measurements and report back. Yesterday stopped at my local electronics supply warehouse and got a 0.1ohm 100W power resistor.
     
  18. #12

    Expert

    Nov 30, 2010
    16,261
    6,770
    LOL! 100 watts! You will not be buying replacements for that one very often!:D
     
  19. Flug540

    Thread Starter Member

    Mar 24, 2012
    60
    1
    Hope not lol :D
    Figured it's better to have good head room.
     
  20. #12

    Expert

    Nov 30, 2010
    16,261
    6,770
    That one has room for head, shoulders, and a fat mother in law!:D

    I hope you got it in the surplus bin and paid about 98 cents.:p
     
Loading...