MOSFET temperature

[Hans B]

Hi all

This will be my first post in this forum.

I build two aleph 5 monoblocks with 32V rails and 2.45A bias, which gives a total dissipation around 160W pr. monoblock. Each
monoblock has a thermal resistance of 0.1C/W, which gives a temperature rise of minimum 16 degrees (measured around 20), and so far so good.

The problem is the temperature of the MOSFETs (IRFP240), which is around 100 degrees C - measured in a little rounding on the TO-247 package (perfect for a probe ).

Each MOSFET dissipates 40W (I use 2 in parallel, not 3) and are placed on a heatsink that has a thermal resistance of 0.4C/W. I use an insulation pad with the rating of 0.4 C/W (TO-3P size). Given 20 degree room temperature (danish weather prevents higher temperatures), the temperature of the MOSFET case should be:
T=T_ambient + PWR_dissipation*(THETA_cs + THETA_pad + THETA_heatsink)

which gives:
T = 20 + 40*(0.24 + 0.4 + 0.4) = 61.6 degrees Celcius.

And the MOSFET junction should be:
T = 61.6 + 40*0.82 = 94.4 degrees celcius

If the case temperature is 100C, then the MOSFET junction should be 100 + 40W*0.82 = 132,8, which I guess is pretty critical on a part rated at 155 degrees.

So my questions are:

What am I doing wrong?

Have I misunderstood the way of calculating the temperatures?

Is it the insulating pad that does not work properly (that is my guess so far)? - First I took a lot of thermal grease and covered the MOSFET with it, and then placed it on the pad. Then I measured and got 105 degrees (all measurements are done after approximately 30 minutes of running). Then I tried to remove all thermal grease and put the pads on without thermal grease and the temperature lowered a little (not significant). I have not yet tried to put a lot of grease between the pad and the heatsink (only the pad and the transistor. The datasheet of the pad can be found at:
http://www.elfa.se/pdf/75/07564602.pdf

Should I go for more MOSFETs?? (I hope not since this would mean a lot of extra work)


Hope you have some answers for me

Regards

Hans

Denmark

BTW: I build a BOSOZ which has 6 3W resistors which also approaches 100C (60V rails, 40mA running through each transistor) - Are these temperatures normal and should I just realize that these temperatures are this high when dealing with SE class A

[Henrik]

I have never ben able to calculate the right values.

But I think you might have a problem with your insulating pads, they shoud alwas be mica and some termal greese when running at the edges at the max. temperature. You can by mica insulators from LCAudio in Randers.

Resistors at 100 degree C shoud be no problem, just let them "fly" 3-5 mm over the boards. The BOSOZ at 60 v rails runs hot, specially the fets in the PSU if you runs two channeles from one supply, then you need more heatsink than described ind the project text.

Yes, dealing with SE class A is dealing with high tempreture.

[UrSv]

Originally posted by Hans B
Hi all

This will be my first post in this forum.

-- snip--

Is it the insulating pad that does not work properly (that is my guess so far)? - First I took a lot of thermal grease and covered the MOSFET with it, and then placed it on the pad. Then I measured and got 105 degrees (all measurements are done after approximately 30 minutes of running). Then I tried to remove all thermal grease and put the pads on without thermal grease and the temperature lowered a little (not significant). I have not yet tried to put a lot of grease between the pad and the heatsink (only the pad and the transistor. The datasheet of the pad can be found at:
http://www.elfa.se/pdf/75/07564602.pdf

-- snip --

The biggest problem might be that that pad should not have ANY grease at all to work well. It is to be used alone with nothing else applied and was specifically designed to replace mica+grease. Which pad or pad/grease combination you use is a little bit of personal taste but they should be used correctly to work as intended. Clean off all grease and try with pad only.

The best solution would be to use one more FET for each place (i.e. 3 instead of 2) but that might be some work as you mentioned.

BTW, welcome to the forum.

[Nelson Pass]

.

[audionut]

I think that running your Mosfets at those temps will cause failures...and failures can result in DC voltages across your speakers which means blown voice coils...I put my alephs in the basement and run a fan over them.

[Nelson Pass]

Something is way off, but your calculation appears correct.
How about measuring the temperature of the heat sink itself
next to the Mosfet?

[Mark A. Gulbrandsen]

Tektronix has neato temperatire probes that will allow measurements right on the device in question. Youmay want to get ahold of one and use that method. There are several DMM's that the probe works with. Thats how I check my devices temperatures.
Mark

[ashok]

You can also use Pt bead probes. These are just insulated wires with a welded bead at the tip. It would be less than 1mm in diameter. Only catch is that the DMM will need a Pt input or you can use a stand alone temp meter with inputs for this probe. I use it all the time and it works very well. I put a spot of heat sink paste on the bead when measuring flat surfaces. This improves the contact. They are not very expensive.
Cheers.

[Henrik]

Hans B´s difference in tempereture between the 40 degree C at the heatzink and the 100 degree C Case temperature at the mosfets is way off like Nelson just said, grease on silicon pads or not.

But any way, what is the best solution, mica/grease or silicon?

Since I have builded the SOZ several times and read Nelsons article many more times, I have alwas used mica/grease, but may be this is not the truth any more.
As I remember, I have read (some years ago) many articles / posts on this issue wich supports Nelsons point of view.

Nelson in the Son Of Zen article:

The Mosfets have electrically live cases, so mounting them on the heat sinks will involve the use of thermal grease and insulators. Mica insulators are the best. You can get insulators made of silicone rubber which do not need the grease, but they are not as good in my experience.

[Hans B]

Hi

Thank you for the responses.

Henrik
Thanks for the tip on LC-audio. I didn't know the sold mica. As you say in the latest post the temperature difference between heatsink and MOSFET are the main concern.


UrSv
As you mention, this pad should not have any thermal grease. I suspected that after using the thermal grease. I removed all thermal grease on heatsink, transistor and used new pads. Unfortunatly it did not improve much. I have tightened the M3 screw as much as I dare on the heatsink (aluminium is a rather soft material) so they sit firmly connected, and with no spacing to the heatsink (of cause).

Another FET could be the solution, however I think that since Nelson Pass states that something is way off, I would think that that might not solve the ``real'' problem.


audionut
I know these temperatures are dangerous, and I can tell you that the monoblocks are no where near my real system. They play on a crappy speaker in another room. Speakers some people refer to as the kind of speakers that sounds the best, when hitting the bottom of a trashcan , however they serve their purpose.

The problem is that I think a have an adequate amount of heatsink (there is never enough). The heatsinks are around 42 degrees C and I would have expected the MOSFETs to be around 20-25 degrees above that.


Nelson Pass
Thanks for the advice, this is my first diy amp, so I don't have much experience in this area. Do you think that 40W is to much (I know its high, but it would mean a lot of ekstra work to change it).

I did not expect the calculations to be incorrect, however I wanted to be sure. The temperature on the heatsink is of cause warmest close to the FET, but not more than a few degrees I would suspect (I will measure when I get home). And when turning the amp on, the MOSFET temperature rises very quickly, but the heatsinks takes a while to follow. I think you once said that most transistors burn, while on a cold heatsink, and that might be what will happen here if I don't get this fixed.

Mark
Measuring right on the device would be the best method, but I would suspect the temperatures to be even higher than the ones I have measured. I now know why Nelson prefers IRF for his transistors. They can take a beating



Any other ideas. So far I am thinking that it is the insulations pads that are not transferring the heat correctly. I think I am going to overgrease (if that is a word) the transistor, pad, and heatsink with
thermal grease and see what happens (so far I have tried only to grease the transistor. After that I am probably going to buy some micas like Henrik suggested.

Any comments would again be greatly appreciated. It might prevent me from doing this (love the icons)

Regards

Hans