Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

Sticking a thermocouple junction to a FET case so as to measure its temperature

Status
Not open for further replies.
We have a D2PAK FET (SPB07N60) in our 146W offline LED light. In our thermal test, the thermocouple on the metal drain was at a temperature of 125 degC. The thermocouple on the plastic case of the same FET was at 118degC.
The FET was dissipating 11.8W

Do you think these temperatures are unrealistically close to each other?

I would have guessed that 2-3mm thick (assuming the heat producing die at at the center) plastic would drop around 10C around 100C. Your value is 125-118=7 drop at 125C. Your readings may be off by a deg. 7-8C drop at 125 looks small but the manufacturer goes lengths to reduce this value for better heat dissipation.

The short answer is NO
 
  • Like
Reactions: treez

    T

    Points: 2
    Helpful Answer Positive Rating
Thanks
Rth(jc) is 1.5 degc/w for the metal drain, but is 3.9degC/w for the Fullpak case (datasheet page 2).
Therefore the drain should have been at Tj + 17.7 degC.
The Fullpak case should have been at Tj + 46 degC.
In other words, there should have been 28 degrees difference between the temperature of the Fullpak case , and the metal drain,…….but we only saw a difference of 7 degrees.
Do you know what’s wrong.?

- - - Updated - - -

SPB07N60 datasheet:
https://www.infineon.com/dgdl/Infin...n.pdf?fileId=db3a304412b407950112b42df065491f
 

Let me try to figure out: remember that these are the max values; typical values will be about 25-50% lower (they are not specifying these numbers because they want you to design for the max values only).

The max difference (your estimates are correct) is about 30C between the drain and case. The typical value is given for the mounted device on PCB and the value is 50% for the minimum footprint (whatever that may be).

But look at graph 2: at 12W the case temp is around 105C; you measured 118. This difference is large but not unusual. At 120C, the dissipation is 8W; the difference is not negligible.

Did you check your thermocouple with boiling water (did it read 100C) - without care there may be upto 2C error in common thermocouple. You can safely assume that the readings are off by +/- 1C.

At 100C, the specified drain current is less than 5A; You are certainly exceeding the limit (I guess).

IS YOUR FET ALREADY FIXED ON THE PCB? Then you can only use the graph 2 on page 5.

My first guess was about 10C difference between the drain and case (I may be wrong!!!) but 30C is also very high (15 would have been acceptable).

But anyway, you are pushing the fellow to the limit. If the FET is mounted on the PCB, the drain temp will reduce and the gap (difference) between the drain and case temp would increase.

No, I am unable to guess what is going on.
 
  • Like
Reactions: treez

    T

    Points: 2
    Helpful Answer Positive Rating
I'm not sure what the assumptions about measured case temperature are based on - without any specifications of the conditions.

Air speed in case of forced cooling respectively horizontal versus vertical orientation of the case surface in case of natural convection will considerably affect the case surface temperature, also cross section of the thermo couple wire.

The other way around, if you want a good guess of junction temperature, cover the measuring point with a piece of foam and use the thinnest thermo couple you can get.
 
  • Like
Reactions: treez

    T

    Points: 2
    Helpful Answer Positive Rating
Thanks,
The D2PAK is on an FR4 PCB, there is a heatsink beneath the PCB (actually on top of it as the pcb hangs upside down on the heetsink) There are no vias in the d2pak pad, the thermal vias are on the thermal copper outside of the package footprint. There is no fan, and there are no ventilation holes inside the enclosure. Tyhe heatsink is the heatsink that the LEDs are also on. The fet is in the same enclosure as the 140w worth of leds.
The two thermocouples were glued with TBS20S thermal glue to the plastic case and the metal drain. The one on the drain was maybe some 0.2mm away from the drain, as we didnt want noise on the reading....but we wonder if noise has capacitively coupled in through the glue to the thermocouple?

- - - Updated - - -

Do yo uknow what the SPB07N60 datasheet is referring to when it keeps mentioning "FullPAK"?
Also, the graph 2 on page 5 (kndly pointed out by C_Mitra) says we shoudl not go above 110degC....but our drain is at 125degC. The contractors tell us to ignore the datasheet, and just note the fact that the led light was still working throughout its 3 hour stint in the thermal chamber.

- - - Updated - - -

On page 2 of the SPB07N60 datasheet it gives two values for Rth(JC)...one is for "fullpak" (3.9K/W)......does the "fullpak" one refer to the plastic of the case, and the other one refer to the metal drain tab?
SPB07N60 datasheet..
https://www.infineon.com/dgdl/Infin...n.pdf?fileId=db3a304412b407950112b42df065491f
 

It is rather common: manufacturers often are fond of naming their versions- I guess the FullPak and the D2PAK are the same but there may be minor differences.

On the last page they describe only one package spec: PG-TO263 (see the first page). So the different specs for thermal parameters are perhaps for different mounting conditions (I may be wrong!!).

But the graph 2 is certainly much more definitive. According to mouser, the same part is available also in TO220 package (it is not mentioned in the datasheet you have attached).
 

Thanks
Rth(jc) is 1.5 degc/w for the metal drain, but is 3.9degC/w for the Fullpak case (datasheet page 2).
Therefore the drain should have been at Tj + 17.7 degC.
The Fullpak case should have been at Tj + 46 degC.
In other words, there should have been 28 degrees difference between the temperature of the Fullpak case , and the metal drain,…….but we only saw a difference of 7 degrees.
Do you know what’s wrong.?
You calculations seem to assume that 11.8W is flowing through each path (for a total dissipation of 23.6W). In reality, the dissipated power will be divided among each path depending on the temperature difference between them. Assuming your measurements (and interpretations on the meaning of the two Rjc values) are accurate, 7.23W is flowing through the metal tab, 4.57W is flowing through the plastic case, and Tj=135.82C.

However this is moot because the two numbers you referenced describe completely different packages (fullpak is the name of infineon's TO220 with an insulated tab and wider pitch). Presumably the Tc values describe heat transfer through the tabs, not the plastic cases.
 
Last edited:
  • Like
Reactions: treez

    T

    Points: 2
    Helpful Answer Positive Rating
However this is moot because the two numbers you referenced describe completely different packages (fullpak is the name of infineon's TO220 with an insulated tab and wider pitch). Presumably the Tc values describe heat transfer through the tabs, not the plastic cases.
Thanks, i thought it must be something like that. But i wonder why the TO220 has such a poorer Rth(jc) than the D2PAK? (3.9K/W vs 1.5K/W)

Also, i cant understand why the datasheet for the D2PAK is going on about a TO220 device.
 

  • Like
Reactions: treez

    T

    Points: 2
    Helpful Answer Positive Rating
Also, the graph 2 on page 5 (kndly pointed out by C_Mitra) says we shoudl not go above 110degC....but our drain is at 125degC. ... ...

In the light of more recent information, I now think that we should be discussing about graph 1 instead; but the D2PAK is supposed to handle 20W at 120C!

Something is not apparently not right. At 100C it can handle 35W.

But anyway, the datasheet is badly written; it could have been as well in Chinese.
 
  • Like
Reactions: treez

    T

    Points: 2
    Helpful Answer Positive Rating
In the light of more recent information, I now think that we should be discussing about graph 1 instead; but the D2PAK is supposed to handle 20W at 120C!

Something is not apparently not right. At 100C it can handle 35W.
Not sure what you're referring to. Available power dissipation should be Ptot=(Tjmax-Tc)/Rjc. That's how the graph is derived.

But anyway, the datasheet is badly written; it could have been as well in Chinese.
While the lack of clarity about the FullPak package is annoying, everything else seems fairly normal.
 

Not sure what you're referring to. Available power dissipation should be Ptot=(Tjmax-Tc)/Rjc. That's how the graph is derived....

But what is available power dissipation? I always assumed that it is Total power dissipation.

The basic problem is that they did not mention the test conditions. At 150C (that is the Tjmax) case temp it does not dissipate any power. It suddenly appears counter intuitive.

In reality the computed graphs are of little use but they provide a go/ no-go rapid test for the engineers. If your point (Tc,Ptot) falls on the left hand side of the graph you are ok (else you are on your own!)

It is still impressive that the can handle 35W at 100C; with a heat sink that would be not a problem (the mounting cond spec on page 3, footnote 3, appear to be for a given spec case only) but as this is a theoretical curve the heat sink does not matter.

By the way, I am still confused whether the case temp refers to the metal mount underneath the belly or the black top?
 
  • Like
Reactions: treez

    T

    Points: 2
    Helpful Answer Positive Rating
By the way, I am still confused whether the case temp refers to the metal mount underneath the belly or the black top?
Thanks, myself also.
As discussed, in our test, the D2PAK was dissipating 11.8W, and the black case was at 118degC and the metal tab was at 125degC. I think something was wrong with our readings....possibly thermocouple too close to metal case, and noise getting capacitively coupled into it.
 

But what is available power dissipation? I always assumed that it is Total power dissipation.
Imagine you can perfectly control the case temperature by use of an external temperature controller. It basically means "if the case temperature is Tc, how much power can the device dissipate before Tj goes over Tjmax."

At 150C (that is the Tjmax) case temp it does not dissipate any power. It suddenly appears counter intuitive.
If your case temperature is already at Tjmax, you can't dissipate any power without Tj exceeding Tjmax.

In reality the computed graphs are of little use but they provide a go/ no-go rapid test for the engineers. If your point (Tc,Ptot) falls on the left hand side of the graph you are ok (else you are on your own!)
Right, they're not very practical, since you don't have a perfectly controlled Tc. And IMO you should stay 20C left of the curve if you want something reliable....

By the way, I am still confused whether the case temp refers to the metal mount underneath the belly or the black top?
Certainly the metal tab. In practice nearly all the thermal flux is going to take that path, so the plastic temperature is practically irrelevant.

Thanks, myself also.
As discussed, in our test, the D2PAK was dissipating 11.8W, and the black case was at 118degC and the metal tab was at 125degC. I think something was wrong with our readings....possibly thermocouple too close to metal case, and noise getting capacitively coupled into it.
I don't see anything implausible about your result. You could probably reverse the difference by blowing on the other side of the PCB, or covering the package with some a convection barrier.
 
Last edited:
  • Like
Reactions: treez

    T

    Points: 2
    Helpful Answer Positive Rating
Status
Not open for further replies.

Similar threads

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top