I wonder if it tells what we want to know...

"Wrap the BS170 case with copper tape and solder it to another copper plate."

In the QMX, only the flat side is connected to the heat sink... (Assuming the washer and screw aren't very significant!) If that were taken into account, heat extraction via the case and via the drain lead might be more comparable, maybe even swapped?

You're making the assumption that temperature at the measurement point is a proxy for power flow on the path. But the actual temperature is a function of the thermal resistance from the measurement point to the environment...

It seems to me that you'd need to standardize the thermal resistance of what you attach the measurement points to, in order to measure relative power flow.

This is giving me a headache! Am I making it harder than it has to be?

Paul -- AI7JR

On 9/11/24 20:42, Sandy via ?wrote:

Hey Tony - That is an interesting result. There is no substitute for actual measurement. This suggests that case contact with the pad is important. I always install the BS170s first with the screw and washer ensuring they are all flat and only then solder in place. Thanks for your efforts! Sandy KB3EOF

On 9/11/24 22:18, mux_folder2001 via ?wrote:

I have attached a PDF file that describes an experiment with the goal of determine where the heat flows in a BS170. The PDF contains pictures of the fixture and a graph of the results. The essence of it is that, given equal opportunity, heat will flow as follows:

1. Best flow is through the case.
2. Second best flow is the drain lead.
3. A distant third is the source lead.
4. Last is the gate lead.

The graph will give you a sense of how much difference there is between them.?

The transistor is configured in a DC circuit and is dissipating 767mw continuously in the test. I should have put that in the PDF file but I forgot to.

Tony

AD0VC

-- 
Paul -- AI7JR

I ran tests with fiber washer/nylon screw and also with metal washer/metal screw. I also ran the TN0110.

Some experimental noise is evident I think but it is too big of a pain to redo everything. Hopefully some useful insight can be gained from this.

Temperature rise at center of identical heatsinks in degrees C
attached to the respective location
	Case	Drain	Source	Gate		total
BS170 max heatsink	4.5	3.7	2.4	1.9		12.5
BS170 Fiber Washer	3.8	2.4	2.1	1.4		9.7
BS170 Metal Washer	3.1	2.5	1.8	1.3		8.7
TN0110 max heatsink	4.1	3.2	1.6	1.4		10.3
TN0110 Fiber washer	3.3	3.3	1.6	1.4		9.6
TN0110 Metal Washer	4	3.6	1.8	1.5		10.9
TN0110 Reverse metal washer*	3	3.8	1.6	1.6		10
*Flat side against washer, round side against heatsink

Tony

AD0VC

---

From:[email protected] <[email protected]> on behalf of Hans Summers <hans.summers@...>
Sent:?Thursday, September 12, 2024 12:21 AM
To:[email protected] <[email protected]>
Subject:?Re: [QRPLabs] BS170 Heat flow experiment with results

?

Very nice experiment, Paul

Nice to see some real experimental data.?

I agree with Paul AI7JR, it would perhaps be interesting to measure with only the flat of the transistor face against the copper, since practically in a transceiver the curved part of the body is not touching the copper. I don't think there's any problem with the?measurements though; all the copper plates are the same size with the same dissipation to the environment so it's a reasonable relative comparison.

73 Hans G0UPL

On Thu, Sep 12, 2024 at 7:47?AM Paul - AI7JR via ?<paul.hanchett=[email protected]> wrote:

I wonder if it tells what we want to know...

"Wrap the BS170 case with copper tape and solder it to another copper plate."

In the QMX, only the flat side is connected to the heat sink... (Assuming the washer and screw aren't very significant!) If that were taken into account, heat extraction via the case and via the drain lead might be more comparable, maybe even swapped?

You're making the assumption that temperature at the measurement point is a proxy for power flow on the path. But the actual temperature is a function of the thermal resistance from the measurement point to the environment...

It seems to me that you'd need to standardize the thermal resistance of what you attach the measurement points to, in order to measure relative power flow.

This is giving me a headache! Am I making it harder than it has to be?

Paul -- AI7JR

On 9/11/24 20:42, Sandy via ?wrote:

Hey Tony - That is an interesting result. There is no substitute for actual measurement. This suggests that case contact with the pad is important. I always install the BS170s first with the screw and washer ensuring they are all flat and only then solder in place. Thanks for your efforts! Sandy KB3EOF

On 9/11/24 22:18, mux_folder2001 via ?wrote:

I have attached a PDF file that describes an experiment with the goal of determine where the heat flows in a BS170. The PDF contains pictures of the fixture and a graph of the results. The essence of it is that, given equal opportunity, heat will flow as follows:

1. Best flow is through the case.
2. Second best flow is the drain lead.
3. A distant third is the source lead.
4. Last is the gate lead.

The graph will give you a sense of how much difference there is between them.?

The transistor is configured in a DC circuit and is dissipating 767mw continuously in the test. I should have put that in the PDF file but I forgot to.

Tony

AD0VC

-- 
Paul -- AI7JR