estimating junction temperature of a power mosfet

[John Larkin]

On Fri, 23 Jan 2015 13:31:20 +0100, "Alain Coste" <coste@irit.fr>
wrote:

Thank you John for the interesting information. The use of mosfets in linear
mode is not very common, and it's more difficult to find data than for
switch mode.

It often makes sense to use more fets, spread out over the heatsink
surface, especially if the baseplate part of the heat sink is thin,
namely has high thermal spreading resistance.

This uses copper heat spreaders to transfer the heat into the aluminum
sink.

https://dl.dropboxusercontent.com/u/53724080/Thermal/Amp.jpg


Now I see what to use _more_ fets means...
For my electronic load I could have used more transistors, but this
increases the number of current sense resistors and operational amplifiers
to control them. For the power I wanted (400 .. 420W), I thought that two
mosfets was a good compromise.

If the hot spot Tt is only 100C at 150W, you should be fine.

If I limit the power to 300W (two mosfets in //) I have enough confidence
that Tj max will not be exceeded.
But when I designed the electronic load I had counted on around 400W.
Well, 300W are not so bad for my needs, but I would like to know more
precisely "how far I can go too far".

It would be fun to somehow remove the epoxy and image the actual chip,
to see its temperature profile.
I removed the epoxy from a bunch of mosfets, but the process was sort
of violent.

https://dl.dropboxusercontent.com/u/53724080/Parts/ExFets.jpg

I think there are organic epoxy removers that might not trash the
silicon.

Now I see what _violent_ means...
Considering the price of my fets, I didn't really want to try it...

We have found the Ixys mosfets to be good for surviving linear-mode
high-dissipation pulses, out in the northeast corner of the SOAR
curve. We blew up a lot of fets to learn that.

Yes, my selection of Ixyx mosfets owes nothing to chance. If you want at the
same time very low theta-jc and guaranteed SOA for DC (and not only for
switch mode) the choice is rather limited. International Rectifier and
Infineon had some fets with guaranteed SOA for DC, but for theta-jc Ixys was
the best. The counterpart is the price, which doesn't encourage to
destructive tests...

Given that heat sinks and huge mosfets are expensive, and opamps and
resistors are cheap, it often makes sense to use a lot of small fets,
to spread out the heat. And dump as much of the power as possible into
resistors.

A switchmode load box would be interesting. Fets switch and dump the
power into a big resistor.


--

John Larkin Highland Technology, Inc
picosecond timing laser drivers and controllers

jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com

 

[George Herold]

On Friday, January 23, 2015 at 11:45:16 AM UTC-5, Jan Panteltje wrote:

On a sunny day (Fri, 23 Jan 2015 07:54:21 -0800 (PST)) it happened George
Herold <gherold@teachspin.com> wrote in
9263cc76-7046-4a61-8a13-280fc4c0d917@googlegroups.com>:

OK first the numbers are almost too good to believe, but here they are.
(I only tested 4 FET's I had to solder wires on to go down into the dewar. The first one I had clip leads, but the dang thick
plastic on the leads froze up, and I was afraid I'd have to break it to get it back out of the dewar neck.)

V V
Room temp. 77K
0.5500 1.0154
0.5504 1.0156
0.5508 1.0157
0.5509 1.0159

What current did you use? Any self-heating?

This was with my fluke DMM (I think ~ 1mA)

George H.

 

[Jan Panteltje]

On a sunny day (Fri, 23 Jan 2015 07:54:21 -0800 (PST)) it happened George
Herold <gherold@teachspin.com> wrote in
<9263cc76-7046-4a61-8a13-280fc4c0d917@googlegroups.com>:

OK first the numbers are almost too good to believe, but here they are.
(I only tested 4 FET's I had to solder wires on to go down into the dewar. The first one I had clip leads, but the dang thick
plastic on the leads froze up, and I was afraid I'd have to break it to get it back out of the dewar neck.)

V V
Room temp. 77K
0.5500 1.0154
0.5504 1.0156
0.5508 1.0157
0.5509 1.0159

What current did you use? Any self-heating?

 

[John Larkin]

On Fri, 23 Jan 2015 07:54:21 -0800 (PST), George Herold
<gherold@teachspin.com> wrote:

On Friday, January 23, 2015 at 8:31:21 AM UTC-5, George Herold wrote:
On Thursday, January 22, 2015 at 3:18:56 PM UTC-5, John Larkin wrote:
On Thu, 22 Jan 2015 11:56:15 -0800 (PST), George Herold
gherold@teachspin.com> wrote:

On Thursday, January 22, 2015 at 8:35:11 AM UTC-5, Alain Coste wrote:
snip
I measured the diode drop on ~10 IRF820's in my part box and they were all within 1 mV of each other. (Not bad) (Don't touch with hands while measuring.)

George H.

That's impressive, less than 0.5 C.

Yeah the diodes (transistors) I use for temp sensors have ~5-10 mV
of variation at room temp.
Hey, I've got a dewar of LN2 next to me..
I'll dunk some in and see how well they track each other.

later...
George H.

OK first the numbers are almost too good to believe, but here they are.
(I only tested 4 FET's I had to solder wires on to go down into the dewar. The first one I had clip leads, but the dang thick plastic on the leads froze up, and I was afraid I'd have to break it to get it back out of the dewar neck.)

V V
Room temp. 77K
0.5500 1.0154
0.5504 1.0156
0.5508 1.0157
0.5509 1.0159

That's amazing. Lakeshore won't like that.

But I've found same-part same-manufacturer semiconductors to be very
repeatable. Which sometimes lets one use measured specs that are way
better than guaranteed specs.

Hey, what should I connect the gate to? (S or D or doesn't it matter?)

S or D but not floating. You don't want the channel to conduct and
short out the substrate diode. You can bias a mosfet ON and disconnect
the gate, and it may stay on for days.


>I left it floating, which was kinda fun 'cause if I scuffed my shoe and touched the gate the channel would conduct.

Reminds me of my fun-with-a-2N7000 experiments

https://dl.dropboxusercontent.com/u/53724080/Parts/Fets/2N7000.jpg

Gate leakage on a good 2N7000 is not many electrons per second.



--

John Larkin Highland Technology, Inc
picosecond timing laser drivers and controllers

jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com

 

[Phil Hobbs]

On 01/23/2015 11:49 AM, John Larkin wrote:

On Fri, 23 Jan 2015 07:54:21 -0800 (PST), George Herold
gherold@teachspin.com> wrote:

On Friday, January 23, 2015 at 8:31:21 AM UTC-5, George Herold wrote:
On Thursday, January 22, 2015 at 3:18:56 PM UTC-5, John Larkin wrote:
On Thu, 22 Jan 2015 11:56:15 -0800 (PST), George Herold
gherold@teachspin.com> wrote:

On Thursday, January 22, 2015 at 8:35:11 AM UTC-5, Alain Coste wrote:
snip
I measured the diode drop on ~10 IRF820's in my part box and they were all within 1 mV of each other. (Not bad) (Don't touch with hands while measuring.)

George H.

That's impressive, less than 0.5 C.

Yeah the diodes (transistors) I use for temp sensors have ~5-10 mV
of variation at room temp.
Hey, I've got a dewar of LN2 next to me..
I'll dunk some in and see how well they track each other.

later...
George H.

OK first the numbers are almost too good to believe, but here they are.
(I only tested 4 FET's I had to solder wires on to go down into the dewar. The first one I had clip leads, but the dang thick plastic on the leads froze up, and I was afraid I'd have to break it to get it back out of the dewar neck.)

V V
Room temp. 77K
0.5500 1.0154
0.5504 1.0156
0.5508 1.0157
0.5509 1.0159

That's amazing. Lakeshore won't like that.

But I've found same-part same-manufacturer semiconductors to be very
repeatable. Which sometimes lets one use measured specs that are way
better than guaranteed specs.


Hey, what should I connect the gate to? (S or D or doesn't it matter?)

S or D but not floating. You don't want the channel to conduct and
short out the substrate diode. You can bias a mosfet ON and disconnect
the gate, and it may stay on for days.


I left it floating, which was kinda fun 'cause if I scuffed my shoe and touched the gate the channel would conduct.

Reminds me of my fun-with-a-2N7000 experiments

https://dl.dropboxusercontent.com/u/53724080/Parts/Fets/2N7000.jpg

Gate leakage on a good 2N7000 is not many electrons per second.

The unprotected 2N7002E parts are going away, though. Better buy a few
reels while they last!

Cheers

Phil Hobbs


--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics, Electro-optics, Photonics, Analog Electronics

160 North State Road #203
Briarcliff Manor NY 10510

hobbs at electrooptical dot net
http://electrooptical.net

 

[whit3rd]

On Friday, January 23, 2015 at 7:54:28 AM UTC-8, George Herold wrote:

On Friday, January 23, 2015 at 8:31:21 AM UTC-5, George Herold wrote:
On Thursday, January 22, 2015 at 3:18:56 PM UTC-5, John Larkin wrote:
On Thu, 22 Jan 2015 11:56:15 -0800 (PST), George Herold
gherold@teachspin.com> wrote:

On Thursday, January 22, 2015 at 8:35:11 AM UTC-5, Alain Coste wrote:
snip
I measured the diode drop on ~10 IRF820's in my part box and they were all within 1 mV ...

That's impressive, less than 0.5 C.

Hey, I've got a dewar of LN2 next to me..
I'll dunk some in and see how well they track each other.

(I only tested 4 FET's I had to solder wires on to go down into the dewar

V V
Room temp. 77K
0.5500 1.0154
0.5504 1.0156
0.5508 1.0157
0.5509 1.0159

So, it's up to plus/minus a part per thousand? That's encouraging, it translates
to a fraction of a degree K (or C) . It's hard to be sure of 'room temp' matching, an oil
or Fluorinert bath would be a prudent enhancement for that column. The 77K numbers
are matched to 3 parts per ten thousand...

Screw torque and bubbles in the heatsink goo are a bigger variable than the
thermometric precision of the MOSFET junction.

Torque on the screw fasteners and dust on the fan blades is a bigger uncertainty than
the

 

[John Larkin]

On Fri, 23 Jan 2015 12:11:26 -0500, Phil Hobbs
<pcdhSpamMeSenseless@electrooptical.net> wrote:

On 01/23/2015 11:49 AM, John Larkin wrote:
On Fri, 23 Jan 2015 07:54:21 -0800 (PST), George Herold
gherold@teachspin.com> wrote:

On Friday, January 23, 2015 at 8:31:21 AM UTC-5, George Herold wrote:
On Thursday, January 22, 2015 at 3:18:56 PM UTC-5, John Larkin wrote:
On Thu, 22 Jan 2015 11:56:15 -0800 (PST), George Herold
gherold@teachspin.com> wrote:

On Thursday, January 22, 2015 at 8:35:11 AM UTC-5, Alain Coste wrote:
snip
I measured the diode drop on ~10 IRF820's in my part box and they were all within 1 mV of each other. (Not bad) (Don't touch with hands while measuring.)

George H.

That's impressive, less than 0.5 C.

Yeah the diodes (transistors) I use for temp sensors have ~5-10 mV
of variation at room temp.
Hey, I've got a dewar of LN2 next to me..
I'll dunk some in and see how well they track each other.

later...
George H.

OK first the numbers are almost too good to believe, but here they are.
(I only tested 4 FET's I had to solder wires on to go down into the dewar. The first one I had clip leads, but the dang thick plastic on the leads froze up, and I was afraid I'd have to break it to get it back out of the dewar neck.)

V V
Room temp. 77K
0.5500 1.0154
0.5504 1.0156
0.5508 1.0157
0.5509 1.0159

That's amazing. Lakeshore won't like that.

But I've found same-part same-manufacturer semiconductors to be very
repeatable. Which sometimes lets one use measured specs that are way
better than guaranteed specs.


Hey, what should I connect the gate to? (S or D or doesn't it matter?)

S or D but not floating. You don't want the channel to conduct and
short out the substrate diode. You can bias a mosfet ON and disconnect
the gate, and it may stay on for days.


I left it floating, which was kinda fun 'cause if I scuffed my shoe and touched the gate the channel would conduct.

Reminds me of my fun-with-a-2N7000 experiments

https://dl.dropboxusercontent.com/u/53724080/Parts/Fets/2N7000.jpg

Gate leakage on a good 2N7000 is not many electrons per second.

The unprotected 2N7002E parts are going away, though. Better buy a few
reels while they last!

Cheers

Phil Hobbs

I don't really care about gate leakage in a 2N7000, except for fun
things.

But I hope the gate protection zener is symmetrical, in case I have
some old application where I pull the gate negative.


--

John Larkin Highland Technology, Inc
picosecond timing precision measurement

jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com

 

[George Herold]

On Friday, January 23, 2015 at 6:13:38 PM UTC-5, whit3rd wrote:

On Friday, January 23, 2015 at 7:54:28 AM UTC-8, George Herold wrote:
On Friday, January 23, 2015 at 8:31:21 AM UTC-5, George Herold wrote:
On Thursday, January 22, 2015 at 3:18:56 PM UTC-5, John Larkin wrote:
On Thu, 22 Jan 2015 11:56:15 -0800 (PST), George Herold
gherold@teachspin.com> wrote:

On Thursday, January 22, 2015 at 8:35:11 AM UTC-5, Alain Coste wrote:
snip
I measured the diode drop on ~10 IRF820's in my part box and they were all within 1 mV ...

That's impressive, less than 0.5 C.

Hey, I've got a dewar of LN2 next to me..
I'll dunk some in and see how well they track each other.


(I only tested 4 FET's I had to solder wires on to go down into the dewar

V V
Room temp. 77K
0.5500 1.0154
0.5504 1.0156
0.5508 1.0157
0.5509 1.0159

So, it's up to plus/minus a part per thousand? That's encouraging, it translates
to a fraction of a degree K (or C) . It's hard to be sure of 'room temp' matching, an oil
or Fluorinert bath would be a prudent enhancement for that column. The 77K numbers
are matched to 3 parts per ten thousand...

Screw torque and bubbles in the heatsink goo are a bigger variable than the
thermometric precision of the MOSFET junction.

Torque on the screw fasteners and dust on the fan blades is a bigger uncertainty than
the

Yeah, the room temperature numbers were drifting all over the place.
Fet's just lying on my lab bench, taped to a piece of paper. Just bolting 'em down to a single hunk of metal would help a lot. I'm not all that surprised by the decrease in variation with temperature. I've seen that before.

George H.

PS. I need to post an epoxy update, the glass beads failed miserably,
Cracks everywhere, as you predicted.

 

[John Larkin]

On Fri, 23 Jan 2015 16:45:23 GMT, Jan Panteltje <panteltje@yahoo.com>
wrote:

On a sunny day (Fri, 23 Jan 2015 07:54:21 -0800 (PST)) it happened George
Herold <gherold@teachspin.com> wrote in
9263cc76-7046-4a61-8a13-280fc4c0d917@googlegroups.com>:

OK first the numbers are almost too good to believe, but here they are.
(I only tested 4 FET's I had to solder wires on to go down into the dewar. The first one I had clip leads, but the dang thick
plastic on the leads froze up, and I was afraid I'd have to break it to get it back out of the dewar neck.)

V V
Room temp. 77K
0.5500 1.0154
0.5504 1.0156
0.5508 1.0157
0.5509 1.0159

What current did you use? Any self-heating?

Cryo-diodes are usually run at 10 uA, but a big mosfet in LN2 probably
doesn't self-heat much.


--

John Larkin Highland Technology, Inc
picosecond timing precision measurement

jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com

 

[Jan Panteltje]

On a sunny day (Fri, 23 Jan 2015 14:35:30 -0800) it happened John Larkin
<jlarkin@highlandtechnology.com> wrote in
<j2j5calp70pu1pk7trqdg2kq74ffhbj2qf@4ax.com>:

On Fri, 23 Jan 2015 16:45:23 GMT, Jan Panteltje <panteltje@yahoo.com
wrote:

On a sunny day (Fri, 23 Jan 2015 07:54:21 -0800 (PST)) it happened George
Herold <gherold@teachspin.com> wrote in
9263cc76-7046-4a61-8a13-280fc4c0d917@googlegroups.com>:

OK first the numbers are almost too good to believe, but here they are.
(I only tested 4 FET's I had to solder wires on to go down into the dewar. The first one I had clip leads, but the dang thick
plastic on the leads froze up, and I was afraid I'd have to break it to get it back out of the dewar neck.)

V V
Room temp. 77K
0.5500 1.0154
0.5504 1.0156
0.5508 1.0157
0.5509 1.0159

What current did you use? Any self-heating?

Cryo-diodes are usually run at 10 uA, but a big mosfet in LN2 probably
doesn't self-heat much.

1.0159
^
Well he is specifying at 10^-4 say 100 ppm....


10uA would probably be OK though, 10uW * Rtj


Math OK?

 

[Jan Panteltje]

On a sunny day (Fri, 23 Jan 2015 12:36:54 -0800 (PST)) it happened George
Herold <gherold@teachspin.com> wrote in
<c4eb58c6-721f-4bc5-84dd-55d3e3b32816@googlegroups.com>:

On Friday, January 23, 2015 at 11:45:16 AM UTC-5, Jan Panteltje wrote:
On a sunny day (Fri, 23 Jan 2015 07:54:21 -0800 (PST)) it happened George
Herold <gherold@teachspin.com> wrote in
9263cc76-7046-4a61-8a13-280fc4c0d917@googlegroups.com>:

OK first the numbers are almost too good to believe, but here they are.
(I only tested 4 FET's I had to solder wires on to go down into the dewar. The first one I had clip leads, but the dang
thick
plastic on the leads froze up, and I was afraid I'd have to break it to get it back out of the dewar neck.)

V V
Room temp. 77K
0.5500 1.0154
0.5504 1.0156
0.5508 1.0157
0.5509 1.0159

What current did you use? Any self-heating?

This was with my fluke DMM (I think ~ 1mA)

George H.

OK, thanks.

 

[Chris Jones]

On 23/01/2015 23:31, Alain Coste wrote:

Thank you John for the interesting information. The use of mosfets in linear
mode is not very common, and it's more difficult to find data than for
switch mode.

It often makes sense to use more fets, spread out over the heatsink
surface, especially if the baseplate part of the heat sink is thin,
namely has high thermal spreading resistance.

This uses copper heat spreaders to transfer the heat into the aluminum
sink.

https://dl.dropboxusercontent.com/u/53724080/Thermal/Amp.jpg


Now I see what to use _more_ fets means...
For my electronic load I could have used more transistors, but this
increases the number of current sense resistors and operational amplifiers
to control them. For the power I wanted (400 .. 420W), I thought that two
mosfets was a good compromise.
[...]

I am building an electronic load using a single IXTN60N50L2. It uses an
unconventional control scheme which in my case was easier to do with a
single large mosfet than with many small ones. I am cooling the MOSFET
with liquid. I was very surprised that nobody seems to sell water blocks
already drilled for SOT-227 packages. I expect that if I drilled the
required mounting holes on any of the widely available CPU or GPU water
blocks, the drill would hit a water channel and it would leak. Therefore
I expect I'll have to make my own water block.

Chris

 

[John Larkin]

On Sat, 24 Jan 2015 23:57:24 +1100, Chris Jones
<lugnut808@spam.yahoo.com> wrote:

On 23/01/2015 23:31, Alain Coste wrote:
Thank you John for the interesting information. The use of mosfets in linear
mode is not very common, and it's more difficult to find data than for
switch mode.

It often makes sense to use more fets, spread out over the heatsink
surface, especially if the baseplate part of the heat sink is thin,
namely has high thermal spreading resistance.

This uses copper heat spreaders to transfer the heat into the aluminum
sink.

https://dl.dropboxusercontent.com/u/53724080/Thermal/Amp.jpg


Now I see what to use _more_ fets means...
For my electronic load I could have used more transistors, but this
increases the number of current sense resistors and operational amplifiers
to control them. For the power I wanted (400 .. 420W), I thought that two
mosfets was a good compromise.
[...]


I am building an electronic load using a single IXTN60N50L2. It uses an
unconventional control scheme which in my case was easier to do with a
single large mosfet than with many small ones. I am cooling the MOSFET
with liquid. I was very surprised that nobody seems to sell water blocks
already drilled for SOT-227 packages. I expect that if I drilled the
required mounting holes on any of the widely available CPU or GPU water
blocks, the drill would hit a water channel and it would leak. Therefore
I expect I'll have to make my own water block.

Chris

McMaster sells cold plates, cheap compared to most others. No problem
drilling this kind.

https://dl.dropboxusercontent.com/u/53724080/Thermal/T222_Coldplate.jpg

https://dl.dropboxusercontent.com/u/53724080/Thermal/T222_Water_Cooled_Pulser.jpg

https://dl.dropboxusercontent.com/u/53724080/Thermal/T222_Copper.jpg


--

John Larkin Highland Technology, Inc
picosecond timing laser drivers and controllers

jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com

 

[John Larkin]

On Sat, 24 Jan 2015 10:14:05 GMT, Jan Panteltje <panteltje@yahoo.com>
wrote:

On a sunny day (Fri, 23 Jan 2015 14:35:30 -0800) it happened John Larkin
jlarkin@highlandtechnology.com> wrote in
j2j5calp70pu1pk7trqdg2kq74ffhbj2qf@4ax.com>:

On Fri, 23 Jan 2015 16:45:23 GMT, Jan Panteltje <panteltje@yahoo.com
wrote:

On a sunny day (Fri, 23 Jan 2015 07:54:21 -0800 (PST)) it happened George
Herold <gherold@teachspin.com> wrote in
9263cc76-7046-4a61-8a13-280fc4c0d917@googlegroups.com>:

OK first the numbers are almost too good to believe, but here they are.
(I only tested 4 FET's I had to solder wires on to go down into the dewar. The first one I had clip leads, but the dang thick
plastic on the leads froze up, and I was afraid I'd have to break it to get it back out of the dewar neck.)

V V
Room temp. 77K
0.5500 1.0154
0.5504 1.0156
0.5508 1.0157
0.5509 1.0159

What current did you use? Any self-heating?

Cryo-diodes are usually run at 10 uA, but a big mosfet in LN2 probably
doesn't self-heat much.

1.0159
^
Well he is specifying at 10^-4 say 100 ppm....


10uA would probably be OK though, 10uW * Rtj


Math OK?

At LN2 temp, the tempco is probably around 2 mV/K, so he's seeing
really good repeatability.

PN silicon diodes act like diodes down to about 20K. Below that the
voltage drop goes way up and they get more resistive, but less
predictable. Some companies, like Lakeshore, test diodes down to
liquid helium temps and sell parts with dependable curves.

Thermal conductivities go to hell at those temps, hence the 10 uA
convention.

A big mosfet in boiling LN2 is going to be very well temperature
pinned, at least short-term, as long as the LN2 stays LN2.





--

John Larkin Highland Technology, Inc
picosecond timing laser drivers and controllers

jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com

 

[Jan Panteltje]

On a sunny day (Sat, 24 Jan 2015 09:40:16 -0800) it happened John Larkin
<jlarkin@highlandtechnology.com> wrote in
<jnl7cadm33hf00saar3grl0lku1sjc8otd@4ax.com>:

McMaster sells cold plates, cheap compared to most others. No problem
drilling this kind.

https://dl.dropboxusercontent.com/u/53724080/Thermal/T222_Coldplate.jpg

https://dl.dropboxusercontent.com/u/53724080/Thermal/T222_Water_Cooled_Pulser.jpg

https://dl.dropboxusercontent.com/u/53724080/Thermal/T222_Copper.jpg

Nice!
Is that the temp sensor in the back of thr last jpg with the black wires?

I am using a BJT as temp sensor here:
http://panteltje.com/panteltje/tri_pic/tritium_decay_experiment_black_box_electronics_top_view_IMG_3873.GIF

How about this for cooling?
http://panteltje.com/pub/tri_pic_with_the_icecream-and_pizzas_IMG_3494.JPG

 

[Chris Jones]

On 25/01/2015 04:40, John Larkin wrote:

On Sat, 24 Jan 2015 23:57:24 +1100, Chris Jones
lugnut808@spam.yahoo.com> wrote:

On 23/01/2015 23:31, Alain Coste wrote:
Thank you John for the interesting information. The use of mosfets in linear
mode is not very common, and it's more difficult to find data than for
switch mode.

It often makes sense to use more fets, spread out over the heatsink
surface, especially if the baseplate part of the heat sink is thin,
namely has high thermal spreading resistance.

This uses copper heat spreaders to transfer the heat into the aluminum
sink.

https://dl.dropboxusercontent.com/u/53724080/Thermal/Amp.jpg


Now I see what to use _more_ fets means...
For my electronic load I could have used more transistors, but this
increases the number of current sense resistors and operational amplifiers
to control them. For the power I wanted (400 .. 420W), I thought that two
mosfets was a good compromise.
[...]


I am building an electronic load using a single IXTN60N50L2. It uses an
unconventional control scheme which in my case was easier to do with a
single large mosfet than with many small ones. I am cooling the MOSFET
with liquid. I was very surprised that nobody seems to sell water blocks
already drilled for SOT-227 packages. I expect that if I drilled the
required mounting holes on any of the widely available CPU or GPU water
blocks, the drill would hit a water channel and it would leak. Therefore
I expect I'll have to make my own water block.

Chris

McMaster sells cold plates, cheap compared to most others. No problem
drilling this kind.

https://dl.dropboxusercontent.com/u/53724080/Thermal/T222_Coldplate.jpg

https://dl.dropboxusercontent.com/u/53724080/Thermal/T222_Water_Cooled_Pulser.jpg

https://dl.dropboxusercontent.com/u/53724080/Thermal/T222_Copper.jpg

Thanks. Since the "cooling" fluid that I will get arrives already at
about 70 deg C, and I want to dissipate a lot of power, I want very low
thermal resistance, so I think I will use something made from copper. I
also need it to be not much bigger than the SOT-227 package because of
the mechanical constraints. I would like to find something already made
for this package, but otherwise I might as well silver-braze some pipes
into a block of copper - either into grooves made with a ball-end mill,
or drill long holes right through the copper (yuck!), drill manifolds in
the other direction, and braze plugs into the unnecessary holes and
braze on inlet and outlet pipes, then mill the mounting surface flat. I
only need a couple of them.

If I could buy a small version of this CP25 thing with threaded inserts
in the right place for SOT-227 then I would be tempted:
http://www.amstechnologies.com/products/thermal-management/liquid-cooling/cold-plates/vacuum-brazed-flat-tube-pin-fin-cold-plates/view/extended-surface-iiTM-cp25/

Chris

 

[Lasse Langwadt Christense]

Den sřndag den 25. januar 2015 kl. 12.37.15 UTC+1 skrev Chris Jones:

On 25/01/2015 04:40, John Larkin wrote:
On Sat, 24 Jan 2015 23:57:24 +1100, Chris Jones
lugnut808@spam.yahoo.com> wrote:

On 23/01/2015 23:31, Alain Coste wrote:
Thank you John for the interesting information. The use of mosfets in linear
mode is not very common, and it's more difficult to find data than for
switch mode.

It often makes sense to use more fets, spread out over the heatsink
surface, especially if the baseplate part of the heat sink is thin,
namely has high thermal spreading resistance.

This uses copper heat spreaders to transfer the heat into the aluminum
sink.

https://dl.dropboxusercontent.com/u/53724080/Thermal/Amp.jpg


Now I see what to use _more_ fets means...
For my electronic load I could have used more transistors, but this
increases the number of current sense resistors and operational amplifiers
to control them. For the power I wanted (400 .. 420W), I thought that two
mosfets was a good compromise.
[...]


I am building an electronic load using a single IXTN60N50L2. It uses an
unconventional control scheme which in my case was easier to do with a
single large mosfet than with many small ones. I am cooling the MOSFET
with liquid. I was very surprised that nobody seems to sell water blocks
already drilled for SOT-227 packages. I expect that if I drilled the
required mounting holes on any of the widely available CPU or GPU water
blocks, the drill would hit a water channel and it would leak. Therefore
I expect I'll have to make my own water block.

Chris

McMaster sells cold plates, cheap compared to most others. No problem
drilling this kind.

https://dl.dropboxusercontent.com/u/53724080/Thermal/T222_Coldplate.jpg

https://dl.dropboxusercontent.com/u/53724080/Thermal/T222_Water_Cooled_Pulser.jpg

https://dl.dropboxusercontent.com/u/53724080/Thermal/T222_Copper.jpg



Thanks. Since the "cooling" fluid that I will get arrives already at
about 70 deg C, and I want to dissipate a lot of power, I want very low
thermal resistance, so I think I will use something made from copper. I
also need it to be not much bigger than the SOT-227 package because of
the mechanical constraints. I would like to find something already made
for this package, but otherwise I might as well silver-braze some pipes
into a block of copper - either into grooves made with a ball-end mill,
or drill long holes right through the copper (yuck!), drill manifolds in
the other direction, and braze plugs into the unnecessary holes and
braze on inlet and outlet pipes, then mill the mounting surface flat. I
only need a couple of them.

If I could buy a small version of this CP25 thing with threaded inserts
in the right place for SOT-227 then I would be tempted:
http://www.amstechnologies.com/products/thermal-management/liquid-cooling/cold-plates/vacuum-brazed-flat-tube-pin-fin-cold-plates/view/extended-surface-iiTM-cp25/

Chris

not exactly cheap but if you only need a few

http://www.customthermoelectric.com/Water_blocks.html

WBA-1.62-0.55-CU-01 drill and tap CPT-2.25-1.62-0.25-AL for the sot-227

or you could just take on of the cheap GPU coolers and reflow
solder a ~10mm plate of copper on top drilled and tapped for the sot-227

http://i01.i.aliimg.com/wsphoto/v0/2027141455_2/New-Water-cooling-Copper-Water-Block-40-x-40-x-10-mm-free-shipping.jpg

-Lasse

 

[John Larkin]

On Sun, 25 Jan 2015 11:16:34 GMT, Jan Panteltje <panteltje@yahoo.com>
wrote:

On a sunny day (Sat, 24 Jan 2015 09:40:16 -0800) it happened John Larkin
jlarkin@highlandtechnology.com> wrote in
jnl7cadm33hf00saar3grl0lku1sjc8otd@4ax.com>:

McMaster sells cold plates, cheap compared to most others. No problem
drilling this kind.

https://dl.dropboxusercontent.com/u/53724080/Thermal/T222_Coldplate.jpg

https://dl.dropboxusercontent.com/u/53724080/Thermal/T222_Water_Cooled_Pulser.jpg

https://dl.dropboxusercontent.com/u/53724080/Thermal/T222_Copper.jpg

Nice!
Is that the temp sensor in the back of thr last jpg with the black wires?

Yup, it's a snap switch type. Shuts things off if we lose cooling.

I am using a BJT as temp sensor here:
http://panteltje.com/panteltje/tri_pic/tritium_decay_experiment_black_box_electronics_top_view_IMG_3873.GIF

Well, that's one breadboarding style.

https://dl.dropboxusercontent.com/u/53724080/Protos/D200_BB_4.JPG

LM35s are nice temp sensors. Or thinfilm RTDs.

https://dl.dropboxusercontent.com/u/53724080/Thermal/RTD_outside.jpg


--

John Larkin Highland Technology, Inc
picosecond timing laser drivers and controllers

jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com

 

[Jan Panteltje]

On a sunny day (Sun, 25 Jan 2015 08:00:32 -0800) it happened John Larkin
<jlarkin@highlandtechnology.com> wrote in
<674acah94ih10mmk3ovtq884tf6p3cdb79@4ax.com>:

On Sun, 25 Jan 2015 11:16:34 GMT, Jan Panteltje <panteltje@yahoo.com
wrote:

On a sunny day (Sat, 24 Jan 2015 09:40:16 -0800) it happened John Larkin
jlarkin@highlandtechnology.com> wrote in
jnl7cadm33hf00saar3grl0lku1sjc8otd@4ax.com>:

McMaster sells cold plates, cheap compared to most others. No problem
drilling this kind.

https://dl.dropboxusercontent.com/u/53724080/Thermal/T222_Coldplate.jpg

https://dl.dropboxusercontent.com/u/53724080/Thermal/T222_Water_Cooled_Pulser.jpg

https://dl.dropboxusercontent.com/u/53724080/Thermal/T222_Copper.jpg

Nice!
Is that the temp sensor in the back of thr last jpg with the black wires?

Yup, it's a snap switch type. Shuts things off if we lose cooling.


I am using a BJT as temp sensor here:
http://panteltje.com/panteltje/tri_pic/tritium_decay_experiment_black_box_electronics_top_view_IMG_3873.GIF

Well, that's one breadboarding style.

Breadboard now been working 24/7 for > 2 years, say 2.5 years.
temp control has been within a half degree C or so all the time.
As _relative_ sensors these transistor junctions are great.


>https://dl.dropboxusercontent.com/u/53724080/Protos/D200_BB_4.JPG

It looks so expensive it scares me to solder on it...

LM35s are nice temp sensors. Or thinfilm RTDs.

https://dl.dropboxusercontent.com/u/53724080/Thermal/RTD_outside.jpg

Yes I have some, or was it LM135 or LM335 is use cold side sensor on my thermocouple amplifier
http://panteltje.com/pub/thermocouple_interface_with_sunshade_IMG_3394.JPG
it is next to the trimpot, I used the adjustment lead of the LM.
In action here with the cryo-cooler:
http://panteltje.com/pub/cryo/

 

[John Larkin]

On Sun, 25 Jan 2015 22:37:06 +1100, Chris Jones
<lugnut808@spam.yahoo.com> wrote:

On 25/01/2015 04:40, John Larkin wrote:
On Sat, 24 Jan 2015 23:57:24 +1100, Chris Jones
lugnut808@spam.yahoo.com> wrote:

On 23/01/2015 23:31, Alain Coste wrote:
Thank you John for the interesting information. The use of mosfets in linear
mode is not very common, and it's more difficult to find data than for
switch mode.

It often makes sense to use more fets, spread out over the heatsink
surface, especially if the baseplate part of the heat sink is thin,
namely has high thermal spreading resistance.

This uses copper heat spreaders to transfer the heat into the aluminum
sink.

https://dl.dropboxusercontent.com/u/53724080/Thermal/Amp.jpg


Now I see what to use _more_ fets means...
For my electronic load I could have used more transistors, but this
increases the number of current sense resistors and operational amplifiers
to control them. For the power I wanted (400 .. 420W), I thought that two
mosfets was a good compromise.
[...]


I am building an electronic load using a single IXTN60N50L2. It uses an
unconventional control scheme which in my case was easier to do with a
single large mosfet than with many small ones. I am cooling the MOSFET
with liquid. I was very surprised that nobody seems to sell water blocks
already drilled for SOT-227 packages. I expect that if I drilled the
required mounting holes on any of the widely available CPU or GPU water
blocks, the drill would hit a water channel and it would leak. Therefore
I expect I'll have to make my own water block.

Chris

McMaster sells cold plates, cheap compared to most others. No problem
drilling this kind.

https://dl.dropboxusercontent.com/u/53724080/Thermal/T222_Coldplate.jpg

https://dl.dropboxusercontent.com/u/53724080/Thermal/T222_Water_Cooled_Pulser.jpg

https://dl.dropboxusercontent.com/u/53724080/Thermal/T222_Copper.jpg



Thanks. Since the "cooling" fluid that I will get arrives already at
about 70 deg C, and I want to dissipate a lot of power, I want very low
thermal resistance, so I think I will use something made from copper. I
also need it to be not much bigger than the SOT-227 package because of
the mechanical constraints. I would like to find something already made
for this package, but otherwise I might as well silver-braze some pipes
into a block of copper - either into grooves made with a ball-end mill,
or drill long holes right through the copper (yuck!), drill manifolds in
the other direction, and braze plugs into the unnecessary holes and
braze on inlet and outlet pipes, then mill the mounting surface flat. I
only need a couple of them.

If I could buy a small version of this CP25 thing with threaded inserts
in the right place for SOT-227 then I would be tempted:
http://www.amstechnologies.com/products/thermal-management/liquid-cooling/cold-plates/vacuum-brazed-flat-tube-pin-fin-cold-plates/view/extended-surface-iiTM-cp25/

Chris

You could get a small copper block and solder a zigzag copper tube to
it. Mill some channels first maybe, like the one in my pic. Low hassle
level, except that copper is a bitch to machine. We use Alloy 110, not
quite as gummy as soft copper. Alloys generally have much reduced
electrical and thermal conductivity.

One GPM has a net theta of 0.0037 K/W, so you don't need much water
flow to cool a few hundred watts of mosfet.

Water has a very high specific heat, which is why long hot showers are
expensive. But worth it.


--

John Larkin Highland Technology, Inc
picosecond timing laser drivers and controllers

jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com