I'll try to replace the MOSFET in my circuit with a pure Gm, see how that works out.
BTW I may appear to split hairs here but in my circuit it is important. The used device has a minuscule SOA so the actual dissipation in the die is critical. So there! ;-)
Jan
BTW I may appear to split hairs here but in my circuit it is important. The used device has a minuscule SOA so the actual dissipation in the die is critical. So there! ;-)
Jan
Does your model include the parasitic bipolar transistor?
View attachment mosfet model 1.pdf
[taken from: Fairchild AN-9010]
-RNM
View attachment mosfet model 1.pdf
[taken from: Fairchild AN-9010]
-RNM
Last edited:
early study of the C characteristics formed with silicon dioxide dielectrics;
http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.877.1955&rep=rep1&type=pdf
What ever values you use in your mosfet models are only good for the conditions under which the C's were measured; dc bias, freq. etc. Your 'in-use' values might be different.
-RNM
http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.877.1955&rep=rep1&type=pdf
What ever values you use in your mosfet models are only good for the conditions under which the C's were measured; dc bias, freq. etc. Your 'in-use' values might be different.
-RNM
Last edited:
You may wish to consider a liquid cooled heatsink such as the ones in overclocked-CPU, high end gaming PCs. Or even better, take advantage of the fact that the SOA is tiny, and use Peltier thermoelectric coolers to keep the case temperature below 7 ºC. Mount a DS18B20 temperature sensor IC, interface to an 8-bit microcontroller, and display heatsink temperature on the front panel. If heatsink temperature exceeds an upper limit: auto shut-off.
For high def audio.... recommendations ---
https://www.grammypro.com/sites/def...ions_for_recorded_music_projects_10_10_17.pdf
-RNM
https://www.grammypro.com/sites/def...ions_for_recorded_music_projects_10_10_17.pdf
-RNM
I think that in the VDMOS model the capacitances are dynamically adjusted depending on Vds and Ids. Just like in real live (see the AN you linked to). Keantoken would know probably.
It's a mess ;-)
Jan
Ian Hegglun developed a thermal emulator for me which is just a bunch of parts that provide the die temperature in real time, taking into account heat sink temperature and dissipation. So the info is available, if I want to monitor it and take measures.
Peltier will increase the in-case dissipation so I am not a fan of that. I'll look into those CPU coolers, good idea.
Jan
Jan, may we know what is the useful application of as you see it from this sought for "knowledge" so we can better understand and maybe help with your inquiry.
If you are concerned with some kind of thermal modulation then, depending on the frequencies running through the device, even all the extreme cooling in the world may not help, there's still a thermal impedance that will always be in the way.
If you are concerned with some kind of thermal modulation then, depending on the frequencies running through the device, even all the extreme cooling in the world may not help, there's still a thermal impedance that will always be in the way.
Last edited:
Its an app for a high voltage gain stage for an ESL direct drive amp. The number of useful devices is very small, and I need to wring every last bit of performance out of it. So I am concerned about minimizing disspation and maximizing heatsinking. An additional issue is that the heat sink capacity (electrical) to the environment is another capacitor that has to be swung 4000V pk-pk, so it's not a good idea to use huge heatsinks.
It's not a matter of linearity, that's taken care of.
So it's really a very interesting project with unusual compromises to be made. I love it ;-)
Jan
It's not a matter of linearity, that's taken care of.
So it's really a very interesting project with unusual compromises to be made. I love it ;-)
Jan
Jan, if you open my file here:
Better power MOSFET models in LTSpice
Go to control panel->save defaults->save subcircuit node and device currents. Now you can probe subcircuit nodes/currents as long as you know their name.
Add this to the schematic:
.save dialogbox
Upon simulation it will open a list of circuit and subcircuit node currents that can be probed. The node names should be interpretable.
Better power MOSFET models in LTSpice
Go to control panel->save defaults->save subcircuit node and device currents. Now you can probe subcircuit nodes/currents as long as you know their name.
Add this to the schematic:
.save dialogbox
Upon simulation it will open a list of circuit and subcircuit node currents that can be probed. The node names should be interpretable.
Actually you can omit the .save command, just right click the waveform and go to "add trace", and you can select from the list circuit and subcircuit waveforms.
Ideal case for water cooling.
You'll be amazed how little water flow you would need.
Just do the maths.
P.
My first thought was 'water and 4000v, what could possibly go wrong', but actually the water cooling kits for PCs seem to be good in that regard and pressure is low. I assume some have tried creating a non-conductive case (say clear perspex) for the block so you can avoid having to put mica or similar between the device and the heatsink. Only a few 10ths of C/W but in a case where every bit counts would be an interesting trade off.
And of course lapping the device and heatsink together (was that your recommendation Patrick or JNeutrons, can't remember).
And of course lapping the device and heatsink together (was that your recommendation Patrick or JNeutrons, can't remember).
There are materials which are thermally conductive but not electronically, also at4kV.
Depending on how exotic you want to go, BeO, AlN, even the common Al2O3, .......
Patrick
Depending on how exotic you want to go, BeO, AlN, even the common Al2O3, .......
Patrick
Hi Jan,
Are you designing amp for Discopete? 😉
You may search aes e-library, you may also consider high-voltage horizontal deflection tubes US Pat 4324950 and the Sanders Amplifier
High voltage tube may be cheaper than hi voltage MOSFET plus heatsink
Are you designing amp for Discopete? 😉
You may search aes e-library, you may also consider high-voltage horizontal deflection tubes US Pat 4324950 and the Sanders Amplifier
High voltage tube may be cheaper than hi voltage MOSFET plus heatsink
Last edited:
Ah, that's cool, I have to admit I never really did pay too much attention to ESL and exotics in the past, that was until after reading one of the more recent threads under SS it got me a bit itchy too.
There really aren't many parts out there for the job of 4kV direct drive, IXTL2N450 could be interesting though, comes already isolated with a ceramic tab (IXYS DCB) incorporated into the package, but at a delicate cost. 🙂
Looking at the SOA limiting curves it suggests these parts can't take HV for prolonged time even at minuscule currents so a HP filter is needed.
Digikey.com IXTL2N450
And little brother..
Digikey.com IXTF1N450
There really aren't many parts out there for the job of 4kV direct drive, IXTL2N450 could be interesting though, comes already isolated with a ceramic tab (IXYS DCB) incorporated into the package, but at a delicate cost. 🙂
Looking at the SOA limiting curves it suggests these parts can't take HV for prolonged time even at minuscule currents so a HP filter is needed.
Digikey.com IXTL2N450
And little brother..
Digikey.com IXTF1N450
Attachments
Last edited:
Just series stack the devices ... lowered voltage across each ... lower dissipation.
[You can make a push-pull stack also]
-RNM
[You can make a push-pull stack also]
-RNM
Last edited:
- Status
- Not open for further replies.