Unfortunately too generic to find the topic you mean...
But maybe the effect has more to do with something like this:
- V-MOSFET thermal coefficient changes rapid if junction temperature reaches e.g. 70 degree celsius?
Complicated stuff and not helped by the usual mosfet charts showing -55, 25 and 150 charts. This paper has some nice explanations about why this is a little more complex than just carrier mobility effect. http://www.springer.com/cda/content...407478-c1.pdf?SGWID=0-0-45-1268751-p174130080
In practice I found that around 70 degrees is where vertical mosfets start behaving a bit more stable from a bias and temp runaway perspective. In other words I want to set my bias such that the sinks are around 50-60 degrees. Still risky without some sort of compensation though.
In practice I found that around 70 degrees is where vertical mosfets start behaving a bit more stable from a bias and temp runaway perspective. In other words I want to set my bias such that the sinks are around 50-60 degrees. Still risky without some sort of compensation though.
Vertical MOSFET - Electro-Thermal Instability (ETI)
An AN from Ixys:
http://www.ixys.com/Documents/AppNotes/IXAN0068.pdf
Thanks for reply. Maybe I found the explanation. As the heatsink was already very hot during longer test cycles the ETI curve may have moved to the left side (lower Vds). Also the mosfet's where not optimally thermally mounted with cheap and many times reused keratherm pads.
An AN from Ixys:
http://www.ixys.com/Documents/AppNotes/IXAN0068.pdf
Electro-Thermal Instability (ETI) can be understood as a result of positive feedback mechanism on the surface of the Power MOSFET forced into linear mode of operation:
- There is a local increase of junction temperature;
- This causes local decrease of Vgs(th) (MOSFET threshold voltage has negative temperature coefficient);
- This causes increase of local current density, Jds ~ (Vgs - V gs(th))2 ;
- Increase of local current density causes increase of local power dissipation and further increase of local temperature.
BR, Toni- There is a local increase of junction temperature;
- This causes local decrease of Vgs(th) (MOSFET threshold voltage has negative temperature coefficient);
- This causes increase of local current density, Jds ~ (Vgs - V gs(th))2 ;
- Increase of local current density causes increase of local power dissipation and further increase of local temperature.
Attachments
Last edited:
SA2015: using vertical mosfets IXTH80N20L and IXTH48P20P
... attached the schematic of the working sample.
First measurements see post # 2171 and post # 2174 above.
Note: Schematic free to use only for non commercial DIY projects.
WARNING: this circuit is using high DC voltages and is therefore dangerous. I am not responsible for any costs, damage and/or injury using these schematics and BOM. Do not use this schematics if you do not have the necessary experience and knowledge.
Have fun!
Toni
... attached the schematic of the working sample.
First measurements see post # 2171 and post # 2174 above.
Note: Schematic free to use only for non commercial DIY projects.
WARNING: this circuit is using high DC voltages and is therefore dangerous. I am not responsible for any costs, damage and/or injury using these schematics and BOM. Do not use this schematics if you do not have the necessary experience and knowledge.
Have fun!
Toni
Attachments
Last edited:
Good to see a current limiter, Toni
! This important detail too often is missing in current designs.Best regards!
... wanted to test the SA2015 also with SEMELAB power horses MG6330 and MG9410 which I have some pieces on stock.
Found no simulation models but instead another bad news for DIYers:
just found out this power horses are already discontinued. So no need for spice models. The IXYS power horses hopefully will be available some time...
BR, Toni
Found no simulation models but instead another bad news for DIYers:
just found out this power horses are already discontinued. So no need for spice models. The IXYS power horses hopefully will be available some time...
BR, Toni
Test sample using 1 pair of IXTH80N20L and IXTH48P20P up und running!
Quick THD+N tests (80k bw) with SOA circuit in situ:
- THD+N 20k@100W@8R: 0.0035%
- THD+N 20k@200W@4R: 0.0042%
Rgate 15R. No oscillations detected. No signs of instability at +/-50V supply.
More tests and tuning data will follow as soon as I have repaired my test sample (a f...y short circuit during a f...y screw driver accident during tests
Dave Zan's simulation models seems to match pretty good with the devices in reality. Bias settings in reality as expected from spice design!
Bias temperature stability similar to SA2015 IRFP240/IRFP9240.
Have fun,
Toni
So I wonder what 2 pair would be like on a shorter PCB, comfortably fitting vertically in a 5U case....
4 x mono build anyone?
SA2015: using vertical mosfets IXTH80N20L and IXTH48P20P
IXYS MOSFETS IXTH80N20L and IXTH48P20P:
Those parts are very expensive for DIY and too expensive to match yourself for an amplifier project. I have bought 10 pieces each and couldn't find one well matched pair ...
Maybe you need to buy 100 pcs to find good matched pairs.
Some prices:
BR, Toni
IXYS MOSFETS IXTH80N20L and IXTH48P20P:
So I wonder what 2 pair would be like on a shorter PCB, comfortably fitting vertically in a 5U case....
Those parts are very expensive for DIY and too expensive to match yourself for an amplifier project. I have bought 10 pieces each and couldn't find one well matched pair ...
Maybe you need to buy 100 pcs to find good matched pairs.
Some prices:
- 1 pc of IXTH80N20L ~ 12 - 13 EUR.
- 1 pc of IXTH48P20P ~ 10 - 11 EUR.
- 100 pcs of IXTH80N20L ~ 900 - 1000 EUR.
- 100 pcs of IXTH48P20P ~ 700 - 800 EUR.
BR, Toni
The small details make a project a perfect project!
Howdy all. My family and I are moving to the US and as a result, I have the two LM317-based power supplies and the astx MCU module shown in the rightmost picture in post 2169 available to be sent for free to someone who can make use of them building Toni's amps. They will need to ship out this week. Let me know if interested.
