Thanks for such an interesting app. note
You are right, I'm using the same old
principle altough the primary of the coupled inductor is connected between the diode and the output capacitors. This forces the diode tab and the IGBT tab to be at the same potential during turn on and allows the primary and the secondary windings to share a connection [like a tapped inductor]
Do you know better alternatives to control diode recovery dI/dt and reduce overall circuit dissipation over a 10% to 99% duty cycle range at the same time?
How is being solved the diode recovery problem nowadays?
[besides using SiC diodes, or allowing for hard switching with 400% current overshoot and then using big EMI filters, extremely overrated MOSFETS and big heatsinks]
I've seen lots of appnotes showing active snubbers but they allways require additional control logic and extra switching devices to be placed on the heatsink. I saw also resonant approaches but they appear to work only for limited duty cycle ranges
In the other hand, the small leakage inductance of the coupled inductor matches very well with the inherently large rise time of the IGBTs [100ns from 20V to 580V for Ic around 6A] so no voltage spike is produced at turn-off and no clamping is required
About thyristors : I'm planning a separate heatsink or some isolating spacers but I feel these are very hard to find
About iron inductors : Are you talking about standard iron laminations found on 50Hz transformers? I happen to have lots of junk iron transformers but how do I gap them?
Also, I like the old bipolar full bridge made with modern fast bipolars and its superior efficiency at 35Khz in comparison with MOSFETs or IGBTs. But MJE13009 has a RBSOA limited to 350V, so in order to get the maximum output, the circuit would require a PFC converter with 340V output and 235V AC max. input instead of the more usual >380V output and 260V AC max. input. Such a bridge circuit may provide 2KW output with 4 TO-220 bipolars and a few watts of losses
I'm considering this combination also as an amplifier PSU, but precise regulation is not mandatory for audio amplifiers, so a transformer-coupled PFC may be used instead [QSC and others already do this in their bigger amplifiers]
You are right, I'm using the same old
principle altough the primary of the coupled inductor is connected between the diode and the output capacitors. This forces the diode tab and the IGBT tab to be at the same potential during turn on and allows the primary and the secondary windings to share a connection [like a tapped inductor]Do you know better alternatives to control diode recovery dI/dt and reduce overall circuit dissipation over a 10% to 99% duty cycle range at the same time?
How is being solved the diode recovery problem nowadays?
[besides using SiC diodes, or allowing for hard switching with 400% current overshoot and then using big EMI filters, extremely overrated MOSFETS and big heatsinks]
I've seen lots of appnotes showing active snubbers but they allways require additional control logic and extra switching devices to be placed on the heatsink. I saw also resonant approaches but they appear to work only for limited duty cycle ranges
In the other hand, the small leakage inductance of the coupled inductor matches very well with the inherently large rise time of the IGBTs [100ns from 20V to 580V for Ic around 6A] so no voltage spike is produced at turn-off and no clamping is required
About thyristors : I'm planning a separate heatsink or some isolating spacers but I feel these are very hard to find
About iron inductors : Are you talking about standard iron laminations found on 50Hz transformers? I happen to have lots of junk iron transformers but how do I gap them?
Also, I like the old bipolar full bridge made with modern fast bipolars and its superior efficiency at 35Khz in comparison with MOSFETs or IGBTs. But MJE13009 has a RBSOA limited to 350V, so in order to get the maximum output, the circuit would require a PFC converter with 340V output and 235V AC max. input instead of the more usual >380V output and 260V AC max. input. Such a bridge circuit may provide 2KW output with 4 TO-220 bipolars and a few watts of losses
I'm considering this combination also as an amplifier PSU, but precise regulation is not mandatory for audio amplifiers, so a transformer-coupled PFC may be used instead [QSC and others already do this in their bigger amplifiers]
Hi,
I am not saying that your snubber idea is bad, rather the contrary. I must admit that I use hard switching with BIG mosfets, and have also tried SiC diode (it IS good). I deliberatly slow down turn on transient by proper gate resistor (hard to do with IGBTs) and get away with it. But my design constrais were a little different (no place for large inductor and free forced air cooling).
Active snubbers are complicated IMHO. Even Unitrode stopped producing control circuit that had provision for control of auxiliary mosfet. Passive lossless snubbers are good especially with IGBTs, but for mosfets and SiC diodes hard switching is acceptable.
Regarding iron cores, I thought cut C cores are still a common thing (we still have more than a ton of various cores purchased some 20 years ago). I had relatively hard time finding producer of such cores nowadays, but here is one. There might be others manufacturers. Since C core max flux density can approach 1.9T in small cores, your gap is 6 times shorter than in ferrite.
Best regards,
Jaka Racman
I am not saying that your snubber idea is bad, rather the contrary. I must admit that I use hard switching with BIG mosfets, and have also tried SiC diode (it IS good). I deliberatly slow down turn on transient by proper gate resistor (hard to do with IGBTs) and get away with it. But my design constrais were a little different (no place for large inductor and free forced air cooling).
Active snubbers are complicated IMHO. Even Unitrode stopped producing control circuit that had provision for control of auxiliary mosfet. Passive lossless snubbers are good especially with IGBTs, but for mosfets and SiC diodes hard switching is acceptable.
Regarding iron cores, I thought cut C cores are still a common thing (we still have more than a ton of various cores purchased some 20 years ago). I had relatively hard time finding producer of such cores nowadays, but here is one. There might be others manufacturers. Since C core max flux density can approach 1.9T in small cores, your gap is 6 times shorter than in ferrite.
Best regards,
Jaka Racman
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.