diyAudio (
-   Class D (
-   -   Body diode bypass, elegant way! (

fredos 1st December 2010 09:51 PM

Body diode bypass, elegant way!
1 Attachment(s)
Here an elegant way to bypass body diode. Just put a tap 1-2 turn before the mosfet node to bypass the Vf of the body diode, ie to get more than the conduction of the internal body diode. In that way external diode conduct way before the slow body diode and all recovery current flow trought the external diode! I have made a prototype with slow APT5010LVR mosfet and IXYS DEE30-12A at +/-250V. Have only done test with 50% duty cycle, but with load swapped to V+ and V- to see worst case load. I got about 15A flowing in the external diode and less than 0.8A in the body diode. I was using a 15 Ohms 1000W load durring test.

lumanauw 2nd December 2010 01:19 AM

Smart thinking, Fredos :D What will happen if the load is lower ohm, like 2 or 4 ohm?

fredos 2nd December 2010 01:34 AM

Should not be a problem...1 Step at the time! I only have 15 Ohms high power resitor to take my time to make measurement.. But even with my 15 Ohms, it's about 3600W feed to them when swapped to V+ or V-! I do test for about 2-3 minutes before I smell my resistor! Tomorow I will drop supply to +/-100V to do test with 4-8 ohms load. My output are rated at 50A, but I will have to limit current to 35A due to IXYS diode. That's just a proto to validate my assumption..

lumanauw 2nd December 2010 03:12 AM

In your output coil, it is 1 or 2 turns from how many turns total?

fredos 2nd December 2010 05:09 AM

Total in my case 27 turn, 1 turn to external diode. I wrote 1-2 turn for anybody that not using same material as me, to try and test! at 250V, I got about 9V difference more that is (lot!) enought to force recovery energy to external diode, at expense of some heating in that turn, so I use 10 gauge litz wire for this one and 12 gauge for the remaining 26 turn. External diode heat a bit, but nothing to worry! So depend on supply voltage, 1, 2 or maybe 3 turn more will be required. Overall, in my case, I got better circuit effiency than any other combination at this voltage.

Workhorse 2nd December 2010 06:23 AM

IXYS is also offering some fast IGBTs now which can switch pretty fast:D

fredos 2nd December 2010 06:41 AM

They have a good turn off spec? Parts number? I have to run my output stage at at least 250Khz, so I need fast device, but as usual, they have better body diode than any mosfet at 500V DS.. By the way, I made a mistake in my first post, I was running at +/-235V..

Workhorse 2nd December 2010 06:47 AM

IXGR60N60C2C1 is tricky part;)

BTW your approach still makes the body diodes to conduct the freewheeling current and did you checked the turn off behavior of your external diodes.

phase_accurate 2nd December 2010 08:05 AM

Fig 5 & 6 ;-) :

espacenet — Originaldokument



darkfenriz 2nd December 2010 05:18 PM

Hi Fredos

I have some doubts about that circuit, let's go back to basics.

Normally during one PWM cycle mosfets force full rail to rail voltage swing, what you have done is you've added one (or two) extra windings in series so that you've created a boosting autotransformer. This way the midpoint of diodes is 'trying' to get the voltage swing above rails, but the diodes clamp it back into rails. Smells like short circuit to me, sorry for that. I guess you may expect some extra current in both the mosfets and diodes during the whole cycle, not only in the dead time body diode conduction period.
This way you might have increased conduction and switching losses a lot.
Keep extinguisher handy if I am right.


All times are GMT. The time now is 01:38 PM.

vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2017 DragonByte Technologies Ltd.
Copyright 1999-2017 diyAudio

Content Relevant URLs by vBSEO 3.3.2