CCRT– Cross Conduction Reduction Topology

[phase_accurate]

The following shall show a means to reduce the effects of shoot-through and body diode-conduction (i.e. reverse-recovery). It will not be a cure-all, but it can give relief to these effects. It works best when the deadtime is around zero. But it is not restricted to zero deadtime. It can actually work from negative to positive deadtime. It can reduce the demands on accuracy of deadtime setting. It does not control the deadtime itself by any means - but it helps to reduce the effects of deadtime tolerances. I have to admit that adaptive deadtime-control would be superior to this but I think that this only makes sense for highly integrated solutions (there are many interesting patents on that subject BTW).

I searched intensively in order to find out if it is already patented (and stumbled over a lot of interesting things while doing so) but didn’t find anything. So it is herewith released to the public domain unless proven otherwise. Feel free to use it. If you use it commercially however, don’t forget to mention its origin! You are of course free to post your experiences with it here.

The idea actually came from some tinkering around adaptive deadtime control. I was thinking about using a small transformer in order to measure cross-conduction. I then remembered that there was an output stage topology by Brian Attwood (Used in the Peavey DECA series of amps), that uses inductors to reduce cross-conduction since he isn’t a fan of deadtime. He also proposed some autotransformer topology to reduce body-diode conduction (I.e. the output inductor was used as auto-transformer). I once tried the latter and it actually works - but it only works well with output coils that have many turns.

Since body diode conduction isn’t that much of a problem with short deadtime - I didn’t actually care for it with the proposed circuit but it seems to be intrinsically taken care of by the circuit as well to some degree.

So here we have it. It is a small autotransformer with very low inductance (<1uH). Both legs are going to the switching devices and the centre tap is connected to the output-filter coil. There are two additional fast-recovery diodes needed as well.

And now have fun !

[Workhorse]

Originally posted by phase_accurate
The following shall show a means to reduce the effects of shoot-through and body diode-conduction (i.e. reverse-recovery). It will not be a cure-all, but it can give relief to these effects. It works best when the deadtime is around zero. But it is not restricted to zero deadtime. It can actually work from negative to positive deadtime. It can reduce the demands on accuracy of deadtime setting. It does not control the deadtime itself by any means - but it helps to reduce the effects of deadtime tolerances. I have to admit that adaptive deadtime-control would be superior to this but I think that this only makes sense for highly integrated solutions (there are many interesting patents on that subject BTW).

I searched intensively in order to find out if it is already patented (and stumbled over a lot of interesting things while doing so) but didn’t find anything. So it is herewith released to the public domain unless proven otherwise. Feel free to use it. If you use it commercially however, don’t forget to mention its origin! You are of course free to post your experiences with it here.

The idea actually came from some tinkering around adaptive deadtime control. I was thinking about using a small transformer in order to measure cross-conduction. I then remembered that there was an output stage topology by Brian Attwood (Used in the Peavey DECA series of amps), that uses inductors to reduce cross-conduction since he isn’t a fan of deadtime. He also proposed some autotransformer topology to reduce body-diode conduction (I.e. the output inductor was used as auto-transformer). I once tried the latter and it actually works - but it only works well with output coils that have many turns.

Since body diode conduction isn’t that much of a problem with short deadtime - I didn’t actually care for it with the proposed circuit but it seems to be intrinsically taken care of by the circuit as well to some degree.

So here we have it. It is a small autotransformer with very low inductance (<1uH). Both legs are going to the switching devices and the centre tap is connected to the output-filter coil. There are two additional fast-recovery diodes needed as well.

And now have fun !

The same scheme is used by Crest Audio intheir LT series Class-D amps with auto transformer inductance beiing 200-300nH, and its already patented by peavey.

[Bender.ru]

Hello Charles, your idea is similar this BCA technology (imho)...am i wrong? can't see any difference

[Workhorse]

In BCA both devices conduct simuntaneously, thats not the case with Charlies scheme.

[phase_accurate]

Hi Kanwar

The topology in the Peavey patent OUTPUT INDUCTOR in an autotransformer fashion. And then there are additional magnetic snubbers (not coupled in any fashion). The original topology used a total of 6 additional components. While Crest only uses four additional ones (RC snubbers not counted).
But none of them uses the snubbers in a coupled (i.e. autotransformer) fashion (my topology uses only tree additional components BTW).

I think it is distinct enough for not being covered by the Peavey patent. There might however exist a patent that describes exactly that but I havent seen any so far. But I must admit that I have only searched class-d audio patents and not DC choppers for motion control, PSUs etc etc.

Regards

Charles

[Workhorse]

Hi Charles

Have a look at this crest schematic.

those inductors are indeed configured as auto transformers on pcb also.

[Eva]

The main drawback of this classic arrangement is that both the switches, the diodes and the inductors have to carry maximum switch current (output current plus reverse recovery peak) during most of the time because there is no quick way to de-energize the inductors. Another drawback is that hard switching is still taking place on the axiliary diodes.

[iand]

Originally posted by Eva
The main drawback of this classic arrangement is that both the switches, the diodes and the inductors have to carry maximum switch current (output current plus reverse recovery peak) during most of the time because there is no quick way to de-energize the inductors. Another drawback is that hard switching is still taking place on the axiliary diodes.

But does it keep the current out of the slow MOS body diodes? (I don't think so)

If it does then the hard switching of the auxiliary diodes isn't such a problem, there are some very fast soft recovery diodes available.

If it doesn't then the switching losses might go up compared to a normal circuit.

Ian

[phase_accurate]

But does it keep the current out of the slow MOS body diodes? (I don't think so)

It does so by not needing deadtime. Body diodes usually get into conduction during deadtime.

If it does then the hard switching of the auxiliary diodes isn't such a problem, there are some very fast soft recovery diodes available.

Exactly !

Regards

Charles

[lumanauw]

Hi, EVA,

Do you have any special trick to make body dioda Qrr (of an off-state mosfet) not shoting-through (crossconduction) to the other side mosfet when that particular mosfet is off?