I am not 100% sure where to put this, but here it goes. I am tasked with designing a DCM flyback converter to be used as an energy recovery converter for an RCD turn-off snubber used in a high-side 120V 400A single quadrant buck chopper. Basically, the flyback is operated in DCM so as to emulate a resistor. The converter maintains the voltage on a large bulk capacitor which absorbs the turn off snubbing energy of the main buck transistor that chops 400A, while the output of the flyback is hard-wired to the main DC bus.


QUESTION: My question is the control method. The flyback operates in DCM. Since the converter controls its own input voltage, I found that peak current mode control is unstable (ie a RHP is present in the plant xfer function)-unless an external stabilizing ramp is added. I found that voltage mode DCM is inherently stable (ie simple first order pole-type response if small L is neglected in DCM). My questions are a follows:

1. Do you think it would be better to try and use peak current mode control with a stabilizing ramp?

2. I am thinking it would probably be a better idea to use voltage mode control....what is your opinion?

3. Maybe it would be simpler to run the converter in open loop? After all, I would be using an ordinary resistor with a huge heatsink if I hadn't decided to take the active snubber approach.

Your opinions would be great. Sorry for the lengthy post! Thanks for all the help these forums are awesome.

Based on what I read... AVERAGE current mode may work best for this application...
Would need a bit more info...

A schematic with power levels would be useful to answer the question.

Also, have you considered processing the power in smaller pieces with parallel buck converters? The overall size will be smaller, and perhaps the big snubber can be avoided with tighter layout.

I'll post a schematic, but just need to figure out where. I wonder if I can just copy and paste it in a comment or something. In any case, I can give a little more detail without a schematic.

My application is a series wound DC motor drive. I am building a buck chopper capable of chopping into 400 amps at a DC bus of 120V, and a switching freq of 8kHz. I am working with laminated bus-bar, so the module approach seems attractive. I am taking an IGBT approach due to the lack of high current MOSFET buck modules. All the high current low-voltage MOSFET stuff is obsolete. I assume this is because IGBTs are taking over in the high current areas. The module(s) I am working with are the APTGT600SK60G TrenchStop IGBT high-side buck modules. The modules are small, low profile, and easy to work with.

In my mind, at this voltage, the ideal approach for the application would be to have a Schottky as the FWD, and a MOSFET or IGBT as the switch. But, nobody makes a module like this, which is a huge bummer. I am not an expert on making chips, but I would think this would be a great product to sell. Oh, well.

As for parallel buck converters...I have thought of that at one point, but wasn't sure. The module I have (APTGT600SK60G) uses a silicon free-wheeling diode... The module datasheet says "Easy Paralleling due to positive TC of Vce,sat," but I believe they are just talking about the IGBT, which is a TrenchStop IGBT. I have also thought of matching the IGBT in the module with a single IGBT, and then placing that IGBT in parallel with the one in the module. As for the diode, not sure. If it is possible to simply parallel the buck modules, that would be great. The power of the drive would be considerably higher, and much more robust.

Ok, now for the snubber idea. I want to turn off snub the IGBT with an RCD load-line shaping turn off snubber. I have in my possession SCM205K601H2 IGBT snubber capacitor modules. It includes the diode and capacitor, which is 2uF. At first thought, I could use a resistor and make the ordinary RCD turn off snubber and be done with it. Well, at 8kHz, 2uF of turn off capacitance, and only 100nH of bus stray inductance, the resistor would have to dissipate about 350W. To me this seemed crazy....

But then I thought that there was a reason why the module only consists of only a diode and capacitor. I got the idea to parallel 7 or 8 50W flyback modules operating in DCM to return the energy from the snubber back into the DC bus. There are a few reasons for my madness: one is that I think that I know DCM flybacks better than any other isolated converter (thats just me), the second is modularity and redundancy. Having one regenerative converter would be less reliable than 8 (unless all 8 blow up I guess). Another reason is that in DCM, the flyback input is a resistor, which is what I would be using anyways. Finally, I think that all 8 flybacks would be lighter than a huge resistor.

Will post a schematic when I figure out how. Thanks for the replies. If its too much of a headache, then I totally understand! I guess I just love giving myself headaches, as long as power devices and snubbers are involved.

Ok. Ill take a look at ACMC. I know a little about it for buck converters. I believe that with ACMC, crossing over into DCM and CCM is no problem, since the average current is controlled.

Just a thought..not knowing the full specs of your app...
You can use a Bi-directional ACMC SMPS on the front-end....This way avoiding the chopper..
If the Bulk cap gets past it's threshold...you start pumping current back into the BUS...

I am not 100% following your thoughts on the Bi-directional SMPS. The only Bi-directional SMPS I know of is the dual active bridge. I have to confess that my knowledge of bi-directional converters is pretty much nil at the moment. I think the DCM flyback would automatically pump current back into the DC bus since amp-turns are always conserved in the flyback "transformer." It is really a coupled inductor, so it will always do whatever it has to do in order to dump that energy stored in the primary (regardless of turns ratio).

If you provide a link to a schematic, it would make it much easier for all of us to truly understand your task.
Regarding your basic question of Flyback control, monitoring switch current is helpful to prevent catastrophic faults and adding a large ramp helps reduce noise susceptibility and only affects gain a small amount. Otherwise, knowing operation is DCM, a very stable method is to use a fixed Ton drive and variable Toff control per a voltage/current loop. Very simple to implement with any timer element such as a "one-shot", or 555, or a CMOS Schmitt invertor gate with a couple of diodes.
Regarding snubbing, with such large currents, it is common to consider dissipation-less snubbers which consist of a small inductor, a capacitor, and two rectifiers.

I have been trying to see if it is possible to neglect sensing switch current for current limiting because the flyback converter will never enter CCM- the secondary of the flyback is hard-wired to the DC bus (a 120V battery bank). In other words, it will never be turned on into a momentary short circuit which other ordinary converters undergo when the output filter cap is initially uncharged. I guess it is possible it could go into CCM for some other reason, causing switch current to increase. I am failing to see why the flyback in this case would need a soft start function or even a current limiting function. Of course, having one would make everything more bullet-proof.

I will look into the constant on-time control. Thanks for the suggestion.

Ahh.... I see. The dissipation-less snubber...I wonder if that would be less headache? Probably so, but isn't this approach cooler? I guess designing something that is bullet-proof is better/safer than putting out something that is cool. Again, Ill have to look into that more, as this seems to be the approach I understand best right now.