Preface: I have a need for a 250-500 VDC input, 48-65 VDC output DC-DC for converter. Right now, it is in the "building one section at a time" stage of power board. I inquired here previously, being rather vague as I'm working on a custom design and I don't want to disclose too much before I've figured it out. Someone here suggested that I try the LTC1922 phase shifted full bridge controller chip. Using the reference design in the datasheet, outlined basically in the first pic, I was able to actually get it working. I have an isolated FET drive I use on any serious H-bridge I build. I very confident in this drive scheme. Right now I have a working board that is mostly identical to the reference design, except I use +12 V, -5 V isolated gate drive voltages for all four quadrants, through a NPN/PNP buffer for each. A TLP250 drives each quadrant from the LTC1922 chip. It works great.

HOWEVER!!: The reference design is powered from primary voltage and the secondary is isolated. I want to opposite. I want the 48 V side to be the logic power, but the 450 V side to be the isolated primary. This means ground references have to swap and that means that the ZVS sensing feedback is on the isolated primary side. I need to isolate the SBUS, ADLY, and PDLY and transfer that info across the isolation barrier. The LTC1922 senses the bus voltage at SBUS and the voltage at the S-D junction of each half bridge with ADLY and PDLY illustrated in the 2nd picture.

The SBUS expects to see around 1.5 V when the primary bus voltage is nominal. And then determines when to switch based on the voltage at each half bridge. Could it be as simple as pic 3, with optocouplers like I've shown?

 

So you want te LTC1922 powered from the secondary side?

 

So you want te LTC1922 powered from the secondary side?

Yes, this will all ultimately end up as a high voltage solar charge controller. I want to power the unit from the 48 volt battery and keep the high voltage solar galvanically isolated. There are signals on the primary side that need to translate across the isolation barrier.

 

normal practice is to have the gate drive transformers safety rated and driving the HV side devices from the LV side - this is what we do in our designs - it is quite common in industry.

 

Yes, this will all ultimately end up as a high voltage solar charge controller. I want to power the unit from the 48 volt battery and keep the high voltage solar galvanically isolated. There are signals on the primary side that need to translate across the isolation barrier.

What happens when the battery is empty?

 

What happens when the battery is empty?

At some point, UVLO kicks in and everything shuts down.

 

normal practice is to have the gate drive transformers safety rated and driving the HV side devices from the LV side - this is what we do in our designs - it is quite common in industry.

That is my intention, except I'm using optical isolation instead of transformers. The quandary that I have is that the LTC1922 driver chip uses 3 sensing inputs to measure the FET bus voltage and the voltage at each half bridge junction in order to adaptively achieve ZVS. I need to find a way to optically get that info across the isolation. I think my idea will work but I would love to have others more knowledgeable than myself ponder it as well.

 

That is my intention, except I'm using optical isolation instead of transformers. The quandary that I have is that the LTC1922 driver chip uses 3 sensing inputs to measure the FET bus voltage and the voltage at each half bridge junction in order to adaptively achieve ZVS. I need to find a way to optically get that info across the isolation. I think my idea will work but I would love to have others more knowledgeable than myself ponder it as well.

In that case it makes more sense to have all the pri side control on the pri side - and just a good opto or two for feedback and protection from the LV side.