High side MOSFET drivers cannot work for Full-Bridge?

[funberry]

high side mosfet driver

I am working on a Full Bridge offline SMPS using high voltages.

I had bought the Fairchild FAN7371 high side MOSFET drivers to drive the high side FETS.

https://www.fairchildsemi.com/ds/FA/FAN7371.pdf
I thought I had this thing worked out and was blissfuly content.

Now Fairchild just came out with a newer high voltage hi-side driver , the FAN7085, which features a "racharge FET"

https://www.fairchildsemi.com/ds/FA/FAN7085.pdf

The only documentation Fairchild provides about this novelty is the sentence:

"Built-in recharge FET to refresh bootstrap circuit is very useful for circuit
topology requiring switches on low and high side of load."

Ouch.. You mean the FAN7371 was NOT suitable for topologies with both hi-and-lo side switches?
I can't use the new FAN7085 because it only goes to 300V, and my high side is 335V DC (220VAC)

Are High-side drivers like the FAN7371 (and many others) generally NOT suitable (or not intended) for full bridge topologies?

Is it because the load is lifted off the ground by the low side switch, and the hi-side driver cannot "see" the ground properly to charge its bootstrap circuit?

Doesn't make sense, since, during the OFF cycle, the low-side MOSFET of the opposide pair of switches grounds the load ( in fact grounds the low-side node of the FAN7371), which should provide the 7371 with plenty of opportunity to conduct current to ground.
I don't get it...

Is there any workaround to use hi side drivers in full bridge topologies?



thank you
Andy

 

[FvM]

mosfet high side driver

You mean the FAN7371 was NOT suitable for topologies with both hi-and-lo side switches?

You should read the FAN7085 description and application circuits more thoroughly. Actually "both hi-and-lo side switches" doesn't mean full bridge operation. I don't see, that FAN7085 is intended or even suited for usual full bridge operation at all.

Generally, bootstrap supply of high side drivers has the restriction of not allowing 100 % duty cycle for the high side switch. This problem also can't be overcome by the FAN7085 recharge circuit, only by reduction of duty cycle.

FAN7085 in constrast is intended for single ended switches where the output isn't pulled down to negative supply in off-state by the load. In some applications, the recharge circuit can achieve this. In cases, where 100 % duty cycle respectively long on times are intended, the bootstrap circuit must be re-charged from an auxilary supply "above" VBUS, there's no other way to design a N-channel high-side switch.

 

[funberry]

fan7371 schematic

Thanks for reply
OK
I read both specs again sevaral times, to the best of my ability, since I'm not an engineer.

I'm beginning to see that the FAN7371 is OK in a full bridge, but would have a difficult time with something like a 2-switch forward.

And the FAN7085 might be suited for just such a situation, the 2-switch FWD.
Am I getting this right?

thanks

 

[FvM]

bootstrap circuit for high side driver

Yes. I think, it's also suitable for some other cases, where the load isn't able to pull-down the switch output. Howver, I understood, that your application is full bridge. So standard level-shifting highside drivers as FAN7371 or IR2xxx are just O.K.

 

[funberry]

high side mosfet driver duty cycle

I am relieved.
thanks

 

[eem2am]

mosfet high side drivers pdf

**broken link removed**

yes you are right in your concern about these DREADFULLY POORLY EXPLAINED bootstrap high side drivers........

i say to you now..............USE A PULSE TRANSFORMER........it is overall less PCB board space.

then bootstrap drivers have a minefield of pitfalls for the unweary................did you read the pages 22 to 28 referred in the above link.
The datasheets dont tell you that they DONT WORK UNLESS you add much protection circuitry with them.

some irf.com app notes detail the protection needed.

 

[FvM]

mosfets in a full-bridge

Did you notice, that the original poster didn't report any actual problems with bootstrap drivers? He was just confused about some remarks in a product advertisement.

Starting a fight against boostrap high side drivers is simply off-topic in this place.

 

[funberry]

Actually my breadboard testing has so far been under ideal conditions - drive pulses from signal generator, resistive dummy load, open loop on the PWM, etc, so if there were problems inherent in these HVIC's, I didn't give them an opportunity to surface.

So I Do want to be prepared for any possible pitfalls.
Thanks for the link to the TI app note.
The additional diodes and zeners suggested are not very much different from all the protection we add around the MOSFETs themselfes anyway. I already have an overkill assortment of diodes and TVS's protecting the MOSFETS and their gates, so the extra parts are not a concern. The space taken by a SOIC-8 and a few (SMD) rectifiers is nothing compared to the 1 cubic inch transformer they are meant to replace.

Pulse XF is precisely the scenario I'm trying to imporve on by going to HVIC.

I have a space constrained application, and need to reduce heat output by about half.
I'm already at near silicon specs with conduction losses, but there is room for improvement in the switching losses. I've wound and tested about half a dozen different pulse transformers, and one of the better performing ones is about 1 cubic inch in volume. And it still distorts my waveform in destructive ways.
The MOSFETS are large devices (putting out 1500W offline), with Cgs over 10nF.
That's a lot of current to pump into that gate.
A toroid of about 2" diameter was able to drive my MOSFET in a way I could call acceptable and not saturate.
But I don't have 2", and not 1" of space, since I am space-constrained.

And a pulse transformer ( of acceptable size ) won't drive the MOSFETS to their theoretical minimum switching losses, but a 4 Amp IC may.

I'm actually at about 96% efficiency with this SMPS, and trying to push it higher.
. Heat sink the size of a fist, triyng to reduce THAT too.
At 1500W gaining 1% in efficiency is a big deal. That's 15W of heat I don't have to remove.

 

[FvM]

I didn't comment the pulse transformer suggestion and the discussion about protection circuits, because I didn't see it related to the original question.

Basically, I appreciate the moderators comment in the quoted dutchforce.com forum. He mentioned, that pulse transformers are much more limited in terms of duty cycle. Also, it's clear, that pulse transformers for IGBTs or larger MOSFETs necessarily involve a pulse shaping and driver circuit at the secondary, so the component count and space requirement are typically considerably higher than with level-shifting high side drivers. In my opinion, the driver effort should be balanced with power components size somehow. In the higher power region, it's state of the art to have individual gate drivers with DC/DC converters and isolated control signals.

they DONT WORK UNLESS you add much protection circuitry

I agree, that the galvanic coupling to the bus voltage opens an entrance for interferences and potentially dangerous overvoltages. The respective manufacturers application notes collect a number of problems that have been observed in real applications and suggest possible counter-measures. It doesn't mean, that they must appear in any application.

If you are designing a power electronics application, it's always a good idea to put it into service with low load and reduced bus voltage. Then you are usually able to identify possible problems before they can damage the circuit.