[SOLVED] High-side driver high power supply

[red_alert]

I'm trying to avoid an isolate power supply for the high-side switches (MOSFET full bridge) so I came up to the only remaining options: bootstrap technique or a charge pump.

The problem is that I need a powerful power supply as I'm going to use high current MOSFET drivers (TC4452 - 12Amps).

I have the possibility to use a bootstrap circuit (as I could limit the duty cycle to 95% or something) but then I need to use quite a bunch of high pulse bootstrap capacitors and a high current ultrafast diode which might become too bulky for my space requirements.

I'd prefere a charge pump circuit but I don't know how to reference its output voltage to the high-side MOSFET source (as its voltage will fluctuate between ground and DC bus). I've seen a few implementations (NE555) but they are low current only. They use a simple resistor to shift the voltage but I need something more efficient (maybe a current source?).

I really need some help here - I spent few days on google with no success.

 

[mtwieg]

The simple bootstrap capacitor is, in fact, a very simple charge pump solution. Anything more complicated is likely going to be less efficient. IMO if that's not enough, then you should jump straight to using a truly isolated gate drive supply. It's not as bad as it sounds. Wurth makes transformers with several 1:1 windings, just drive one with a 15V square wave, rectify the outputs. You don't even need output regulation. For a few buck you can have 4+ isolated supplies.

 

[red_alert]

I guess I was spending too much time avoiding this straight solution. After all, it seems to be the less bulkier one (by putting it away and only run a few cables). I could make use of SG3525 and n-FETs I have for a push-pull implementation. Thanks again for your suggestions.

 

[chinuhark]

I was thinking of that for myself actually. Tell me how the SG3525 solution works out. Dedicated isolated supplies for the gates is always better and hassle free. Especially for such large power applications.
I read somewhere about the whole bootstrap vs isolated supply debate and the article said that bootstrap is good for low power applications as it saves space and reduces cost but it's not worth trying to save on the power supply cost when high power circuits are in question.

 

[FvM]

Gate driver supply design starts with a calculation of average power consumption. 12 A peak gate current is irrelevant in this regard. Charge pump is usually no alternative to bootstrap supply due to it's limited power delivery, it's e.g. used to put up the driver quiescent current for designs with 100 percent duty cycle capability or infrequently switching intelligent high side drivers.

I agree that isolated DC/DC is a straightforward and ruggedized solution. Particularly for higher output power the additional effort will be well compensated by higher design safety.

 

[red_alert]

Gate driver supply design starts with a calculation of average power consumption. 12 A peak gate current is irrelevant in this regard.

You're right, but still the main design constraint is the need for those bulky pulse capacitors to withstand the 12A peak gate current.

Maybe I could design a separate PCB for the (power) bootstrap circuit, using a bunch of high pulse/low esr capacitors and a ultrafast high current diode.

After all, the Vcc voltage for the low-side MOSFET driver is located on a separate module too and you need a decoupling capacitor closer to the IC pins anyway (to sustain the peak gate current).

 

[FvM]

You're right, but still the main design constraint is the need for those bulky pulse capacitors to withstand the 12A peak gate current.

Regular MLCC capacitors will well serve the purpose. They have mohm ESR and no problems to handle several A rms.

 

[red_alert]

I was using only this type of capacitors for bootstrap and DC bus decoupling:

https://www.wima.com/EN/products_pulse.htm

and, more specific, the FKP1 series:

https://www.wima.com/EN/fkp1.htm

"As regards pulse loading capability, WIMA FKP 1 represents the high-end of capacitor technology." they claim and I can confirm their best results.

I have no informations about MLCC pulse capability. Could you recommend any specific MLCC capacitor for such an application?