TPS 40090 -- 100V to 52V 4 phase sync buck design

Allen,

You have a difficult road ahead of you generating a 100V to 52V 4-phase synchronous BUCK converter.

The TPS40090 controller can be used as the core of this design.  You'll need to produce a 5-15V bias voltage to run the TPS40090 controller and you'll need a driver to convert the 0-5V PWM output signals to the gate drives needed for a 100V input synchronous BUCK.

I do have a couple of questions:

1) What current level are you planning to run this supply at?

2) Why are you looking for a synchornous BUCK solution?  This application runs 50% duty cycle.  With a 0.5V Schottkey power diode the efficiency impact of a diode will be approximately 0.5%, this will be reduced by losses in the low-side FET that you select, providing very little benefit for the additional cost of the MOSFET over the Diode and the extra drive circuitry.

I will check to see if I can suggest a >100V synchronous BUCK driver 

Hi Peter,

Thank you for your response. Our input voltage will vary from 60 to 100 V mostly around 70 V.

Current is about 120 Amp so if the system works in 4 phase then each phase will have 30 Amp.

I simulated the system and based on the calculation I found out the best solution is synchronous buck converter; however, I have difficulty finding the controller for the voltage and current level.

Thank you,

Allen,

You're going to find a lot of difficulties with this application I'm afraid.  Most Multi-phase BUCK converters run from lower voltages (under 30V) and most 100V supplies using a BUCK style topology run at much lower currents.

At low line (60V) the duty cycle will be ideally 86.6% which will mean 28Arms in the high-side FET.  You'll need a >100V MOSFET with a low enough Rdson and high enough thermal dissipation to support that kind of power dissipation.  Driving that kind of FET (or likely several FETs in parallel) will require a very strong driver, TI doesn't offer such a driver currently, though National does offer 100V capable synchornous BUCK drivers.

Another issue will be current sense.  The TPS40090 controller's current sense inputs have a 4V common mode range, so you'll need to monitor the inductor current, reference it to a lower voltage (0.7V or higher recommended) and feed the current sense as a differential signal between the CSCN and CSx pins of the IC.  Since the current sharing function requires a DC value of the current, this can't be done via a current sense transformer unless the converter will be operated in Discontinuous Conduction Mode (DCM) which a synchornous BUCK doesn't support.

Hi Allenm

You can design auxilary power supply which will convert the 100vdc input in to 12V dc to power up the TPS controller o,nce power converter starts you can have power f

rom other auxilary winding and you can disconnect the  power source, else you can continue using the auxillart power suorce but be carfull with grounding connection.

Regards

Piyush

Piyush,

Thank you for the comments.

Yes, this type of application would likely want a low current 100V to 6-12V DC/DC buck converter to generate the bias voltages to run the controller and the high-voltage synchronous FET drivers.  Switching 30Arms current 100V MOSFETs will require a fair amount of current, likely several hundred milliamps, if this is converted to gate drive voltages via a linear supply, it would consume tens of watts of power.

It should be possible to build a 100V to 12V @ 1A switching converter with at least 70% efficieny, delivering upto 12W of driver power with only 17W of input power.

However, that still doesn't solve the major issues:

1) Locating 100V MOSFETs capable of efficiently carrying 28Arms current

2) Locating a 100V MOSFET H-Bridge Driver capable of  efficiently switching the above MOSFET

3) Converting the phase current information from the 50-70V output to a low voltage that the TPS40090 controller can use (I suspect that this will require a Hall Effect based current sensor)

While solutions are available for item 3, 2 and 1 will be more challenging.