Buck Controller Bootstrap Woes

[ACharnley]

Part is NCP81151. The boostrap is fed internally by diode to VCC, which is VBUS. Input is 0-40V.

Consider the system has been in sleep and turns on. The output is already near VBUS. An input voltage presents itself and the MCU sends a PWM to the input. The 81151 is in auto diode emulation mode. The bootstrap doesn't charge through either the lower FET or the output (which is too close to VCC) and subsequently the converter fails to start switching.

I've been going around in circles with this one.

1. ditch it and go PMOS on highside, which is more challenging to drive when factoring in gate protection.
2. create a voltage doubler and feed to the NCP81161 which iirc is rated about 17v (though current consumption is double), so Vgs will always be v=BST-load.
3. try to do the same using a buck all-in-one, though I'm curious how they resolve it or if it's a systemic issue with the NMOS highside design.
4. blip the lowside NMOS off briefly at the start of operation to allow BST to charge, and do it quickly to ensure reverse flow doesn't fry the FET. This won't help if the NCP81151 switches low side to diode emulation later on (wouldn't know).
5. Stick a detect on the lowerside FET, if the output voltage is close to VCC (MCU measures it) and this FET hasn't been turned on in a while then do the blip of (4).
6. Something else?..

 

[ronsimpson]

Consider the system has been in sleep and turns on. The output is already near VBUS.

This tells me there is no load. You need a load to pull the output down to near ground. Then the diode boot strap thing will work.

 

[ACharnley]

Which is a problem as that will be detrimental to standby efficiency. The load are very large super-capacitors, they may already be charged at or near VBUS so the difference in voltage across the bootstrap capacitor will be minimal.

 

[ronsimpson]

After some sleep, if you turn on the bottom MOSFET, you will charge up C9. Now you will have some energy to turn on the top MOSFET. Maybe it needs two strokes of 50% to get started.
What dose sleep mode look like? Are you using the enable pin?
How are you monitoring current? That is pretty key to this charger.

 

[ACharnley]

I'll have to be careful and do it with hardware control, there's a huge amount of capacitance ready to discharge through the lower FET and cause some smoke.

I've since added a detect onto pin 5 of the IC. I'm using a timeout and one-shot timer to create a single pulse if a high signal isn't detected on the lower FET. The pulse sets deadtime to the mpptPWM and sets mpptBlip low, effectively creating 0V at the PWM pin which turns the low FET on.

 

[rjenkinsgb]

An input voltage presents itself and the MCU sends a PWM to the input.

Is the input literally switching on? How fast?

If it's a reasonably fast rate of rise, a simple capacitor - resistor - diode (& zener if needed) circuit could be used to "add" the input voltage to the output & charge the bootstrap cap?