Yet More Buck Questions

[ACharnley]

I'm having to make a SEPIC working between 6-100V and I've focused on the buck for today.

I have it working and it's crawling along "nicely" thanks to the 1MHz speed.

When the next prototype goes to fab this section will be controlled by a pic's PWM, ADC and deadband peripheral. I've mimiced these in the circuit.

As you can see the feedback is set for 5.1V.

Two comparators will stick 18V into the gates (so yes, I'll not actually be working with any real current over ~12V but it won't fry either).

First question, R1 and R5 are to prevent ringing but at anything greater (i.e 10 oh)m the pwm is badly distorted. Should I just remove them and rely on the comparators push/pull current limitation?

Second question, you can see that there's some current movement through L1 which reduces with a larger inductor. At 1MHz I expected to use something around the 10uH mark. Any recommendations here? I'm happy with 60mA @ 5V but wonder if it can be bettered.

Third question, I stuck R6 in to see if the deadband it working correctly, which it is, but when the M11 (high side) fet switches there's a short burst of current through R6. It's cut off the picture here but it is 3.5A. Any idea what causes it and should I be concerned?

Simulation attached.

Thanks, A.

 

[ronsimpson]

Is this a project where you are charging a super cap to 6 volts?

 

[ACharnley]

Yup.

 

[ACharnley]

Few additional questions:

1. I need to generate 18V without using the PIC (out of pins), what popular boost ic do people frequently use? (ideally it has to be small, integrated other than LC and will run from 3.3v). I need no more than 50mA.

2. I'll drive the fet's using the cheap AZV3002 comparators. The PIC PWM has push/pull output but do the negative/ref pins for a comparator have to be held above 0v or can I just ground them (I wonder what it does when both are at 0v)?

 

[ronsimpson]

do the negative/ref pins for a comparator have to be held above 0v or can I just ground them (I wonder what it does when both are at 0v)?

A comparator compares two voltages.
If the (+) pin is larger than the (-) pin then the output is high.
If the (+) pin is smaller than the (-) pin then the output is low.
More or less; if (+) = (-) then the output is unknown.

For the comparator to work the two inputs must be within the "common mode range". This part: that range is from supply to supply. 0 to 5V Many parts will work to a slightly negative level. Most parts will not work near the upper supply. ( 5V-1.5= 3.5V)

This part has hysteresis of about 9mV. Most do not have!
>If one input is at 1V and the other starts at 0V and ramps up to 2V the output will switch at input = 1.009V. (input going up)
>If one input is at 1V and the other starts at 2V and ramps down to 0V the output will switch at input = 0.991V. (input going down)
If there was no hysteresis the two points would be very close to 1.0000000? volts.

"Input offset voltage": This is an error voltage. The part can not be made perfectly. This part typically has an error of 0.0005 volts but could be as far off as +/-0.03V.
> One input is at 1.00000 volts and the other input is varied to see what happens. Worse case the comparator might switch as high as 1.03 volts or as low as 0.97V. (+/- 30mV)

PIC output: The output likely will not reach 0V. Very close.

Why you should not connect one input to 0V:
>0V is at the very lowest the part will work at. common mode range
>The other input will need to go 9mV below ground to get the part to change output states. hysteresis Also Hysteresis has a min/typ/max range.
>The input offset range will cause the part to see 0V as anywhere from -30mV to +30mV.

Connect one input directly to the PIC pin. Connect the other pin to a supply that is 1/2 supply. This will solve all the above problems.
>For CMOS parts like the PIC use 1/2 supply. For TTL pars use a supply that is in the center of the data range.
>Example using TTL: low = 0 to 0.8V, high = 2 and above. So set the compare voltage to be about 1.5 volts.

 

[ACharnley]

Cheers, the reference being V/2 is what I suspected.

Just a suitably small fet boost ic to find now.

 

[ronsimpson]

Cheers

I see where you are getting power. What are you powering? How long do you want the caps to hold up for?

 

[ACharnley]

It's for driving N fets (high side).

Edit: messed up, thought the op-amp had a high voltage rating. On the hunt for a push-pull op-amp working up to 20v, possibly I could use TLC3704 and lower the fet drive voltage.

 

[ACharnley]

Doh it's what I don't know. Discovered the high-speed mosfet driver like https://www.microchip.com/wwwproducts/en/TC4427A