• AVR Freaks

Hot!Preventing Buck backflow using PIC16/18 comparator

Author
acharnley
Super Member
  • Total Posts : 268
  • Reward points : 0
  • Joined: 2016/05/01 06:51:28
  • Location: 0
  • Status: offline
2019/02/25 00:34:17 (permalink)
0

Preventing Buck backflow using PIC16/18 comparator

Hi,

I have a voltage mode buck as shown. 
There are two super-capacitors towards the right. One scenario is they are almost fully charged and the duty cycle begins to ramp up. The bottom FET would be on for a greater share of the time, the inductor would saturate and current (a lot) would flow to ground.

I could just not turn on the FET until a particular duty cycle, or I could attempt to detect the flow and shut it off (sounds nicer). I came up with the above but now I realise the FET will be in oscillation. Is the correct method to compare the voltage both sides of the inductor and turn it off that way?

I'm thinking a PIC16 comparator and the PWM fed into CLC.

Regards, Andrew
 

Attached Image(s)

#1

9 Replies Related Threads

    crosland
    Super Member
    • Total Posts : 1568
    • Reward points : 0
    • Joined: 2005/05/10 10:55:05
    • Location: Bucks, UK
    • Status: offline
    Re: Preventing Buck backflow using PIC16/18 comparator 2019/02/25 01:42:54 (permalink)
    +1 (1)
    acharnley
    There are two super-capacitors towards the right. One scenario is they are almost fully charged and the duty cycle begins to ramp up. The bottom FET would be on for a greater share of the time, the inductor would saturate and current (a lot) would flow to ground.

     
    OK, I will show my ignorance, but why would that even happen? The control loop should be sensing the output voltage and controlling the FETs appropriately. Perhaps you should be using a device designed to do the job.
     
    You should have individual charge balancing resistors across the supercaps.
     
    What are you actually trying to achieve?
    #2
    acharnley
    Super Member
    • Total Posts : 268
    • Reward points : 0
    • Joined: 2016/05/01 06:51:28
    • Location: 0
    • Status: offline
    Re: Preventing Buck backflow using PIC16/18 comparator 2019/02/25 10:45:18 (permalink)
    0
    Why would it happen? Well at a low duty the bottom FET is on for a much greater amount of time. If the inductor was to saturate the FET will get nice and hot complements of the super caps discharging. I'm fairly sure sync buck chips turn the bottom FET off at low duty cycles (the ones that don't tend to have bad efficiency at light loads). 

    To do it dynamically I can only see sensing the current through the inductor and turning it off. Maybe I could do it through math if I know the input voltage, output voltage and duty. The control loop is whatever I implement, so feel free to suggest as that's exactly what I'm asking!

    PS) there's no buck that'll do low voltage to 100V. LT comes close at 7V and that chip is very expensive. I guess this range isn't really practical as the current sensing dividers won't give the necessary accuracy. I'm likely to have the same issue if I try to do detection with the PIC comparator.

    PPS) Aware of the super cap leakage resistors, but thanks!



    #3
    acharnley
    Super Member
    • Total Posts : 268
    • Reward points : 0
    • Joined: 2016/05/01 06:51:28
    • Location: 0
    • Status: offline
    Re: Preventing Buck backflow using PIC16/18 comparator 2019/02/25 11:03:18 (permalink)
    0

    Attached Image(s)

    #4
    acharnley
    Super Member
    • Total Posts : 268
    • Reward points : 0
    • Joined: 2016/05/01 06:51:28
    • Location: 0
    • Status: offline
    Re: Preventing Buck backflow using PIC16/18 comparator 2019/04/05 11:33:32 (permalink)
    0
    Still after suggestions on this one.

    I hypothesis if the duty is say 10% (so the inductor would charge some and the 90% would be the discharge cycle). On the discharge cycle the voltage from the inductor decreases until it's marginally greater than the output and there'd be little current transfer - so reverse transfer wouldn't happen for a measure of time, which would be the rate of inductance decay. 



    #5
    Bob White
    Super Member
    • Total Posts : 239
    • Reward points : 0
    • Joined: 2010/11/06 19:52:38
    • Location: Denver, Colorado
    • Status: offline
    Re: Preventing Buck backflow using PIC16/18 comparator 2019/04/05 13:32:59 (permalink)
    +1 (1)
    One the attributes of a synchronous buck regulator is that the inductor current can at times flow from the nominal output to the input.  This happens when the average output current is less than one half the peak to peak ripple current.
     
    If you really do not want this to happen, then you need to switch to what I call the "classic buck" converter where the freewheel device is a diode and not a MOSFET.  This guarantees that the inductor current can never be negative.  The cost to achieve this is lower efficiency.
     
    When you do this, when the average output current is less than half the peak-to-peak ripple current the converter goes into discontinuous conduction mode (DCM).  This works fine but the dc conversion ration becomes nonlinear and the small signal control-to-output transfer functions also radically (from a two-pole to single pole characteristic).
     
    There are PWM and gate drive ICs that detect the reverse flow of inductor current and turn off the synchronous switch/MOSFET.  This mode is called the "diode emulation mode".  One Microchip gate driver IC that allows you to do this is the MCP14628.  Using such a gate drive IC would probably be the simplest and most effective way to prevent the buck converter operating such that at some part of the cycle the inductor is driving current into the input.
    #6
    acharnley
    Super Member
    • Total Posts : 268
    • Reward points : 0
    • Joined: 2016/05/01 06:51:28
    • Location: 0
    • Status: offline
    Re: Preventing Buck backflow using PIC16/18 comparator 2019/04/05 15:26:43 (permalink)
    0
    Thanks Bob. Regarding the MCP14628 I need 100V ideally (at very low duty) and best I could do was the MIC4605/85V.
     
    So my options are to remain in CCM (which I can do by ensuring the input voltage is a margin above the output voltage) or when entering DCM ensure the bottom FET is turned off. Question then is without current sensing the inductor, can I determine the CCM to DCM duty point by equation?

    Things I know are; input voltage, output voltage, input current (fixed 0.5A), frequency, inductance, inductor + fet resistance.

    Note the output is capacitive not resistive. One of the reasons for home brewing this is I'm using maximum power transfer theorem and so need digital control of the duty to drive the input voltage to a particular point.



    #7
    Bob White
    Super Member
    • Total Posts : 239
    • Reward points : 0
    • Joined: 2010/11/06 19:52:38
    • Location: Denver, Colorado
    • Status: offline
    Re: Preventing Buck backflow using PIC16/18 comparator 2019/04/05 19:41:44 (permalink)
    +2 (2)
    You simply replace the bottom MOSFET (the synchronous switch) with a diode.  That would assure no negative inductor current.
     
    Shameless plug: Are you coming to the MASTERs conference in August?  If so plan on attending the Fundamentals Of Switch Mode Power Conversion Classes (PC1 and PC2) I will be teaching.  I am teaching on both Wednesday and Thursday mornings so whatever fits your schedule...
     
    #8
    acharnley
    Super Member
    • Total Posts : 268
    • Reward points : 0
    • Joined: 2016/05/01 06:51:28
    • Location: 0
    • Status: offline
    Re: Preventing Buck backflow using PIC16/18 comparator 2019/04/06 00:45:15 (permalink)
    0
    Yeah that's the easy solution, but I'm not about the easy. If I can gain a few more % at low voltage operation I'm going to try! Is there a means to work out the CCM / DCM boundary point mathematically?

    Don't know anything about that conference. I'll do a google later.
    #9
    Bob White
    Super Member
    • Total Posts : 239
    • Reward points : 0
    • Joined: 2010/11/06 19:52:38
    • Location: Denver, Colorado
    • Status: offline
    Re: Preventing Buck backflow using PIC16/18 comparator 2019/04/06 12:48:03 (permalink)
    +1 (1)
    CCM/DCM boundary is when the average output current (Iout, Vout/Rout) equals one half the peak-to-peak ripple current in the inductor.
     
    I_ripple_pk_pk = (Vin - Vout) * Ton/L = (Vin - Vout) * D/(Fsw * L)
     
    where, D, the duty cycle, equals Ton/Tswitch
     
    MASTERs conference is Microchip's annual customer training conference.  In the US it is a several day/week long event held in August in Phoenix (August does not matter, you will be freezing in the nuclear powered air conditioning).  Not inexpensive but a great experience (note that the cost includes everything from the time they pick you up at the airport until they drop you off at the airport - airport transportation, hotel room, all meals, the conference itself)

    https://secure.microchip.com/usmasters/home.aspx
    #10
    Jump to:
    © 2019 APG vNext Commercial Version 4.5