Page 1 of 2 12 LastLast
Results 1 to 10 of 14

Thread: High voltage boost using UC3845. Need some assistance!

  1. #1
    diysmps Member
    Join Date
    Mar 2014
    Location
    Lancashire
    Posts
    12

    High voltage boost using UC3845. Need some assistance!

    I'm pretty new to SMPS.I am trying to design a boost converter using the UC3845. I want to step up from 19V (e.g. laptop PSU, but currently using a bench power supply) to about 250V. It would be nice to get 50mA out of it, but I would settle for less.

    I have a circuit that is sort of functioning, except its load regulation seems unusually poor. Here's the problem:
    Switching frequency ~39kHz
    296V into 100k load, duty cycle 20.2%
    250V into 25k load, duty cycle 29.5%
    242V into 20k load, duty cycle 31.8%

    Can anyone suggest why the output is so poorly regulated, and how to fix this?
    sdfddfs.png
    Last edited by daviddeakin; 06-19-2014 at 10:45 AM.

  2. #2
    diysmps Member
    Join Date
    Mar 2014
    Location
    Lancashire
    Posts
    12
    Just noticed a typo: R3 is actually 1k, not 10k.

  3. #3
    diysmps Senior Member
    Join Date
    Oct 2012
    Posts
    266
    To get the voltage ratio you want means a duty cycle of 0.924. The UC3845 is internally limited to D<0.5 which limits your output voltage to 2xVin=38V. UC3842 or UC3843 would be more capable here. It's still a big ratio to ask of a boost converter. you will also need slope compensation as the duty cycle is high and if this is operating in continuous conduction mode, you need to account for the right hand plane zero.

    I'm making the assumption that your inductor and output capacitor have been calculated correctly. Compensation could probably also be optimised for a better loop response.

    Matt
    Buck derived converters in regulated power supplies require output inductors
    and the following control modes (voltage or current mode control; bold is recommended).

    Buck (CMC/VMC) | Forward (CMC/VMC) | 2-switch forward (CMC/VMC) | Half Bridge (VMC) | Push-pull (CMC) | Full Bridge (CMC). Click for basic schematics.
    ON Semiconductor TL431 design guide

  4. #4
    diysmps Senior Member
    Join Date
    Oct 2012
    Posts
    266
    I went through the textbook calculations for a boost converter power stage (continuous conduction mode, but you could probably still do a DCM one at the required 12.5W output power, peak currents would be bigger, but no RHP zero and diode speed doesn't really matter; there are pros and cons). This was done at 100kHz, the frequency at which I always start, and there's not much reason not to have a higher switching frequency than you already have in a non-isolated DC-DC conveter.

    Duty cylce is 0.924 at 100% efficiency, accounting for 80% efficiency it would be 0.94. I've assumed the input voltage is a perfect 19V source, if it dips below 19V, the duty cycle increases. [With such a high duty cycle, you may find it worth considering a flyback converter instead. Quite a similar topology but with the help of a turns ratio to lower the duty cycle.]

    The inductor value I got out was between 0.68mH-1.3mH depending on the ripple current you want to allow in the inductor(20-40%). 1mH is proably a good value here and gives you a choice in inductor selection, but I personally usually use 20% so that I can keep it in continuous conduction down to 10% load (below which it transitions to DCM, the loop response changes, it generally becomes more stable but has a worse transient response.)

    The peak current though inductor, switch, diode was somewhere in the 800-900mA range, 860mA or 880mA I think (I'm writing this from memory the next day). This means your diode needs to be rated for 50mA average, but 900mA repetitive peak. UF4007 is not a fast diode by SMPS standards (trr=75ns), and there may be a better choice, but i haven't found one without going to FREDs (typically at least twice as fast and not too expensive) or SiC schottkys (expensive) It might not be too significant a problem, and you might be able to handle it with an RC snubber etc. Of course if you do a DCM converter, the diode speed shouldn't matter because it there's no inductor current when it switches off.

    For the output capacitor, you can choose this based on the acceptable ripple voltage. My initial calculation got me around 4.7uF, but at this output voltage it's difficult to find an aluminium electrolytic capacitor anywhere near this with low ESR. doing the calculations on the ripple due to ESR, I found it looks like there's less ripple by using a polypropelene cap of lower capacitance but with negligable ESR. I found a 470nF 400V one for 0.53 (with a pin pitch of about 18mm, they're quite big). That would give an output ripple of 1V (0.4% output voltage) before accounting for the larger high frequency switching spikes, which could both be further reduced with a second LC filter outside the control loop.

    Current sense resistor's probably around 1 ohm, but I haven't checked that. Input capacitor, I've not done any calculations. Looking at it by eye, I'd guess it's probably around 47-100uF.

    I haven't done the feedback compensation calculations yet other than based on the worst case bias current of the FB pin, the total resistive divider could add up to about 1.25Mohm with <1% bias error. fixed resistors of 3x330k on top and 10k below might be convenient values since they're in the E6 range. The power lost in the divider would be 0.5% of output power. having the total resistance around 500k is also perfectly valid. I'd tweek in this range while working out the compensation network.
    Last edited by KX36; 06-27-2014 at 09:51 AM.
    Buck derived converters in regulated power supplies require output inductors
    and the following control modes (voltage or current mode control; bold is recommended).

    Buck (CMC/VMC) | Forward (CMC/VMC) | 2-switch forward (CMC/VMC) | Half Bridge (VMC) | Push-pull (CMC) | Full Bridge (CMC). Click for basic schematics.
    ON Semiconductor TL431 design guide

  5. #5
    diysmps Senior Member
    Join Date
    Oct 2012
    Posts
    266
    Just calculated the RHP zero and it's not great news. At D=0.924 its around 3.5kHz, but the small increase to D=0.94 drops this to 2.2kHz. As I said, boost converters are really stretched at such a voltage ratio and duty cycle. Realistic practical boost converters aren't really comfortable above about a 3x-5x boost voltage maximum

    By the way, what software are you using for your schematics? It looks identical to that in a couple of books I have by Merlin Blencowe http://www.valvewizard.co.uk
    Last edited by KX36; 06-27-2014 at 10:30 AM.
    Buck derived converters in regulated power supplies require output inductors
    and the following control modes (voltage or current mode control; bold is recommended).

    Buck (CMC/VMC) | Forward (CMC/VMC) | 2-switch forward (CMC/VMC) | Half Bridge (VMC) | Push-pull (CMC) | Full Bridge (CMC). Click for basic schematics.
    ON Semiconductor TL431 design guide

  6. #6
    diysmps Senior Member
    Join Date
    Oct 2012
    Posts
    266
    My gosh I'm generous. Not even sure you're still there...

    Buck derived converters in regulated power supplies require output inductors
    and the following control modes (voltage or current mode control; bold is recommended).

    Buck (CMC/VMC) | Forward (CMC/VMC) | 2-switch forward (CMC/VMC) | Half Bridge (VMC) | Push-pull (CMC) | Full Bridge (CMC). Click for basic schematics.
    ON Semiconductor TL431 design guide

  7. #7
    Hello all,
    This one is from a friend. I did some reverse work but I do not know what IC is. It could be the same chip as previous post. It gives about 15W regulable Out V and It can delivers 250V.
    Any suggestion will be welcome.
    Thanks and regards.top boost 001.jpg

  8. #8
    diysmps Member
    Join Date
    Mar 2014
    Location
    Lancashire
    Posts
    12
    Hey KX36, thanks for the input! I posted the same question on another forum, and the general consensus was "you need more inductance". However, I can't really grasp why it doesn't work as is.
    I mean, I AM getting suitably high output voltages using my inductor, and the IC has no shortage of loop gain. I know it CAN go up to 50% duty cycle, yet when I load the output it refuses to make the duty-cycle go high enough to maintain the same output voltage.
    296V into 100k load, duty cycle 20.2% <<OK, great.
    242V into 20k load, duty cycle 31.8% << the chip is capable of going up to 50%, so why won't it do it! Why is it stubbornly sticking at a measly 31.8% in this case?

    (And yes, I stole some digrams from valvewizard.co.uk because I think they look cool, and I've started using them to make my own diagrams with a simple image editor!)

  9. #9
    diysmps Senior Member
    Join Date
    Oct 2012
    Posts
    266
    There is a fixed equation for duty cycle in a boost inductor. Vout/Vin=1/(1-D), or D=(Vout-Vin)/Vout. The question is how does your circuit seem to break the equation. I believe the answer is that it doesn't. The inductance is low enough that it works in DCM and completes both parts of the switching cycle well under half the period of the oscillator. And that's the period you're mistakenly measuring as duty cycle. Peak currents will be unnecessarily high.

    You'd still be better off with a bigger inductance or faster switching frequency, an IC which allows a higher duty cycle and proper loop design to match it. Or just use the working one I designed for you of course.
    Last edited by KX36; 07-01-2014 at 01:31 PM.
    Buck derived converters in regulated power supplies require output inductors
    and the following control modes (voltage or current mode control; bold is recommended).

    Buck (CMC/VMC) | Forward (CMC/VMC) | 2-switch forward (CMC/VMC) | Half Bridge (VMC) | Push-pull (CMC) | Full Bridge (CMC). Click for basic schematics.
    ON Semiconductor TL431 design guide

  10. #10
    diysmps Member
    Join Date
    Mar 2014
    Location
    Lancashire
    Posts
    12
    Hullo KX36! Yes I'm dragging up this old thing again - groan! I've ordered some UC3843 so I can try your circuit finally.

    While I wait for them to arrive I have been trying to understand why you say (and I believe you) that the circuit is so borderline, struggling to acheive the required step up ratio, when the following simple circuit manages an even higher step-up ratio using the same principle:


    OK, in this case the 555 is being used as a pulse position modulator, rather than PWM, but is that such a crucial difference? I mean, as a PPM it still can't vary the duty cycle to anything more than what the UC3843 can! How is it that such a cheap circuit can do apparently with ease what a 'proper' UC3843 only just acheives while being 'not recommended'?

Similar Threads

  1. Replies: 13
    Last Post: 06-26-2014, 08:27 PM
  2. Using an ATX transformer "backwards" to get high voltage?
    By daviddeakin in forum Power Supplies
    Replies: 1
    Last Post: 03-25-2014, 12:47 AM
  3. DC Boost converter
    By tonigau in forum diysmps
    Replies: 1
    Last Post: 02-06-2012, 11:51 PM
  4. 4-10kw boost converter/pfc
    By staska in forum diysmps
    Replies: 0
    Last Post: 01-14-2012, 09:55 AM
  5. High voltage ZVS supply.
    By carampman in forum Power Supplies
    Replies: 0
    Last Post: 07-26-2011, 01:42 PM

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •