Connect with us

I can't explain this by myself...

Discussion in 'General Electronics Discussion' started by OhMy, Jun 18, 2014.

Scroll to continue with content
  1. OhMy


    Apr 5, 2014
    At first I built this circuit "First try.gif" for a (hopefully) 250W inverter DC boost stage, but it was very unsatisfactory: the current sense transformer seemed to make a mess, so it got excluded; the main problem was burning of IRFZ44 mosfets.
    Next I re-made the circuit and the PCB into "Circuit.gif" and "PCB.gif". Now it seemed to be a bit harder to burn mosfets (although not impossible :) ), but the never ending problems came out at the output rectifier.
    The diodes started to die one after another, sometimes when a 60W 230V bulb was connected, sometimes a bit later; sometimes, when the converter got reconnected to a power source. For no apparent reason. I even searched for some used (had no new ones) MUR460 and UF5407 diodes, but very soon MUR460 past away, and not long after one UF5407 too. I put a snubber (50R + 2n2) at the output too, but with no improvement.
    But THE MYSTERY (at least for me) is in the fact, that 95% of the time faulty appeared only one or two of the diodes, which are marked with red dots in "PCB.gif". Does it make any sense?

    Please, take a look at the pictures, maybe there are some apparent mistakes?

    Thank you.

    Attached Files:

  2. duke37


    Jan 9, 2011
    The battery is connected the wrong way round.
    Diodes VD1 and VD2 do nothing.
    I assume these are just drawing errors.

    Is the frequency high enough to stop the transformer going into saturation?
    Is the primary bifilar wound?
    Is the frequency low enough to keep the dissipation in the rectifiers reasonable?
    Can you look at the drain voltages?

    Is the dead time long enough?
    Do you need the IR2110? You do not need a high side driver.
    Last edited: Jun 18, 2014
    KrisBlueNZ likes this.
  3. OhMy


    Apr 5, 2014
    Thanks, Duke37.

    It's always been confusing for me to know, which way is correct :). But I can assure, that I've seen it drawn in both ways in many cases; same goes for capacitors. That's why I hate symbols in schematics with no polarity indication.
    Sorry, was in a hurry, when made some adjustments in the circuit. At first I had only one 40V schottky at each shoulder, but they got shorted almost immediately, so now there are two in series. So far they hold.

    The frequency is about 50kHz, but it can be adjusted somewhat with a 10k timing trimpot, but the results doesn't seem to change much.
    The primary winding's both halves are split in two and wound below and above the secondary with 10 strands of 0.56mm wire, so that in each halve period there are two turns below the primary and two turns above it "Coil.gif".
    The output rectifiers don't get hot, at least for the time periods they stayed alive- just slightly warm.
    If you mean oscilloscope, then no, I don't have one, just a DMM with frequency and duty cycle measuring capabilities.
    The dead time should be enough, because the duty cycle doesn't go above 47.5%.
    Probably not, but I tried to drive mosfets directly from SG3525 and through BC337 + BC327 pairs, but the results were somewhat bad (six or more burned mosfets). I had some IR2110, so used it as a dual low side driver, as seen in Fig.11 here- Mosfets seem to prefer this way noticeably better. So far only one died...

    Attached Files:

  4. duke37


    Jan 9, 2011
    I am somewhat at a loss here.
    The components will fail either from excess current or voltage. The rectifiers are unlikely to have excess current so that leaves excess voltage.

    In order to limit the voltage on fet switch off, there must be close coupling between the drain connections. It looks as if you do have bifilar winding which is essential.

    Without a scope you could place a diode on each drain, charging a capacitor and so measure the peak drain voltage. It should not be above 30V. A fat zener across the capacitor could limit this voltage at the expence of power loss.

    I find that a cheap inductance meter is helpful when winding my own transformers. You could measure coupling between the two primary windings.

    EdiI: I have had another look at the diagram, there is no bypass capacitor close to the transformer, this could give high voltages from rapid switching transients. Try 1000µF in parallel with 100nF.
    Last edited: Jun 19, 2014
Ask a Question
Want to reply to this thread or ask your own question?
You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.
Electronics Point Logo
Continue to site
Quote of the day