Mosfet driver critiques wanted

Discussion in 'The Projects Forum' started by Jimmeh30, Aug 4, 2011.

  1. Jimmeh30

    Thread Starter New Member

    Mar 4, 2011
    22
    1
    Hello again folks :D

    Just wondering if anyone would be interested in reviewing and making suggestions as to wether or not/how well this setup would work.

    Reason I'm asking is that while testing a seperate circuit on the same board (temp switch for fans on the heatsinks), the TL494 shorted and burst open and I can't find why. Both pot's were at 50% at the time of failure and unless I've missed something, there are no shorts on the proto board.

    Also, at the time of the failure, the outputs (pins 9&10) were open/not connected and the circuit was running fine for 2 or 3 minutes before the fault.

    Before the failure, the scope showed clean square waves on both outputs, freq from 10Khz-well over 150Khz and the dead time control was working perfectly.

    Can anyone see a reason for this failure? Or is it likely that there was a short somewhere that caused it?

    Also, I haven't got as far as testing the totem poles and final drivers and am unsure about the diode configuration in the final stage.

    I intend to use the circuit to drive TV type flyback xfmrs.
    Any help would be greatly appriciated.
     
  2. mik3

    Senior Member

    Feb 4, 2008
    4,846
    63
    There are lots of mistakes in your circuit.

    Read the datasheet carefully and take advice from the example circuits there on how to use the output transistors. Also, your totem pole pairs are not connected the right way. The PNP shall be on top and the NPN at the bottom.
     
  3. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    The BD681/BD682 pairs are configured as emitter followers, which is acceptable - however, the way the TL494 output is not connected correctly; the emitter follower pairs will never return to ground once pulled up by the TL494.
     
    Jimmeh30 likes this.
  4. Jimmeh30

    Thread Starter New Member

    Mar 4, 2011
    22
    1
    Thanks for replying Mik3.

    This is about the tenth attempt at this, all previous attempts have failed due to poor (or no) transient supression and were made with 555's as a signal source. This circuit is one I've adapted from two circuits, one being from a "O. Hunt" from Hvlabs.com and the snubber network from Jan Martis' driver circuits which can be seen on youtube under "super flyback driver final version"

    Could you point out a few of the mistakes you're referring to?

    That totem pole design I got from Fig. 10 on page 12 of this document, http://focus.ti.com/lit/ml/slup169/slup169.pdf

    It works quite well, though it clamps the gate voltage to the voltage at the two base's inputs. I believe this is because they're connected in that configuration thought I'm not sure about that. (if anyone knows I'd be interested as to why)

    As for the rest of the circuit, am I on the right track?
    I've attached a revision of the one in the original post.
     
  5. Jimmeh30

    Thread Starter New Member

    Mar 4, 2011
    22
    1
    Also, Thanks to you also Sgt Wookie :) I was wondering about that.

    Do you think that the snubber arrangement is a good start?
    Maybe you could suggest something more suited/robust?

    This is far beyond my scope and becomming a fairly expensive learning curve.

    My biggest problem is that I've only got 1x probes for my o'scope atm and I'm not game to try and measure the transients until I can learn enough to be confident in choosing some 100x probes. (1st scope I've owned and I'm still learning how to use them)
     
  6. iONic

    AAC Fanatic!

    Nov 16, 2007
    1,420
    68
    Jimmeh30,
    I'll be keeping an eye on your thread, but can't say that I'll have much to offer.
    What is the application for this circuit?
     
  7. Jimmeh30

    Thread Starter New Member

    Mar 4, 2011
    22
    1
    iONic,
    I've got a few objectives, but mostly it's for the learning curve.

    This circuit is the first attempt at a: driving anything with a MOSFET, b: trying to learn about these fandangled tl494's :p and c: building a reliable HV power supply.

    Aside from that, a 250KV+ plasma speaker would be puuuurdy and I have a local source of new cheap tv flybacks without rectifiers built in. If I can get it to work reliably I have several comicly large incandescent light globes that would make good plasma balls and I'd also like to drive a few rectified flybacks to play around with Ion wind & thrust generators, maybe a lifter or two, perhaps try to wind and tune it to a magnifier coil. The usual HV suspects :D

    It's also the first time I've stepped away from robotics and logic. Making any sense at all of datasheets has been a learning curve in itself.
     
  8. Jimmeh30

    Thread Starter New Member

    Mar 4, 2011
    22
    1
    Sorry,

    Just realised I hadn't uploaded the revised circuit, so here it is...
     
  9. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    I didn't even pay attention to your input on the left this morning; I was in somewhat of a hurry.

    I guess you're trying to make a Class D amplifier. You can't simply feed audio into the TL494; that just plain will not work. You need to convert audio signals to digital first, and ensure that you will have sufficient "dead time" or the MOSFETs will go up in smoke.

    Rather than try to fix up this one, you should have a look over on Rod Elliott's website, "Elliott Sound Products", here:
    http://sound.westhost.com/

    Pay particular attention to the Projects and Article sections.
    In the Articles section, under Amplification, the PWM amplifiers article link is:
    http://sound.westhost.com/articles/pwm.htm

    In Projects, under Power Amplifiers, you'll find this Class D amplifier project:
    http://sound.westhost.com/project114.htm

    Note that he's used a ColdAmp module.

    If you continue attempting to make your own design, you will be working on it for a long time, and you will burn up a lot of parts and spend lots of money in the process.
     
  10. Jimmeh30

    Thread Starter New Member

    Mar 4, 2011
    22
    1
    Thanks again Sgt.

    I'll have a read through it now.

    Cheers :)
     
  11. Jimmeh30

    Thread Starter New Member

    Mar 4, 2011
    22
    1
    Having read most of that, it would be a much better way to go. I will defenitely spend more time studying that site. Allways wondered about class D operation.

    With the circuit I'm proposing tho, shoot through can't happen because both pairs of FETs are switching the same rails. My thinking was it would allow me to run it in half bridge at a later date, and for now will reduce the stresses on the FETs because they're only on for half the time. The extra two diodes in the first revision, I placed thinking that they might be needed to stop the conduction through either pair while the gate charge falls in the event that the dead time was insufficient.

    I'd wager I'm right off there. Also they're shotkey types from old computer psu's (3 leg rectifiers >|< wired parallel as one diode) and only good for 45Vfwd so most probably, the back emf will destroy them anyway?

    Definately though, you're right about the input. The threshold for the DT Control is 3-3.3v (0%Duty) I should atleast put an opamp infront of it to shift it from line level.

    Do line level outputs swing negative? or are they only 0-2v typ? Either way, I'd need to do more reading because I've never used an op amp either.

    I did consider leaving the DTC variable and modulating through one of the error amplifiers, but again, I've no idea how to go about that. Would that not digitise the audio in much the same way as the class D's you referred me to? The block diagrams are very similar, sort of.
     
  12. Jimmeh30

    Thread Starter New Member

    Mar 4, 2011
    22
    1
  13. Jimmeh30

    Thread Starter New Member

    Mar 4, 2011
    22
    1
    Just a quick update, still banging my head against this thing to no avail.

    Mik3, you were right, ordinarily that totem pole would work, but the BD681/682's have back emf protection diodes built in, so I found tonight when I looked back at the data sheets again.

    I built and tested (with a speaker) the tl494 class D amplifier, it does work, quite well, but I couldn't get it stable nor could I get the speaker to sit in it's "neutral location". It did however work as an amplifier very well. I'm leaving that for another day though.

    I've built another modulated sig generator from two 555's and so far I've got nice lean square waves from 10k- over 100K and duty adjustment from 50% up. The audio modulation is both lengthening/shortening the on/off respectively AND vice versa.

    Tonight I'll rebuild the totem pole in the reverse polarity, test it and then try to remember all the test's I've run in the past fortnight to try to figure out just how many components I've scrapped, thinking that totem driver was functioning correctly. >:/
     
Loading...