What caused excess current in this amplifier?

Discussion in 'Analog & Mixed-Signal Design' started by Veracohr, Oct 27, 2018.

  1. Veracohr

    Thread Starter Well-Known Member

    Jan 3, 2011
    660
    100
    I'm working on a 200W audio amplifier. I've made several mistakes along the way, but the most recent leaves me puzzled. I wired the output wrong and it caused two resistors to burn up, but I can't figure out where that much current came from.

    It's designed as a bridge amplifier. The image below shows just the output sections of each side of the bridge, from the simulation I've been using. I wired the outputs wrong, so that both NPNs went to one output and both PNPs went to the other. That's reflected in the image below as the "outp" and "outn" outputs. The power transistors and 0.1Ω resistors are off the PCB. About 15 seconds after I flipped the power switch (no input signal and no speaker load), R42 let the smoke loose, and I even saw a small red flash come from it. Upon further inspection, R14 is discolored as well. Everything else is fine.

    Screen Shot 2018-10-27 at 6.55.10 PM.png

    These are 1/4W resistors, so it had to have been over 50mA to burn them up. In the simulation these resistors only get 3mA through them when the output is wired wrong. I didn't get a chance to measure any voltages before the smoke. The power rails for the power amp section are unregulated and come from a 48V center tapped transformer. I've attached the LTSPICE file with the power supply section and output section. There's more to the whole circuit, but there's no DC connection between the preamp and power amp sections.

    What could have caused that much current to go through only those two resistors, and not the other two 100Ω resistors? I want to figure out if there's any other problems before I fix the wiring and try again.
     
  2. crutschow

    Expert

    Mar 14, 2008
    20,067
    5,652
    I don't understand how it was miswired.
    Can you draw that on the schematic.
     
  3. Veracohr

    Thread Starter Well-Known Member

    Jan 3, 2011
    660
    100
    It should have been like this:

    Screen Shot 2018-10-27 at 11.58.01 PM.png


    But instead it was like this:

    Screen Shot 2018-10-27 at 11.58.58 PM.png
     
  4. MisterBill2

    Well-Known Member

    Jan 23, 2018
    1,547
    272
    WE still have no idea which two resistors burned up. and does it have the problem now??? Spice simulations only have a chance of being correct if the actual circuit is the same as the circuit in the simulation. AND it is only by guessing that the drive voltages are presumed. OR did the thing burn up before it was driven?
    NONE of us are mind readers. We need the information, please.
     
    ArakelTheDragon likes this.
  5. ArakelTheDragon

    Active Member

    Nov 18, 2016
    967
    72
    Also:
    Please mark the important parts in red.
    Did you calculate the equivalent resistance and voltage drop of the circuits without a load? Never connect a circuit without a load.
    Are all the resistors the same wattage?
    Please put exact values (Vcc = 12VDC) and put all markings(what are the bases connected to).
     
  6. Alec_t

    AAC Fanatic!

    Sep 17, 2013
    8,755
    2,021
    With the mis-wire shown (but not with the correct wiring), loss of feedback from the emitters of Q5/Q6 to the base of Q2 would account for R14 and R42 frying, since they then pass ~240mA according to the sim.
     
    Last edited: Oct 28, 2018
  7. Veracohr

    Thread Starter Well-Known Member

    Jan 3, 2011
    660
    100
    Ah, I see. The feedback comes from the output connector back to the PCB through a MTA connector. I checked that connection before posting and it was good, but I guess I'll assume there was an issue with it when I first turned it on. Thanks.
     
  8. ebp

    Well-Known Member

    Feb 8, 2018
    2,060
    726
    The 3904's/6's are for current limiting. When the voltage across the 0.1 ohm resistors reach about 0.7 V (so 7 amperes) the small transistors begin to conduct, "robbing" the output transistors of their base drive.

    The output stages are presumably biased for class AB operation (can't tell from the partial schematic). This means both the output transistors are trying to conduct some current. Normally that current flows through both of them and the current limit sensing resistors from the positive supply to the negative supply. When you broke that path, the bias current would "try" to flow through any available path - which in the case of the right hand output stage would be R42 and mostly the base-emitter junction of Q32. This would turn Q32 on very "hard" which should turn Q26 and A31 off. But worse than that, the bias current of Q15 would also be looking for a path to the negative supply via R2, R40 and R42. Of course turning Q32 on harder does nothing to reduce the current through Q15.

    I'd recommend replacing Q32.

    The circuit should survive without a load. Had the load been connected in this instance, the bias current would have found a path through the load, which might have been OK - or not!
     
  9. ebp

    Well-Known Member

    Feb 8, 2018
    2,060
    726
    should have mentioned:

    I''m assuming the fact the current sense resistors are off-board is for some particular purpose. It is always perilous to use this sort of configuration because of the potential for connection issues, though much less so for something that lives in a closed box where the connections can't be messed with once properly made initially. For something like a power supply with remote sense it can be a big problem if connections get messed up.

    In this case, it might be prudent, if possible, to put higher resistance sense resistors on the board. Something in the range of an ohm or 2 or 3 might be acceptable and not interfere too significantly with the external resistors. It isn't totally foolproof, but might be worth considering if there is any risk of connection issues during normal use.
     
  10. Veracohr

    Thread Starter Well-Known Member

    Jan 3, 2011
    660
    100
    The sense resistors being off-board is basically because the power transistors are off-board. At my local electronics surplus shop I found two big heat sinks drilled for 4 TO-3 transistors each. They originally each had MOSFETs in them and a driver board attached. I tossed the driver board, kept the MOSFETs for some future use just in case, and I'm using one of the heat sinks for the 4 output transistors. The MJE182/172 driver transistors are on the board (with small heat sinks themselves).

    Since I had to go off of the PCB to the power transistors, it seemed silly to me to bring it back in just to go to the sense resistors, just to have to go back out again to the output connectors. So I left them off-board too, and wired them directly to the output connectors. On my first attempt I used the type that are encased in ceramic, and they were free-hanging between the transistor wire leads and the output connectors. That original version had an unrelated problem (I messed up the pinout of the bridge rectifier). On the second revision I decided to use the type of power resistors that come in their own heat sinks, and they're now bolted to the chassis. I'm going out from the PCB to the power transistors (the emitter connection is really just there for the current sensing), from the power transistor emitter to the 0.1Ω resistor, and from the resistor directly to the output connectors. Then I have wires going from the two output leads back to the PCB for the feedback. That's where the issue must have arisen.

    I don't think I can use higher sense resistors. Based off of the simulations, where I of course had to make an assumption about the power transformer, my current design shows the voltage across the load dropping to 40V at the heaviest load, which is just enough to allow 200W output into a 4Ω load (two 8Ω speaker loads in parallel). Of course, I'm simulating a resistive load, but a realistic model of a speaker is kind of complex, and I don't have a real speaker to model just yet, so I decided to just use a resistive load for the simulation. If I increased the current sense resistors, the output would start limiting before 200W. I may not NEED 200W, but that's the design goal I'm going for, and this project is at least partly a learning opportunity.

    I suspect you may now be tempted to ask: why such a low voltage transformer so that it has to be designed this way? Essentially I was looking for the cheapest option that would give me a 200W amplifier, and a high-current, low voltage transformer with a bridge design was at least $100 cheaper than a higher voltage, single-ended design, as far as the power transformer goes.

    It's funny, because before I created this thread I tried simulating with various leads unconnected. I assumed the current had to have come from an imbalance somewhere. I think I even simulated with BOTH of the feedback connections missing, which does not result in excessive current. If I had tried simulating with only ONE feedback connection missing, I'd have seen the problem.
     
  11. ebp

    Well-Known Member

    Feb 8, 2018
    2,060
    726
    The higher value sense resistors wouldn't replace the external ones, but remain in parallel. Clearly with your construction the problem would not be expected to occur again once everything is assembled, so it really is a non-issue.

    Using the bridge makes sense in view of the price of the transformer.
     
Loading...