overheating power transistors

Discussion in 'The Projects Forum' started by malcolm, Mar 22, 2011.

  1. malcolm

    Thread Starter New Member

    Mar 6, 2010
    7
    0
    my last communication was last year with what turned out to be soldering problems. Your advice was gratefully recieved and since I have soldered on (pun intended).
    I now have a whole bunch of operable passive second order filters and I'm building power amps for each one.
    Currently then, I have built a Jean Hiraga Class A whose power transistors get so hot the solder melts off the emmitter tag, the power supply fuses blow and the junction resistances go to zero. What's going on.
    I have a heat sink which dissipates 1.3W/degC.
    Driver transistors etc show high junction resistances and all passive components and circuit paths seem to check out. What have I missed?

    I only have a multimeter.
    Can you help?
     
  2. ifixit

    Distinguished Member

    Nov 20, 2008
    639
    108
    Hi,

    With the input shorted to ground and no load on the output, check the output transistors quiesent current (Measure the voltage across the .33Ω resistors).

    The design calls for alot of current. Here is a quote from someone who built one...

    "My version of the Jean Hiraga Super Class-A Amplifier is running at a bias of about 1.65A @ 35V resulting in about 58W of continuous dissipation per transistor in the output stage (just over 1/3 their rating of 150W). As you can imagine, the heat sink runs quite hot, approximately 40 Celsius (100F) above room temperature."

    Check your design.

    Regards,
    Ifixit
     
  3. malcolm

    Thread Starter New Member

    Mar 6, 2010
    7
    0
    As you suggested, input to ground, output open circuit.
    Replacement fuses, replacement power transistors.

    Power on. Result. Ominous column of smoke somewhere near the middle of the circuit board. Source unknown.

    I could leave it all on untill something bursts into flames then I would learn more about the source, but I've decided not to do that.

    Does this result have meaning to you? If so what's the next step.
     
  4. ifixit

    Distinguished Member

    Nov 20, 2008
    639
    108
    Hi,

    I'm making reference to the http://diyaudioprojects.com/Solid/Jean-Hiraga-Class-A-Amplifier/ site for a schematic.

    Is this where you got your schematic? If not, post your schematic.

    Something is likely not biased, or built properly, or something has failed and is causing other things to overheat.

    Can you power up the circuit with limited power rail voltage, say ±10V. This may give you some time to measure with your meter without having something burn up. When I troubleshoot this kind of thing, I use a variac to bring power up slowly. Use your finger (carefully) to feel for hot components. This will give you a clue what part of the circuit is unhappy.

    Regards,
    Ifixit
     
  5. Adjuster

    Well-Known Member

    Dec 26, 2010
    2,147
    300
    With a new home build lots of things are possible, including bad joints, wiring errors, faulty components or components connected the wrong way round (or in the wrong place, e.g. an NPN transistor in place of a PNP), wrong resistor values... A really careful check of the circuit would be a good starting point, if possible by another person - we do not always find it so easy to spot our own errors.

    There may also be "layout" problems - did you use a professionally designed PCB? One way that an amplifier can misbehave is to oscillate, perhaps at a frequency which you cannot hear. Normally an oscilloscope would be required to rule that possibility out.

    Quite possibly, you may simply have a grossly excessive bias current. Is this is adjustable, and if so did you ensure that the setting was at its minimum before switching on?

    I would strongly suggest obtaining the use of a variable current-limited bench power supply for testing this thing, as just switching on the main power with the circuit in its present condition is likely to lead to a pile of ruined components.

    It is also a pity that you could not see what was smoking. Is there no evidence of something having overheated?
     
  6. malcolm

    Thread Starter New Member

    Mar 6, 2010
    7
    0
    Adjuster

    Thanks for that.

    I have mounted the circuit onto track board, so definately not a professionally designed PCB, although I have reflected the recommended PCB architecture with a few jumper leads added.

    I have thouroughly checked all the components passive and active and everything is correct. The circuit pathways also seem to be OK and I have also checked for continuity (crossed tracks etc.)

    There is no third person know to me who would be even remotely interested, let alone capable of running a check on what I've done, so that option's closed out.

    I believe that bias current is set to minimum, that is, the midpoint of the board mounted pot. Is this correct?

    I only have a multimeter. No oscilloscope and no variable current-limited bench power supply and actually no bench either.

    The power supply that I am using is self built and involves two no. 250W torriodal transformers which give dual rail voltage +/- 36.7V.

    I'm going to run with the idea of a reduced rail voltage and take the pd across the 0.33 Ohm resistors. I'll do that with a voltage splitter by trial and error (no idea of load resistance) untill I get approx. +/- 10V as suggested by ifixit.

    I strongly suspect that the driver transistors are smoking but I'll try the touch for hot test.

    This will all happen tommorrow as currently I've managed to blow all my spare fuses and under current circumstance feel reluctant to jump across the fuse holders. The fuses are 5amp.

    Help me out guys. I got very close today to firing up my sallen key sub base 4th order, variable fc (20 - 100Hz) low pass but didn't quite get there.
     
  7. Adjuster

    Well-Known Member

    Dec 26, 2010
    2,147
    300
    Do not even think of bridging out your fuses - you risk wrecking your power supply and perhaps starting a fire. The circuit should be drawing less than 2A, so if it is blowing 5A fuses you have a serious problem.

    Some means of reducing the current is necessary to test this more safely. If all you can get is resistors, remember that they will need to be rated to cope with what may turn out to be pretty much a short-circuit.

    Having seen the circuit, the bias current seems to be fixed. That pot. is for DC balance, and yes, mid-point would be the default position.
     
  8. malcolm

    Thread Starter New Member

    Mar 6, 2010
    7
    0
    Gentlemen,

    I have found some more fuses and run the circuit at +13.3/-12.0.
    Disconnected, the voltages are 30 both sides. So there is a slight slew to the positive in the circuit but that may be an irrelevance at this point.

    I find that there is no pd across the 0.33ohm resistor at all. Also the reduced rail voltages to earth are reflected exactly both sides of the 0.33ohm resistor.

    The column of smoke may well have been due to my connecting the driver transistor to the power transistor (2SC2240) incorrectly, that is, emmitter to base and collector to collector. OOPs!

    The above readings were taken with the transistors correctly connected.

    A resistance check on said transistors would indicate no lasting damage (high resistance across all three pins?)

    The Jean Hiraga web site mentioned is the article that I've used.

    I'm in your hands gentlemen.
     
  9. ifixit

    Distinguished Member

    Nov 20, 2008
    639
    108
    Hi Malcolm,

    Does the amp work now? When checking transistors, the forward biased B-E junction should look like a diode drop of .65V, or so, on the diode check position of your Multi-meter. An ohm reading can be seen as well, so long as the voltage used to measure ohms is high enough to make a transistor junction conduct.

    I have an LTspice schematic I put together for a reference, however I had to subsitute transistors. See the attachment.

    The circuit simulation worked very well, it came up right away and was very stable. The quiescent current was only 150mA, but seems to still produce a good output. 1.65A for quiescent is too high (IMO) the author suggests 1 Amp or less, I think 0.5 Amp would do, but crossover distortion will be higher. Your ears will be the deciding tool to use to determine the best setting.

    Adjust Quiescent Current:
    Set the pot to center, short the input and no load on the output. Adjust the value of R26 and R23 to change the quiescent current. Increasing the resistance, increases current. The transistor gains may not be equal between the positive and negative halves so these two resistors may end up being slightly different values to get the output to be close to zero at a particular quiescent current setting.

    Remove the input short. The output DC level may change a little because of the way the circuit is biased. Now adjust the pot to get the output to be <100mV DC.

    Caution! This design is DC coupled, so I would consider AC coupling the input signal to avoid getting DC on the output load. A 1uF cap will give a good flat frequency response from 20Hz to 20KHz.

    Have fun,
    Ifixit
     
  10. malcolm

    Thread Starter New Member

    Mar 6, 2010
    7
    0
    Thankyou you very much for you collective wisdoms.
    I now have an operating amp with the following parameters after fiddling about over the past day or so.

    R26 180ohm
    R23 100ohm

    This seems to hold the temperature down.

    I'm getting approx 60mV across R5 and R8 and 100mV output to ground.
    However, the pot is heavily biased towards the positive rail and there is distortion.
    To get a clean sound it seems to like the bias set slightly towards the negative rail which shoves the output to ground into the 800mV area.

    So, good result, although R26 and R23 values are in reverse to the original design.

    I have AC coupled the input as suggested with an electrolytric that I found in my odds box.

    One further problem!

    The input, which is CD fed, through a passive filter net and then lifted through a simple OP amp (rails at +/- 8V) produces a buzz which sounds damaging. This is not mains, its higher than 50Herz. Is this impedance mismatch?
    Your thoughts on this would be welcolmed.
     
  11. ifixit

    Distinguished Member

    Nov 20, 2008
    639
    108
    Hi Malcolm,

    The electrolytic is likely polarized and if the AC input signal reverses the polarity you may get signal distortion at low frequencies. What value are you using?

    Referring to my schematic I hope you mean the .33Ω R2 and R4.

    Something is still not quite right here. Perhapes you need more quescient current. More fiddling required:).

    Audio frequencies are too low to worry about "impedance mismatch". However, inefficient power transfer and driver (opamp) output overload is something to investigate.

    The amp input impedance is approximately 33KΩ so any decent opamp should able to handle that. Some opamps don't like to drive into a capacitive coax cable unless they are compensated for that kind of load. Are you driving a long coax cable?

    Happy listening,
    Ifixit
     
    malcolm likes this.
Loading...