You said the heat sink isn’t getting hot. That might actually be the problem. TO220 packages are notoriously difficult to cool since the tab is so small, resulting in poor heat transfer.
Instead of feeling the sink, put your finger on the mounting screw. I’ve had situation where the sink is barely lukewarm and the screw burned my finger.
You also want to make sure that Q15 is mounted on the same heatsink at the output transistors. This will prevent thermal runaway of the bias.
Instead of feeling the sink, put your finger on the mounting screw. I’ve had situation where the sink is barely lukewarm and the screw burned my finger.
You also want to make sure that Q15 is mounted on the same heatsink at the output transistors. This will prevent thermal runaway of the bias.
Thanks 🙂 So its not bad at all actually. This is the response and phase response, squarewave at 10kHz 1 watt into 8 ohms. A few changes. You don't need R9
Q15 is mounted on the sink and also while i was checking how hot the heatsink is i check everywhere including the plastic tab and screws and it didnt really get dangerously hot? The transistor just shorted itself like that without a reason(or atleast i cant find one???) and the 2 times this happened the power output wasnt more than 15-20W so i doubt i overloaded it?
Thanks for the analysis Mooly! R9 is not in the schematic i wanted to check something in the simulation and forgot to remove it, but it isnt present in the actual circuit board. Also may i ask how you came with the values for R1 and R2?
The degen resistors on Q3 and 4 need to come down too, to get vce up a bit on the mirror. But that’s not going to cause anything to blow up.
The maximum average power dissipation in each transistor is about 8 watts, and that’s worst case (which is less than full power output, more like 2/3) into a sine wave. Ball park for music, turned up to gross distortion, is about 2/3 of that - or a little over 5 watts. TO-220’s can handle that. 20 is about the max real world they can take before you just can’t get the heat out of the package anymore, and you are just not there yet.
Peak current at low (about 5V) Vce will be roughly 3 amps. If you can handle half that into a zero crossing (vce=25) there won’t be any issue with second breakdown. Even de rated for temperature TIP41/2 can take the required amp and a half. (There are highly desirable fast types that CAN’T take this such as D44H10, and yes they blow up). As you approach full vce the current falls even more, and since the duration is so short (less than 1/4 of a cycle) you can effectively use the 5ms curve for that portion of it. 4 ohm loading looks a lot more iffy in that situation - it’s flirting with the edge of destruction. All of those old budget solid state receivers running around this voltage/power had warnings of “8 ohm minimum load” and they meant every word of it. With just 1pair of speakers hooked up they would run forever.
The maximum average power dissipation in each transistor is about 8 watts, and that’s worst case (which is less than full power output, more like 2/3) into a sine wave. Ball park for music, turned up to gross distortion, is about 2/3 of that - or a little over 5 watts. TO-220’s can handle that. 20 is about the max real world they can take before you just can’t get the heat out of the package anymore, and you are just not there yet.
Peak current at low (about 5V) Vce will be roughly 3 amps. If you can handle half that into a zero crossing (vce=25) there won’t be any issue with second breakdown. Even de rated for temperature TIP41/2 can take the required amp and a half. (There are highly desirable fast types that CAN’T take this such as D44H10, and yes they blow up). As you approach full vce the current falls even more, and since the duration is so short (less than 1/4 of a cycle) you can effectively use the 5ms curve for that portion of it. 4 ohm loading looks a lot more iffy in that situation - it’s flirting with the edge of destruction. All of those old budget solid state receivers running around this voltage/power had warnings of “8 ohm minimum load” and they meant every word of it. With just 1pair of speakers hooked up they would run forever.
The easy way for those emitter degeneration resistors 😉 they looked far to high at 330 ohm. I didn't try and calculate anything. Its all very tweakable and the current you are running the input pair at seems perhaps a bit low (R4 value will change that) but these are all things you can thrash out in a simulation.
Thank you for the reply but i didnt understood what you mean? What do you mean by "As you approach full vce the current falls even more"? by looking at the datesheet for the safe operating area at max of +25v it can safe operate with more than 2A of current wich is about the peak current draw from my 8 ohm speaker? Sorry if im sounding dump but also why do i need to look at the 5ms curve if its operated in the safe space? (again sorry i sound dumb but this is kinda confusing)
20W is a lot with poor heat transfer. But if the transistor package didn’t get extremely hot, then that’s probably not the issue.
I built a linear lab supply using TIP121 a while ago and I had a hell or a time getting them to release their heat to the sink. That’s why I brought this up. (I ended up using a TIP142 instead for better heat transfer)
As designed if the contact on the pot opens, the output current can suddenly rise to a very high level. The transistor could blow quickly, before any significant temp rise occurs at the tab or heat sink.
Its easy enough to move the pot from base to collector to base to emitter. Regardless of the cause of your transistor failure, this change should be implemented to rule out such a cause for destructive failure.
The current that is drawn by (or sourced from) the load will come down rapidly when the voltage switches polarity. With a resistive load, the current comes all the way down to zero when the voltage crosses through zero. With a fully reactive load it draws maximum current when the voltage is zero, but a fully reactive load cannot supply any power into the loudspeaker motor and hence no sound. Once that current peaks, it will come DOWN as Vce goes above 25V. And it’s duration will be short. Real world speakers are somewhere between the two extremes - and usually closer to the resistive case, so maximum current is not drawn an half supply voltage. Near loudspeaker bass resonance (but not at it) the impedance phase can be significant but it’s magnitude is also running high, which reduces the current drawn.
It’s a lot to take in, and to fully understand why this all happens requires college level math. Suffice it to say rule of thumb that works in most cases is to be able to handle half maximum peak current and full quiescent vce simultaneously. You might also be shocked to see just how much stuff there is on the market that violates this. And of course, full peak current at low vce must be dealt with, without running yourself out of hFE. But that just affects distortion.
It’s a lot to take in, and to fully understand why this all happens requires college level math. Suffice it to say rule of thumb that works in most cases is to be able to handle half maximum peak current and full quiescent vce simultaneously. You might also be shocked to see just how much stuff there is on the market that violates this. And of course, full peak current at low vce must be dealt with, without running yourself out of hFE. But that just affects distortion.
Power output is not dissipation. Maximum is vcc^2/(pi^2*RL). Real world with music is less than half. Even 8 watts per device won’t run the case temperature more than 25 degrees C above the heat sink temp - and that’s for a poor mounting.
Yes, I'm aware of that. But I know from experience that TO220 packages are terrible at dumping heat to the sink. But it doesn't seem to be the issue here, so it's a moot point.
I think we agree bulk heat is not an issue and neither is SOA. That leaves 3 possibilities. Oscillation (which would make the zobel resistor hot), impedance is lower than he’s letting on, or the TIP42’s are NFG. Once the first two are ruled out that only leaves the third.
so from what i understood everytime the sine wave reaches the half of vcc(or 0 if its dual supply) there will be a huge current draw for a short period and that damages the transistor since in this short period its running out of the soa?
The only way to draw the maximum current while crossing through zero is with a fully reactive load. A pure inductor or capacitor. If a speaker were to be fully reactive it can not produce sound at all. The power factor is zero, therefore no average power delivered. What really happens is it stores energy, and sends it into the output transistor on the opposite quarter cycle. Real world speakers draw about half their peak current (or less) at the zero crossing because it is somewhat, but not fully, reactive. If you account for this, it’s usually fine. Yes, weird speakers are exceptions. It does happen. But your typical woofer in a box, crossover, and voice coil type tweeter, no.
It probably IS a current draw issue. But it shouldn’t be if the TIP42 is any good. If you used at TIP32C it would likely die. Also distort like hell trying to reproduce a 3A peak, and having its gain fall to 10 at that 3A. If you used a D44H10, it would sound fine, and also die, because those are fragile as all hell (power handling takes a dump at 15V).
well at first i had 2 theories as of why it kept failing. 1st it was a design flaw and 2nd the transistors were fake. I cant really confirm how real the transistors are but its weird that i got both the 41 and 42 from the same place and only the 42's keep failing while i havent had any problem regarding the tip41c? Im not sure if the shop i bough them from gave me real ones but you said that if they were real it wouldnt be a problem so i guess maybe the problem is fake transistors?
Has this question been asked? Where did he buy the transistors? I did the back of the napkin math, and he should be within the SOA. Maybe not with a ton of margin, but good enough. If the answer to my question is "AliExpress", then that's THE answer!