Overheated transistors at 14V

Thread Starter

mike_canada

Joined Feb 21, 2020
239
I did end up upgrading my speakers to higher-watt capacity and better sound. I tested this particular circuit with about 6V and it worked well. Then I tested it with 14V and about 2 minutes in powered at 14V, something smelled like it was overheating yet the music was still playing. It turns out the transistors were very hot even though the TIP41's were connected to heatsinks via nuts and bolts. I then reduced the voltage back to 6V and the circuit worked again. Only thing I may need to change is the LED resistors and/or have the music always cranked up for the LED's to respond to the music.

Is it normal for heatsinks to be so hot? I don't think the smell was normal. It almost smelled like burning plastic.

b.png
 

Thread Starter

mike_canada

Joined Feb 21, 2020
239
I was playing with some parts and was lowering voltage so unfortunately I don't have the original spice file anymore. But I attached the next equivalent file.
 

Attachments

Audioguru again

Joined Oct 21, 2019
5,183
Q5 has no series base resistor to limit its base current. The current must be massive.
Then Q1 has a high collector-emitter voltage across it and the massive Q5 base current in it causing heating.
Q2 is trying to drive the base of Q5 to a negative voltage much more than is allowed, due to the reverse breakdown voltage of the emitter-base junction of Q5, causing heating in Q2.

Therefore the base of Q5 must be fed the signal through a voltage divider and a current-limiter.

If Q4 and Q5 are replaced with an opamp then the sensitivity can be much higher.
 

Thread Starter

mike_canada

Joined Feb 21, 2020
239
Q5 has no series base resistor to limit its base current. The current must be massive.
Then Q1 has a high collector-emitter voltage across it and the massive Q5 base current in it causing heating.
Q2 is trying to drive the base of Q5 to a negative voltage much more than is allowed, due to the reverse breakdown voltage of the emitter-base junction of Q5, causing heating in Q2.
I don't understand how the voltage and current is massive at Q5. Maybe spice doesn't either. And this is with the input audio set at 5Khz frequency and 0.4V amplitude which is close to normal listening level (according to wikipedia).

I didn't check Q2 right after circuit operation. Just Q1 and Q3 only. Maybe first pull-down resistor R11 needs a better value (again) for better biasing.
 

Attachments

Thread Starter

mike_canada

Joined Feb 21, 2020
239
and don't transistors have some resistance? like Q1? I mean, sure R1+R2 can be the external current-limiting resistance but isn't there some sort of resistance even when the transistor is turned on like maybe in the order of ohms?
 

sghioto

Joined Dec 31, 2017
3,416
I suspect the bias current maybe too high on Q1 and Q3 when using 14 volts. Try adding a resistor labeled Rb in series with D1 and remove D8 as I show in the modified schematic . I would start with a 100 ohms and increase as needed. The idea is to reduce the quiescent current through Q1 and Q3 with no signal. Normally this current is set to just above what is required to eliminate crossover distortion in the output.
1663722246806.png
 

crutschow

Joined Mar 14, 2008
30,804
isn't there some sort of resistance even when the transistor is turned on like maybe in the order of ohms?
Of course, but the transistor is not fully on, it is only on enough to generate the speaker voltage for the music being played (it acts sort of like a variable resistor between the supply and the speaker).
Thus the transistor dissipation is largely determined by the speaker current times the difference between the supply voltage and the speaker voltage at any instant in time (ignoring the dissipation from the bias current as mentioned by sghioto).
 

Audioguru again

Joined Oct 21, 2019
5,183
I don't understand how the voltage and current is massive at Q5. Maybe spice doesn't either. And this is with the input audio set at 5Khz frequency and 0.4V amplitude which is close to normal listening level (according to wikipedia).

I didn't check Q2 right after circuit operation. Just Q1 and Q3 only. Maybe first pull-down resistor R11 needs a better value (again) for better biasing.
5kHz is a very high frequency where the capacitance of C1 passes some signal to the very low resistance of the base-emitter junction of Q5 (your highpass and lowpass filter capacitor values are completely wrong). Since you have no series base resistor to limit the base current in Q5 then the base current when C1 is charging is massive and heats Q1.
When the signal in C1 goes low it causes another massive current when the emitter-base junction of Q5 has Avalanche breakdown and heats Q3.
Q5 is probably burned out.

Your texts are mixed up. R7 is the DC pull down resistor and R11 is the series input resistor that helps set the AC gain.
 

Thread Starter

mike_canada

Joined Feb 21, 2020
239
You have the very low capacitance of C1 driving the very low base resistance of Q5, then Q5 gets no signal.
I increased the capacitance of C1 and reduced the frequency.
The transistors past C1 make up part of the VU meter. I have that kind of circuit duplicated 2x (3 units altogether). One of them has a series capacitor that is much higher to attract the lower frequencies. I may play with capacitor values such as the one you mentioned later down the road.

Thus the transistor dissipation is largely determined by the speaker current times the difference between the supply voltage and the speaker voltage at any instant in time
Ok in all tests, the speaker is 8 ohm. In that failed test, the voltage is 14.4V.

So I guess that means 1.8 amps at the speaker output when music is going which means 25.92 watts.

If I'm not mistaken, the TIP transistors can handle up to 40-65 watts with heat sinks attached. then again I didn't use heatsinks on the PNP's. When I tested everything on 6V, nothing smelled like it was smoking.

As per Audioguru's circuit, the input resistor i had for R7 is 4.7K and R11 initially was 22K.

I'm probably better off just dropping the voltage to 7.2V and changing the audio input transistor (Q2) to 2N3904 because when I experimented with that in spice, I saw a perfect amplitude (of the output audio signal to the speaker) going from -3V to 3V whereas a Pn2222 only allowed the amplitude go from -2.5V to 2.5V. I bet its because the max VBE value of the 2N3904 is much lower than that of the Pn2222.
 

Thread Starter

mike_canada

Joined Feb 21, 2020
239
Ok so I experimented with that capacitor in spice to make it in the uF range and that emitter voltage went past 6. I think I'm better off just to lower my supply voltage at least for the time being.
 

Audioguru again

Joined Oct 21, 2019
5,183
The transistors past C1 make up part of the VU meter. I have that kind of circuit duplicated 2x (3 units altogether).
All 3 circuits are severely overloading the audio output transistors because their bases have nothing to limit their current, as I have been saying over and over.

I guess that means 1.8 amps at the speaker output when music is going which means 25.92 watts.
No. The max peak current in the speaker is almost 7.2V/8 ohms= 0.9A. Then the max RMS output voltage is 14.4V/[2 x (the root of 2)]= 5.1V. The speaker power is 5.1V squared/8 ohms= 3.25W and since the diodes at the bases of the output transistor cause them to operate in class-AB then each output transistor heats with 3.25V/2= 1.6W only when a continuous tone is playing at barely clipping, not when playing music that has various lower levels.

I'm probably better off .....
No. Don't the output transistors not overheat when the overloaded VU meter transistors are not connected? Then add base resistors to the VU meter input transistors to limit their base current and base voltages.

6V peak-to-peak when using a 7.2V battery produces a max low distortion output power of only 0.56W into 8 ohms.
 

Thread Starter

mike_canada

Joined Feb 21, 2020
239
I probably have burnt one transistor because the LED isn't responding. then again when I measured the transistors in-circuit, with a diode meter, it seems the pins are OK because the meter shows a number in the hundreds one way and infinite the other way. Then again I wonder if I'm using poor capacitors (not accurate enough ones).

At 7.2V nothing is hot. The TIP4x transistors are warm-ish but not hot. then again I'm not at max volume. Maybe volume 1/10 on a 10 bar volume setting.
 

sghioto

Joined Dec 31, 2017
3,416
At 7.2V nothing is hot. The TIP4x transistors are warm-ish but not hot
With that data I'm still convinced that the bias is to high. Have you tried running the modified schematic in post #7 through Spice?
Typically the current through Q1 and Q3 would be around 10 to 20 ma with no signal present.
 

Thread Starter

mike_canada

Joined Feb 21, 2020
239
Typically the current through Q1 and Q3 would be around 10 to 20 ma with no signal present.
I disconnected the signal but at 7.2V the current is 99mA and at 14.4V the current is 318mA. This was measured in spice and that's with the two diodes still present. That just made me think that the 2n3904's were BAD for VU portion of the circuit because I thin they can handle a max of 200mA and I was probably driving 318mA into them through capacitors when I used 14.4V.

I also fixed my problem with the LED. it turned out I had the capacitors in the wrong spots.
 

Audioguru again

Joined Oct 21, 2019
5,183
The 2N3904 transistor has a maximum allowed collector to emitter current of 200mA. Its base-emitter current is always 1/10th (20mA) or less than its collector-emitter current.
Your circuit is missing a resistor in series with the base of Q5 to limit its peak base current to 20mA (in your circuit the base current is very high because C1 in series with the base-emitter of Q5 are a dead short circuit until C1 is charged.

Your circuit is also missing a resistor or two to limit the negative-going base-emitter voltage of Q5 from having avalanche Breakdown at a very high current.
 

Audioguru again

Joined Oct 21, 2019
5,183
The heating of the output transistors will be reduced if the two diodes are in thermal contact with them.
When the output transistors get warm then they conduct more current which makes them hotter.
When the two diodes get warm then they cause less current and less heating in the output transistors.
 

sghioto

Joined Dec 31, 2017
3,416
I disconnected the signal but at 7.2V the current is 99mA and at 14.4V the current is 318mA.
That's about what I suspected the standby current is way too high at 14.4 volts. The problem is the two diodes, you need to remove one and replace with a resistor as I suggested. Start at 100 ohms and adjust as necessary to get the current around 20ma.
A complimentary output such as this with the TIP41 and 42 only require one diode to set the bias but sometimes a resistor is added to allow trimming the bias voltage. The TIP41 and 42 are bipolar transistors not darlingtons which would need two diodes in series.
That just made me think that the 2n3904's were BAD for VU portion of the circuit because I thin they can handle a max of 200mA and I was probably driving 318mA into them through capacitors when I used 14.4V.
Q5 has nothing to do with the outputs getting hot. Q5 is isolated from the output by C1 and there is no DC current through C1.
 

Audioguru again

Joined Oct 21, 2019
5,183
2 diodes are common in low power class-AB amplifiers. Maybe cheap out-of-spec diodes were bought from ebay or AliExpress.
The peak current in C1 and the base-emitter junctions of the output transistors is massive and is unlimited and must be reduced.
 

Attachments

Top