This is my PCB layout, which from Plamen's posts, I can already see has a lot of issues.

Unfortunately the LTspice component names don't match this PCB, but the layout is the same. I attached another schematic where the part names are the same for reference.

Petkan:
After setting the idling current do you still overheat?.

I will re-test later today with all your recommendations, and see if the temperature continues to rise with just idle current.

 

Linear amplifiers have low efficiency like 30% and less. Heat dissipation rises with output power.

Internal heat dissipation increases with output power, but not at a fixed relationship. The output stage efficiency is worst when the amplifier is producing about 40% of its maximum output power. As the output power increases above that, so does the efficiency. This is an indirect manifestation of the Maximum Power Transfer theorem.

ak

 

Petkan:
gradually increase the input level while observing the output waveform.
If the output follows the applied input undistorted keep increasing the level; (so far unloaded) until you see slipping.

What would be a safe level to expect the input signal to be?
I had assumed 5vpp should give me some room if the intended input is to be from a CD player?

Or should I calibrate lower at 1 or 2vpp?

Thanks

 

Petkan:
apply input sine initially at very low level and observe by scope the output.
Keep monitoring the consumption and gradually increase the input level while observing the output waveform.
If the output follows the applied input undistorted keep increasing the level; (so far unloaded) until you see slipping.

Here's what I tested so far tonight with NO LOAD on the output other than the scope;
Magenta - Positive Rail
Cyan - Negative Rail
Yellow - Input Signal
Dark Blue - Output Signal

1kHz Input Signal
From 1vpp to 3vpp I was unable to clip; Each rail drawing 37mA:



At 4vpp I was able to start clipping; Each rail drawing approx 40mA


Next I ran higher frequencies on the input to see how it looked:


I believe this means my first few stages, though not great, are functional correct?
Am I correct in thinking that this range (up to 5vpp) would safely cover the input signal from a CD player?

Tomorrow I will attempt to add a load to the output and start verifying some currents.

Thanks again for all the replies, it is a big help!

 

What would be a safe level to expect the input signal to be?
I had assumed 5vpp should give me some room if the intended input is to be from a CD player?

Or should I calibrate lower at 1 or 2vpp?

Thanks

Petkan:
The gain of the amplifier is set by the resistor divider in the main negative feedback (Gac=1+R3/R2).= 1+100k/10k = 11.
The DC gain is only 1.(caps removing R3 at dc.)
Clipping point is defined by power rails headroom. (+-35V). The output peak peak value cannot exceed the power frame.
Divided by ac gain we can estimate the max input voltage level corresponding to clipping point.
With +-35V rails, (considering some limitations to closeness to rails) most likely the output cannot exceed say 33V peak corresponding to 3Vp at the input or 2Vrms (5V is too much indeed).
Once we confirm the output can swing close to the rails, we can go back to low level to check for distortions due to insufficient biasing.
If the main gain setting divider defines too little gain - distortions may occur in the input stage due to over-driving. (not in this case).

Capacitors C9, C11 directly across the output (no ballast resistor) are a surprise to me. The network to the right (series RC) is commonly used to compensate for the predominantly inductive nature of the speaker as a load (resulting in phase shift between output voltage and current, compromising stability).

 

I will attempt to remove C9 and C11 today and re-test. Those capacitors were added during my breadboard testing stage to help remove oscillation I was seeing at the output. They're probably no longer required since being built on a PCB.

Thank you

 

What I also found interesting was that when running at 30-50kHz my output transistor heatsinks were getting warm even without a load connected to my output.

I wonder if I'm getting some type of inductive heating from the output inductor being mounted too close to my heatsinks on the PCB?

 

What I also found interesting was that when running at 30-50kHz my output transistor heatsinks were getting warm even without a load connected to my output.

I wonder if I'm getting some type of inductive heating from the output inductor being mounted too close to my heatsinks on the PCB?

Petkan:
At high frequencies the output current flows into C9, C11 and to some extent to the RC network to the right.
Although there is no external load, these caps act as internal load.

 

Petkan:
At high frequencies the output current flows into C9, C11 and to some extent to the RC network to the right.
Although there is no external load, these caps act as internal load.

So there should be no harm in removing C9 and C11 correct? It should actually help if anything?

 

Petkan:
At high frequencies the output current flows into C9, C11 and to some extent to the RC network to the right.
Although there is no external load, these caps act as internal load.

Also note that bandwidth is not defined at high level. Amplifiers have limited slew rate (V/us) i.e. the max frequency declines with level.
Typically frequency bandwidth is defined at lower level (-20dB). An audio amplifier is not meant to operate a 20kHz sine for long time, let alone 50kHz. High frequency spectral content in the music is way lower compared to 500Hz to 5kHz band.

 

So there should be no harm in removing C9 and C11 correct? It sh

There is great harm leaving them in. Amps with negative feedback, like yours, tends to be unstable when presented with reactive loads.

As a matter of fact, a typical torture test for high quality amps is to connect a capacitor, typically less than 2uf, to the output and see if the amp oscillates. Most amps are unable to pass that test.

By putting those two capacitors in, you are essentially torture testing your amp permanently.

It is airacle that it hasn't disintegrated.

 

By putting those two capacitors in, you are essentially torture testing your amp permanently.
It is airacle that it hasn't disintegrated.

The output transistors have disintegrated...twice
Thanks for the info, I will be removing them shortly.

I'm seeing that there is a huge difference between breadboarding with wires all over the place and using a PCB for construction...

Thanks again

 

The output transistors have disintegrated...twice
Thanks for the info, I will be removing them shortly.

I'm seeing that there is a huge difference between breadboarding with wires all over the place and using a PCB for construction...

Thanks again

Petkan:
You can put that last line in bold. Layout is serious and delicate matter. Minor layout flaws can result is serious misbehaviour

 

So after many months, I've finally assembled this into a case and thought it was finally complete. I set it up and started a listening session and I've yet again blown up the output stage

I have removed the capacitors on the outputs and upgraded to Aavid Thermalloy 530002B02500G heatsinks (2.6 deg C/W) and changed the emitter resistors to 0.68 ohms.

I've attached the final schematic.

I'm still not sure what I'm exceeding on the TIP41/42 but the metal tab measured 35 deg C when she blew up this time (using a thermocouple).

I am considering replacing all the TIPs with FJP5200 / FJP1943 transistors, or MJE15028 / MJE15029. Hopefully that will help.

Can anyone tell me if there is any other major problems with the design than can be blowing up the output stage, or is it just too small of heat sinks? I'm really hoping this design can be saved, since I've already built up the PCBs and a custom enclosure, but maybe it has to be scrapped all together?

Please help,

Thank you!

 

Here's some pictures to show the board layout and give an idea of the heat sink physical size:

 

You have a HUGE problem. It looks like you have no thermal feedback for the bias regulator.

But first, there are good an bad reasons for having fuses in each rail. I have them in my amp, but there is a circuit that if one rail goes down the other one does too.

First an foremost, put an AGX very fast acting style fuse in series with your load (Speaker).

You have two basic figures of merit
1) DC offset 0 usually comes out in the wash (-+0.75V or less,my amp about +-10 mV or less
2) DC bias - thermally compensated
3) The voltage between the bases of Q8 and Q9. This voltage has to change as the output transistors heat up.

Q7 is a vBe multiplier, but how many Vbe's do you have? . Look at adding some thermally mounted diodes in the regulator for thermal compensation. Three, plastic packages, are also easy to mount. Two holes, for two diode bodies, press fit, and one on the outside.

The best way to set DC bias is a distortion meter. If you don;t have one. Idle the amp, no load. Monitor the current across an emitter resistor. If set too high, the current will run away.

I made version of The Leach Amp in the 1980's: http://leachlegacy.ece.gatech.edu/lowtim/

 

So should I mount Q8 and Q9 off-board and maybe epoxy them right to the output transistor heat sinks? Would that provide the thermal feedback you mentioned?

 

So after many months, I've finally assembled this into a case and thought it was finally complete. I set it up and started a listening session and I've yet again blown up the output stage

I have removed the capacitors on the outputs and upgraded to Aavid Thermalloy 530002B02500G heatsinks (2.6 deg C/W) and changed the emitter resistors to 0.68 ohms.

I've attached the final schematic.

I'm still not sure what I'm exceeding on the TIP41/42 but the metal tab measured 35 deg C when she blew up this time (using a thermocouple).
I am considering replacing all the TIPs with FJP5200 / FJP1943 transistors, or MJE15028 / MJE15029. Hopefully that will help.

Can anyone tell me if there is any other major problems with the design than can be blowing up the output stage, or is it just too small of heat sinks? I'm really hoping this design can be saved, since I've already built up the PCBs and a custom enclosure, but maybe it has to be scrapped all together?

Please help,

Thank you!

Your output transistors have ample margin on voltage and current. On power dissipation you monitored the temperature and did not see thermal run away. Typically BJT transistor fail short collector emitter (or excessive leakage between them). However if short B-E junction is seen (without C-E short) it may point to exceeding reverse B-E voltage. Your output stage is dual stage emitter follower (unity gain). The only way to exceed reverse B-E voltage is if the emitters are held by capacitors while the base is declining fast. This does not happen during listening music, where no high level high frequency is applied. This circuit has no current limiting and shorted load may be a risk. Can you examine your speaker cable/connectors for potential for shorts.

 

The TIPs failed with Collect to Emitter shorted this time. One of them actually split in half when it let the smoke out.
There isn't a short in the cable or connectors. I verified nothing was touching or shorted, and the cables and speakers are fine when connected to a different amp.

If it adds any info, R16 was burned and failed open at the same time.

Thanks

 

I just hooked up one of the TIP41C transistors (that did not fail) to a DC power supply and run 15W through it for an extended period of time. The temperature of the metal tab settled at around 98 deg C and the heatsink settled around 90 deg C.

I thought for sure it was that the heat wasn't getting dissipated fast enough with these heat sinks...