If you replaced the output transistor in question look at the date and batch marks on the top face to see it this is different from others of its type. It is possible the driver transistor has been compromised in the melt down. I have already told you there are suspect solder joints on your board - too many for me to list.
If you took your amplifier into to a service workshop this would be the first thing they would attend to.
A couple questions re the schematic: What transistor types are Q1 and Q2? Am I guessing correctly that the R18 site can be either a 22K resistor or a 15V Zener? You've opted for the Zener? (I believe either would be ok.)
After scrutinizing the noise waveform in post 19, I'm less inclined to suspect a leakage/breakdown defect--- the glitch amplitude is too uniform, I think.
I'm suspicious of the differential input stage and its input bias: assuming a balanced stage, in round numbers, each Q1 and Q2 would be biased to half of the 3.75mA tail current. Assuming Beta of 150, base current would be about 12.3uA. That base current across R3 (33K) yields 0.407V at the base of Q1, a LOT of offset error to apply to differential stage. In contrast, the base sees R5 (820 ohm) as source resistance, 40 smaller.
So here's a quick experiment: install a temporary jumper between the C1-R2 connection and RCA ground shell, thus making R5 the bias resistance. Re-adjust R17 for 0V at the output and look for any evidence of misbehavior.
If we're lucky and this proves to be the problem, an appropriately chosen resistor from Q1 base to C11 B+ is an easy remedy.
After scrutinizing the noise waveform in post 19, I'm less inclined to suspect a leakage/breakdown defect--- the glitch amplitude is too uniform, I think.
I'm suspicious of the differential input stage and its input bias: assuming a balanced stage, in round numbers, each Q1 and Q2 would be biased to half of the 3.75mA tail current. Assuming Beta of 150, base current would be about 12.3uA. That base current across R3 (33K) yields 0.407V at the base of Q1, a LOT of offset error to apply to differential stage. In contrast, the base sees R5 (820 ohm) as source resistance, 40 smaller.
So here's a quick experiment: install a temporary jumper between the C1-R2 connection and RCA ground shell, thus making R5 the bias resistance. Re-adjust R17 for 0V at the output and look for any evidence of misbehavior.
If we're lucky and this proves to be the problem, an appropriately chosen resistor from Q1 base to C11 B+ is an easy remedy.
Q1/Q2 are MPSA18's, the diode is a 15V 1N5245B, and yes this diode is used instead of the 22K R18. There is a bit of a covid isolation situation going on where I am working on this. It's proving difficult to get things done. It might be a bit of time before I am able to work on this. I shall try that when I get the chance.
Sorry to hear of your precarious circumstances hope the situation resolves quickly.
The circuit diagram you posted shows R18 the 22k resistor with a 0.22u capacitor in parallel.
Resistors don't introduce as much noise as zener although the latter may have caused no trouble for most builders. The fact there is a resistor for R18 suggests there have been odd cases with zener noise issues and that for new builds the resistor is preferred.
Re the LTP Q1 and Q2 base bias under dc conditions is set by the value of R3 and R6 which are equal in value. R2 and R5 also equal in value are in series with blocking capacitors and come into the picture only for ac purposes.
The circuit diagram you posted shows R18 the 22k resistor with a 0.22u capacitor in parallel.
Resistors don't introduce as much noise as zener although the latter may have caused no trouble for most builders. The fact there is a resistor for R18 suggests there have been odd cases with zener noise issues and that for new builds the resistor is preferred.
Re the LTP Q1 and Q2 base bias under dc conditions is set by the value of R3 and R6 which are equal in value. R2 and R5 also equal in value are in series with blocking capacitors and come into the picture only for ac purposes.
Thank you, everybody is ok. It's just as a precaution at this point. Here is a link to my build thread. I did it quite a while back and don't remember some of the smaller details. I've read it through again.
Original Build Thread
There were quite a few options, most people told me to choose the zener diode. For some reason the build guide from the store is no longer active, that makes it harder for everybody to follow along with how this amp is supposed to be put together, and all the available options. I don't have the file on my computer any longer, but I do have a print out of the guide.
I'm afraid I must respectfully disagree. To accommodate bias current errors, each input needs to have equal DC source resistance. Please see:
https://www.analog.com/media/en/training-seminars/tutorials/MT-038.pdf
This may not be the culprit in this amp's misbehavior. Just a guess.
Best,
Steve
It you are talking about base input currents for Q1 and Q2 in the LTP the current gains of these may be different the adjustment trimpot R17 is there to change the resistance between the either half of the track as required to vary the levels of current from the tail reaching the emitters of Q1 and Q2. You can look up simulations to see what levels of Ib current are involved or you can look at this in terms of alpha (Ic/Ie) which equals beta/beta +1. beta of 150 is very low for a small signal transistor.
A device like MPSA18 is likely to be four or more times higher than that at say 600 so alpha would be 600/601 or almost 100%. Even a beta of 150 gives 99.3% alpha leaving less than 1% for base current.
I have looked at your thread briefly while a lot of members swear by the Dim Light bulb test it is possible that on power up the reaction the start up of a 600 VA toroid could have been violent enough to have caused instability resulting in your amplifier having the melt down - before the 22R safety resistors in the supply rails fused out. I'm assuming you checked by measuring resistances that all your power transistors were isolated from the heat sinks.
That thermal run away incident happened due to a few mistakes I made. One was that one of the output transistors didn't make good enough thermal contact, which caused the bias to fluctuate unstably. The second and most critical mistake was that my meter defaulted to ac instead of dc when I turned it back on. I then turned the bias pot to it's extreme because I wasn't getting the correct reading, causing the meltdown. One output transistor was dead shorted. I tested all the rest by running 12volt through a current limiting resistor through collector and emitter turning the transistors off and on checking for leakage. They all tested fine including the driver transistors, and I replaced only one; checking for one with a similar Vbe.
If the set up procedure was done without F1 and F2 in their holders R53 and R54 would have burned out and limited the impact. If the fuses were in their holders the situation is rather different - it is easy to be wise after the event. Is the dead short across the collector emitter junction of a PNP transistor and was there any short involving the base junction.
All your output transistors have 0.5W 2R2 base resistors which serve a dual purpose as safety fuses to protect the driver in the event of a base to emitter or base to collector short in any one of these, also providing an even distribution of base current to the output transistors.
In the case of the dead short transistor if this involves the base connection in any way, I would replace the 2R2 base resistor as the original part could have overheated and become noisy.
Last edited:
Q21 was a PNP output transistor, and as I recall all leads were shorted. Had to have involved the base. I'll check out it's base resistor and give that a try.
Problem seems to be fixed... after taking the board off 20 times and testing transistors and resistors to no avail, i just bulk replaced transistors after finally receiving parts. Can't say I know what the problem was except it was a small signal transistor. I replaced q1-q6 + q9. All noise is gone.
Thanks lol. I labled each transistor i took out and maybe will test each one some how when bored at some other time.