Gibson G-10 (solid state) Output Stage Problem

[Lee_ranaldo]

I have this old but good sounding G-10 Gibson solid state amp

By design this amp uses an Push/Pull Output Stage with one Germanium PNP power transistor (2N2147/48).

Since is not longer in production the replacement (2N2955 Silicon TO-3 Power BJT) do the job very well but the amp has a constant "distortion" in the background as "noise" that doesn't raise with the volume at all

My thinking says that it's cross over distortion since i've replaced or tested all the transistors and they all are ok.

The only thing , the 2N2955 used as a replacement gets warmer compared to tne 2N3055 used in the output stage.

Since the voltage of a GE juntion is 0,2 - 0,3 volts to start conducting and the SI juntion is 0,7 aprox. maybe it's a problem and i need to mod this output stage to make it work properly and symetrically.

I include the picture of the schematic (and old one values can't be seen correctly) if someone can guide me.

[Enzo]

You have to re-bias the circuit for silicon.

See that diode between the bases of the output transistors? Add another in series with it. Possibly could need a third. See how that does.

[Lee_ranaldo]

Thanks Enzo! you helped me before with a fender amp

i was thinking the same.

I'll have to push the juntion to stay "connected". I'm gonna try 1 or 2 1N4007 diodes in series and then a LED diode just to see if it works.

Thanks for your suggestion

If it works i'm gonna post the corrected schematic here.

Thanks!

[Lee_ranaldo]

Ok i was simulating the circuit with circuit maker

adding another diode in series with D1 (shown on schematic) just shuts the output stage off.

it works on Push Pull output stages with symetrical power supply (Vcc , -Vee), since this amp is configurated as a push pull but the power supply goes from GND (0 Volts) to VCC (Called V on the schematic) i don't know if it works with the same principle as any push pull design.

Any suggestions?

[Enzo]

The differnce between split power supplies and single supply is largely in the speaker connection - it needs a large cap in series to block the DC. What happened when you varied the bias control after adding a diode?

I added the diode to get the output bases approximately 2 junction drops apart to eliminate crossover. If other measures are also required, I'd have to think on it.

If the added drop is too much, try also adding a small pot, like 100 ohms, in parallel with the extra diode, so you can dial in less effect from it.

[Lee_ranaldo]

well after adding 2 diodes in series 1N4007
the "bias" control on Q7 didn't work. no sound at all. and then i had to replace the driver for the 2N3055 transistor (BC160) because i had almost 70 volts on the colector of this transistor (that means something was shorting out).

V is almost 31 volts on this amp.

the replacement i've used for BC160 is 2N2905 which has almost the same ratings as the original.

The output transistors didn't had any mica spacers and thermal grease on the sinks so i've added it. The amp has a constant distortion but when i raise the volume it acts almost "normal".

The sinks weren't touching the case at all. but that's the only problem i have , i still have cross over dist with 2N2955 and 2N3055.
i'm gonna search for a design of a output stage with this kind of power supply stage.

However i'm gonna mod as you suggest on the previous post.
and i'll post the results.

Thanks for sharing your knowledge
i really appreciate it.

[Lee_ranaldo]

news.

in simulations i've done on software adding another diode in series doesn't affect at all the output signal, only increasing the voltage that the PNP power transistor receives by an small amount of 1 volt or less.

in other designs the voltage needed for the PNP power transistor to be right biased is took from a resistor from the output (after the large cap), actually is like 2K2 or less, and that how this works from a single supply.

The diode D1 actually is replaced by a jumper.

it works the same on simulations.

the thing that affects it completely and the symetry of it is the 1M (390K on schematic) resistor from V+ before the first transistor that is connected to the bias circuit.

It changes the signal completely.

under simulations the max RMS output power is 15 Watts RMS on 8 ohms load.

i can include the simulation made on circuit maker if anyone is interested.

since circuitmaker doesn't have the equivalent for the 2N2955 power transistor i had to use an ECG equivalent (ECG398 if i'm right) that is actually a TO-3 150W PNP power transistor.