Bias question

[amr2]

Hello

I finished a rebuild of a Yamaha M80 months ago thanks to this site. Output devices, drivers, pre-drivers, bias transistors, bias pot, all diodes and some resistors have been changed, on both channels. Seems to work fine, but something related to the bias setting is annoying me, so I have decided to create a new thread.

The amp is biased to 10mV across the emitter resistors. When I turn the amp on R channel starts around 6mV, and L channel around 2mV. Both rise slowly close to 10mV in less than 5 minutes. The voltage remains quite stable over time without input signal.

If I play a sine signal (1KHz) using a dummy load, and stop it after a while, R channel returns to the nominal bias faster than L channel. It remains about 6mV higher, and slowly it goes down to the nominal value (1-2 minutes)

Perhaps any component that I have not changed needs to be revised, so I would like to understand why one channel starts higher than another, and what components of the schematic generates the initial drift when the amp is powered on. I have a digital oscilloscope to measure whatever. This is the schematic:

http://akdatabase.com/AKview/albums/userpics/10004/Yamaha M80 Schematic with TSB66.pdf

Any suggestions are greatly appreciated. I apologize for my basic English.

 

[petehall347]

i wouldn't worry about it as long as its stable at operating temp .prob just transistors and diodes warming and cooling at different rates because of where they are in the amp . make sure whatever tracks the temp. are in good contact with the heatsink

 

[gslikker]

maybe amplification difference in the bias circuit sensing transistor.

Having 2 millivolts in your case you might have measureable distortion, but this would settle very rapidly...now the dummy load thing is handy to see whether your amp can have a thermal runaway OR take away all the bias when running hot.

Myself when I refurbish an amp, I'll run it until the heatsink will be something like 55 degrees celcius (something like 130/140 fahrenheit), take the load off and see if bias is acceptable being around normal or maybe 1 or 2 times higher but not alarming.

Some old amps, like my Luxman SQ507X, had some inferior tracking by design.

 

[amr2]

OK, I'm going to do that test, thank you.

I forgot to mention a thing. All components changed are genuine, except the bias transistors (2SC1953) that are theoretically old stock, but I cannot be 100% sure. They are very similar to the originals, and the body is dark grey, not black like several counterfeits that I have seen. The original was 2SC1953-Q and these are 2SC1953-R and 2SC1953-S. I installed the R types, but one thing that I found strange is that the suffix “S” doesn’t exit in the datasheet:

http://www.semicon.panasonic.co.jp/ds4/2SC1953_BED_discon.pdf

An option could be change them for an equivalent, like the KSC3503D...

 

[ConradH]

The channels probably aren't exact thermal copies of each other, just physical layout and airflow issues. The transistors probably aren't perfectly matched for B-E voltage. I wouldn't lose any sleep over it if the bias is stable and recovers nicely from a hot condition. Bias can be a funny thing and it helps a lot to have the amp on a THD analyzer to see where the bias really needs to be, often not quite where you'd think.

 

[amr2]

Perfect. I have just removed the cover to track the bias voltage. I will see how it recovers from a hot condition and I will report back. What about using a hairdryer to heat up the heatsink? :scratch2:

On the other hand, this amp seems to be very sensitive to airflow. I can see how the voltage descend a few mV only blowing on the PCB!

 

[ConradH]

Being a bit impatient, I use a hot air gun all the time on heat sinks.


An electronic salesman walked in and enthusiastically said "heat sinks!" I threw him out because any fool knows heat rises.

 

[gslikker]

heat shrinks, too. sometimes, cold shrinks also ;-)

 

[amr2]

The hairdryer test:

- Hairdryer at minimum velocity, amplifier with no input signal. The little airflow that was going to the PCB was causing to drop the bias quite fast…until 3mV…then I stopped the test. The heatsinks were quite hot.
- Now the bias began to rise, until 12mV, to slowly drop close to 10mV

I don’t understand why very little airflow in the PCB causes to drop the bias ¿?

 

[petehall347]

its not the flow of air its the heat .

 

[amr2]

But it's strange that blowing air at room temperature it also drops. It's just a curiosity, I'm not going to put a fan on the top...

I will try to do a similar test with the dummy load.

 

[SoCal Sam]

When you compare the signal output from both channels, is there any difference? If none, be happy.

 

[amr2]

Do you mean comparing a sine wave from both channels on the oscilloscope? If so, no, there is no difference. I cannot measure THD, but the signal seems to be clean...

By the way, the bias have been adjusted without load. One channel has a slight DC offset, and there is a little difference if you leave the load connected.

 

[amr2]

I did some tests using a dummy load, and although one channel have a thermal drift slightly different from the other, I think that it's time to put the cover on and enjoy it. Thank you guys.