Sony TA-8650 bias adjustment

[MattiaM]

Since there is no unique source of information on the bias for the Sony TA-8650, with no mention of changes from the 125mV in the original Sony Service manual in any service bulletin, and rumors from a Japanese forum of a required new value of 30/40mV, I decided to do a quick and by no means scientific experiment to see if I could spot any differences between a bias set to 35mV and 100mV (125mV make my amplifier way to hot for me to feel comfortable).

I set the segnal generator with a 1kHz signal at 800mV peak to peak directly into the power amplifier, and these are the results from the output (with 8 ohm load):

- there is no visible crossover distortion even at 35mV bias.
- there is a slight difference in output, from 9.27Vrms with bias set at 35mV to 9.47Vrms with bias set at 100mV.
- The second order harmonics seem to be slightly more pronounced at 100mV.
- Distortion from third order harmonics appears to be slightly lower with the bias set at 100mV.
- At 35mV bias the amplifier runs MUCH cooler.
- At 35mV (preamp connected), the sound seems to me to be slightly more muddy, but it might just be my own bias...

(Out of curiosity, I also tried to see how a square wave would be amplified, knowing that they're quite hard on audio amps... Piece of cake for the Sony).
(IMG_4243 and 4241: 35mV; the other 2 100mV)

 

[elnaldo]

Do you see any reason to bias the amplifier at 100mA instead of 35/40mA?

 

[MattiaM]

I am still not entirely sure; to me the sound seems better at 100mV. The drawback there is much more heat, which in the long term will likely shorten the life of the amplifier. When I have time I will check the entire frequency response with the two bias settings and also the spectrum at low and high frequencies, to see if that adds to the story.

 

[elnaldo]

You could try a colder point, like 60mA, less heat, perhaps similar performance.

 

[rottalpha]

I am still not entirely sure; to me the sound seems better at 100mV. The drawback there is much more heat, which in the long term will likely shorten the life of the amplifier. When I have time I will check the entire frequency response with the two bias settings and also the spectrum at low and high frequencies, to see if that adds to the story.

 

[dr*audio]

Best way would be to measure THD.

 

[MattiaM]

Alright,
I finally got some time to play with the bias of the TA-8650.
I set the bias at 40mV and 100mV, the preamplifier disconnected, with a sinewave input directly into the amplifier to produce an output of approximately 10W rms.
I checked the frequency response in a 5Hz-50kHz range, and the THD at 500Hz, 1kHz and 10kHz (avg. of 50 values).

THD @ 100 mV bias: 0.037% (500 Hz), 0.024% (1 kHz), 0.036% (10 kHz).
THD @ 40 mV bias: 0.046% (500 Hz), 0.031% (1 kHz), 0.044% (10 kHz).
There is no visible difference in the frequency response.

Does 100mV sound better? To me yes. As I stated in a previous post, 40mV in my opinion sounds more muddy, which could be the consequence of the higher distortion across all frequency, although we can argue whether or not the human ear can hear such small differences. To me, these results reinforce the feeling that at 40mV the amplifier just didn't sound as good as it could.

Hope this helps whoever is working on this amazing amp!

(40mV 1kHz, 40mv 10kHz, 40mV 500Hz, 40mV f. resp; 100mV 1kHz, 100mv 10kHz, 100mV 500Hz, 100mV f. resp)

 

[z900nl]

Wow I just got an 8650 on my own, asking myself the same questin as your starting post. I set it to 80mV for now (was just a feeling). I got the 8650 for a couple of days now with the 5650 and 4650 allready in the house.

 

[moodsman]

thanks a lot MattiaM
These measurements confirm what I preach for years, those V-Fets are no simple transistor amplifires

V-Fets sound like a tube. So they need a high propensity to function properly, and they should be as hot as tubeamps
a cold v-fetamp had a malfunktion or is set incorrectly.
I think it's not good to reduce the bias to 40mv, as the V-Fets could possibly die, and this is the worstcase to these Amps.
I use a TAN-8550 since 1986 with 125mV bias and only had one problem in all that time.