Diamond Output buffer eats trannies (thermal)

[Andypairo]

Hi Chris,
I will try to simulate how much does current in the input stage change at loud levels (20Vrms), to see if they do get starved.

As I supposed even a 20V signal doesn't change that much the current of the input bjts (fluctuation is less than 20uA and they have nearly 2.5 mA)

Will try replacing with matched ones.

Cheers

Andrea

 

[anatech]

Hi Andrea,
You're lucky, my configuration starved the inputs. I solved it.

-Chris

 

[Andypairo]

Hi,
I swapped the input bjts with reasonably matched pairs and now offset is much lower.
It seems that I overlooked the importance of using selected components.

In any case all the credits for this interesting project go to Mtlin12

Cheers

Andrea

 

[Andypairo]

Listening session

I had a brief listening session and I am pleasantly surprised from how it sounds.

At first glance it doesn't impress much but then you find yourself listening to things you never noticed before.
Choirs seem much less spatially compressed and different voices singing together are now more recognizable. There is also a very enjoyable sense of rythm.

It sounds pretty different to what I am used to, in a few days I will be able to give more impressions.

Cheers

Andrea

 

[anatech]

Hi Andrea,

It sounds pretty different to what I am used to, in a few days I will be able to give more impressions.

It only gets better. Detail will be very good. Large dynamic signals are the weak spot.

-Chris

 

[Andypairo]

Do you mean that it doesn't handle large signals well (or just that the difference form a "normal" amp is smaller?

Cheers

Andrea

 

[anatech]

Hi Andrea,

Do you mean that it doesn't handle large signals well (or just that the difference form a "normal" amp is smaller?

I found it was an odd effect. Almost like it was current limiting dynamically. This occurred with heavy dynamics. Otherwise it was very fast with incredible detail.

SY got a chance to listen to it briefly while it was grafted into a Counterpoint SA-100 chassis and voltage amp stage. I'm still not happy with the combination although the power diamond buffer shows some promise.

-Chris

 

[Andypairo]

May I ask which schematic are you employing?

And then, with heavy dynamics you mean loud listening levels (power?) or something else?

Cheers

Andrea

 

[anatech]

Hi Andrea,
Think of a standard diamond buffer. Use a darlington connection at both ends with speedup resistors (also increases current in the input transistors). Each polarity output is a matched pair also. I find that adding more output pairs cuts down on delta I and sounds better. The bias level is around 100 mA per output device. The standard resistor position is using a current source with red LED bias.

The input impedance is very high and can be driven with a tube follower easily. Clamp diodes to each rail comprise the protection for the input. It's capacitively coupled with a relay shorting the input to the buffer until the tubes settle down.

Performance also suffers at lower load impedances. I've run it into 2 ohms.

And then, with heavy dynamics you mean loud listening levels (power?)

Yes.

-Chris

 

[Andypairo]

I think that a schematic with component values would really help.

Which rails are you using? I don't need high power so I settled on +/- 33V, that should be enough for 50W/8 Ohm (my speakers are 4 Ohm). Bias is about 200mA (but 30 flow in the drivers).

Cheers

Andrea

 

[anatech]

Hi Andrea,
My schematic is a moving target as I mess with it. Nothing secret and very basic. It is as I described. My rails for experimental purposes right now run from 35 VDC to 50 VDC. Once I have something I like it will be expanded to a 70 ~ 80 VDC version as well.

I don't have anything drawn up in an electronic format for it.

-Chris