a couple of questions on diamond buffer

[Stefanoo]

Hi Folks,

I have attached the schematic i am currently running.
I really like how it performs.
Bandwidth is over 1MHz (i will precisely measure it later on).

In order to not have any overshoots the output transistor's bases as to be directly tight to R145/146.
With this set up the AC performance is great.

I have a couple of questions:

1) Is it true that this buffer ONLY sounds good if it's heavly biased? and if so, what is the suggested current at the output?

2) I don't know if there are any mistakes on the schematic i have drawn, but the input sits at +11V istead of 0V and so does the output. Do you guys have any suggestion with this regard?

Thank you very much for the attention.


Best,

[Stefanoo]

this is the schematic

[aparatusonitus]

Originally posted by Stefanoo
2) I don't know if there are any mistakes on the schematic i have drawn, but the input sits at +11V istead of 0V and so does the output. Do you guys have any suggestion with this regard?

Place 1k in series and 47k to ground at input transistor bases...but some tens of uV of output DC offset will remain still.

[CBS240]

1) Is it true that this buffer ONLY sounds good if it's heavly biased? and if so, what is the suggested current at the output?

That makes sense because class B (AB) bias will have some crossover distortion, so for best perfomance would be class A, no?

[ostripper]

After foolin' with the diamond for a while, unless the circuit is
implemented foolishly , it will always run class A , even at
very low currents.

As shown in the attachment below, it is always conducting
regardless of signal or current. The only issue about
running at low currents is it's ability to drive the Vbe and
drivers and it becomes a little bit non-linear. this
would depend on the choice of devices and biasing scheme.

I tried both CCS's (like in attachment) ,and simple Re's with
the diamond. With Re's above 10K (instead of the CCS's)
the diamond would stop conducting at high output.

Steve dunlop's "krill" and the nakamitchi 620 both use the
current sources and in simulation this makes for very
good foolproof performance.
OS

[Lumba Ogir]

Stefano,
I like the way you`ve done this one. I would use 2SC5171/2SA1930 (instead of those ugly MOSFETs).

[Stefanoo]

Hi guys,

thanks for your posts.
The Diamond buffer works great, wide bandwidth and great linearity.

A very stupid question.
The diamond will be used on my phono stage and therefore the buffer has to drive a preamplifier.
The question is: what type of output devices would feet better the application?
I am currently running them for debugging with general purposes TO92 BC550/560C.
I mean: do I need TO220 devices and high bias, or i can just Bias as much power as the TO92 can take (wich is about 300mW)?

Usually low power transitors are faster and less noisy then medium power ones.

But my questions comes from the fact that as far as i have seen things around, they all come with TO220 output devices to drive the external world.


In case do you guys have any suggestion with regard to the semiconductors?


Thanks

[Lumba Ogir]

Stefanoo,

The diamond will be used on my phono stage and therefore the buffer has to drive a preamplifier.

Most importantly, it has to drive the RIAA network, right?

do I need TO220 devices and high bias

I say yes.

[Stefanoo]

Originally posted by Lumba Ogir
Stefanoo,

Most importantly, it has to drive the RIAA network, right?

I say yes.

humm..why would an high impedance network as the RIAA network constituite a problem for the output stage?
Currents demands are really low.
The only think i am thinking is if a capacitive load, as the RIAA network on the feedback loop, would cause any issue, but i can't see it happening.

What drawback would I have by using the TO220 devices as the audio toshiba devices 2SC2240 and 2SA990 (I think i got the name right... )?


P.S. to answer to OS: the output stage runs in class A depending upon the Load.
I have notices that with a 10mA CCS and with a load of 100ohm the stage is still in class A but with a little higher load it switches to Class AB, but that will never happen since it will drive high impedance i guess.

[Lumba Ogir]

Stefanoo,

The only think i am thinking is if a capacitive load,

Exactly, current is needed, say at least 15mA, depending on heat sink.
You could use 2SC2240/2SA970 for Q45/48, 2SC5171/2SA1930.for Q46/47.