John Curl's Blowtorch preamplifier

[Edmond Stuart]

Ovidiu, I'm waiting for the next edit.

Anyhow John understands my buffer.

PS: A simple complementary buffer is also feasible, although they have an offset voltage as no truly complements are available.

OTOH, if both offset are equal no problem at all because they are nullified in the next differential stage. In case they are not equal, the servo comes at rescue, that is, if the difference is not too large.

 

[john curl]

However, Edmond, at this time, it is the worlds most expensive buffer. It is elegant, but impractical. How can we get the same results, and maybe even better, with less?

 

[Edmond Stuart]

John, see my last edit.

 

[Edmond Stuart]

Okay, here's a simpler version. Now it looks rather conventional, except the servo, of course.
Offset is no problem if dual fets are used. THD figures are slightly better, although the 7th harmonic is a bit higher.

 

[syn08]

Okay, here's a simpler version. Now it looks rather conventional, except the servo, of course.
Offset is no problem if dual fets are used. THD figures are slightly better, although the 7th harmonic is a bit higher.

Now you are coming home... no more edits required. So zinsula and yours truly were right in assuming the current sources in your first schematic do basically nothing?

So far, nobody complained about the servo, it's very nice.

Though, an edit: On top of my head, you could replace the dual opamp with a diff in diff out opamp like OPA1632. Unfortunately it's not FET input, so it may pose some problems though.

 

[john curl]

So far, so good. Now, what if we removed the input servos and only used the amp servos? Saves money, parts. How about using 2SK170 and 2SJ74's, as they are cheaper too?

 

[Edmond Stuart]

Now you are coming home... no more edits required. So zinsula and yours truly were right in assuming the current sources in your first schematic do basically nothing?

Basically nothing regarding AC. However, I had added them for two reasons:
1. to zero the offset (btw there are no complementary jfets with Vos=0)
2. in an attempt to not deviate too much from the original design. IOW, I tried to look like it was an typical JC design.

So far, nobody complained about the servo, it's very nice.

Thx. BTW, a second op-amp might look overkill, although it doesn't, because it saves one large cap (1uF).

Though, an edit: On top of my head, you could replace the dual opamp with a diff in diff out opamp like OPA1632. Unfortunately it's not FET input, so it may pose some problems though.

I'm afraid not the most suitable op-amp for a servo, just like you said, no fet input, so too much bias/offset current.

 

[Edmond Stuart]

So far, so good. Now, what if we removed the input servos and only used the amp servos? Saves money, parts. How about using 2SK170 and 2SJ74's, as they are cheaper too?

The input servo is the amp servo. So better not to remove it.

Tomorrow I'll simulate the circuit with the other fets. Now it's bedtime.

Cheers,
E.

 

[Edmond Stuart]

.............
Just by simulating, I don't see much advantage for the "Edmond Buffer" even at fairly low load resistors.
But it makes offset trimming easier, as there is no need to screw around with the source resistors.

Nice idea.

Tino [/B]

Tino,

My apologies for my rather rude reply. You were right: the current sources doesn't make much sense. Moreover, the servo takes care of removing the offset (at least in this application). Apparently I overlooked an important feature of my own design!

Cheers,
E.

 

[Blues]

I'll put NP on the spot...

"Servos just aren't elegant. I much prefer to make the circuit
behave itself in the first place..." from http://www.diyaudio.com/forums/showthread.php?s=&postid=145516&highlight=#post145516

 

[lumanauw]

Hi, Edmond,

With little modification, the current sources can be usefull. Look at patent #4160216 Barry Thornton.

 

[SY]

Certain features look familiar. Here's a 1977 circuit with a rather... special n-FET. (from www.davidberning.com)

 

[Wavebourn]

Here is my concept of a buffer. However, it requires a coupling cap on input which JC probably would not like... If to use FETs for followers R1 and R3 may be huge.

 

[lineup]

what is really disturbing
to make this one really aestitic design, as those Nelson Pass and Me likes,
is that opamp
think it amplifies 'backwards' in some perculiar way

and how about PSRR with those R1/R3
directly connected between power supply and transistor base

naaahhh, this is not my cup of tea
i can feel bad vibrations coming up from my stomach
and the subjective sound i already have prepared my mind for
to translate from my ears to my mind impression
is not very promising

sorry, Wavebourn, you have done much better amplifiers than that
in my not so humble opinion
(but when we deal with audio prophets, we have to use some more strict criteria)

Lineup

 

[Wavebourn]

Where did you see PSRR?
It is an emitter follower with Vce and Ic kept constant. Just one possible implementation of the concept.

 

[Wavebourn]

One more version: stable current, but variable Vce. Also, combination of feedback and feed-forward is used.

 

[Edmond Stuart]

I'll put NP on the spot...

"Servos just aren't elegant. I much prefer to make the circuit
behave itself in the first place..." from http://www.diyaudio.com/forums/showthread.php?s=&postid=145516&highlight=#post145516

Hi Blues,

As I said before, this input circuit wasn't my first choice. See also: http://home.tiscali.nl/audio/
But in this special case, I just followed the philosophy of JC. If designed with care, both techniques (servo vs. big cap) perform equally well.

Cheers,
E.

 

[Edmond Stuart]

Please Wavebourn, specify component values, supply voltage inputs and outputs. Without these essential attributes, it's just not a schematic.

 

[john curl]

Wavebourn's schematic is also off topic.

 

[Edmond Stuart]

Hi, Edmond,

With little modification, the current sources can be usefull. Look at patent #4160216 Barry Thornton.

Hi David,

Thanks for the link. Interesting how B.T. solves the cross-over and 'hole storage effect' issue. However, I have a comment: he just moved the switching nasties from the BJTs to the diodes. But admittedly, it could help, in particular if fast (Schottky) diodes were used.

To stir the pot and OT: I disagree with this phrase: "bipolar transistors are essentially current operated devices".
Essentially, it is a voltage controlled current source (VCCS). In practice however, it is used as current controlled current source(CCCS) due to the limited current gain (beta). The latter is just a consequence of unwanted side effects like non-ideal emitter efficiency, base transport factor, depletion layer recombination etc.
Ideally, a BJT would behave as a kind of FET, practically, as FET with a sort of leaky gate. At least, that's my view.

Those who disagree are welcomed (our specialist Ovidiu?)

Cheers,
E.