John Curl's Blowtorch preamplifier part II

Status
Not open for further replies.
If that of Courage is allowed, then I hope I can post the circuit:

(if not, let me know, please)
 

Attachments

  • preamp.PNG
    preamp.PNG
    21.5 KB · Views: 744
Thanks PMA, that should be enough for intelligent discussion. The values are on target, and make sense. This is the closest approximation to the BT yet, and I suspect that it can be improved, as well, with a little constructive discussion. As you know, people have already taken the early schematics and made them into commercial products, so putting this here, isn't going to give away anything.
 
One question I have to this circuit is the gain balance. I guess the gain is set by the ratio of R20/R21 to R19.
Beyond maximizing dynamic range, I guess there's no special requirement to match R20 to R21, am I correct?

Fold the circuit up/down. Then fold again left right. You got a circuit with two transistors. The gain is approximatively G=(Rd/Rs)*gm*Ro where Rd=R5/4, Rs=R19/2, Ro=R21/2 and gm=gm4*4 Total is G=gm4*R21*R5/R19.

Done by visual inspection, so hope I didn't miss anything. Add the distortion cancellation common to all matched symmetrical circuits and you got the entire picture. Offset, drift, PSRR, CMRR are all horrible, that's the reason why the gain stage is 25% of the entire design (rest goes to servos, power supplies), you would also need 0.1% matched resistors. You also need a stiff metal box to keep the **** out of the common mode and keep the temperature more or less constant.

A substandard design by any criteria of the 21st century.

P.S. You could of course cascode the JFETs and get some decent PSRR, at the cost of some headroom.
 
Last edited:
A first thing I would do is to lower the R11 to R18 impedances. FET's have nonlinear capacitance and lower resistance seen by the gates are a good thing I think.

Depending on layout, 100 Ohm gate stoppers (close to the MOSFET gates) would be a good idea too

Tino
 
Last edited:
AX tech editor
Joined 2002
Paid Member
One question I have to this circuit is the gain balance. I guess the gain is set by the ratio of R20/R21 to R19.
Beyond maximizing dynamic range, I guess there's no special requirement to match R20 to R21, am I correct?

jd

I think that at least 1% tolerance resistors should be used, but perhaps this is an area where better matching might bring improvement. What do you think, PMA? I'm thinking of a small value bulk metal trimpot between the 1K loads and wiper to ground.

What kind of improvements could we then expect?

jd

Anybody going to answer my question, or are we continuing to jump all over the place in order to avoid a coherent discussion?

jd
 
I think that the question of perfect tolerance resistors or a trim, should be addressed by worst case analysis using a sensitivity matrix. Does Spice have such a matrix? Sorry, I know ECAP better in this situation.
Would you like to do the calculation, Jannman? Your 'guess' is as good as ours, at the moment.
PMA, I noted, just yesterday, in testing a CTC Blowtorch for distortion, that the 2'nd harmonic was only about 20dB below the 3'rd harmonic in magnitude. This is with single ended drive, and balanced output into a ST1710 THD analyzer. Of course, I would expect the ST analyzer's input impedance to be matched to about 1% for both inputs, and I would expect a typical amplifier to have the same sort of matching. In any case, it should be about 50 times less (typically) than the mismatch sensitivity of the 1K resistors. So it might be useful to add a trimpot between the 1K resistors. Bulk metal, mind you, not some cheap thing.
 
Last edited:
For everyone else, we are discussing a distortion specification 'improver' technique, not something that would be really obvious in listening. However, it is a start.
Also, please remember we are trying to discuss different approaches to amplifier design within the context of the first 2 stages. Details, such as gate resistors are deliberately left out, because they just complicate the schematic and confuse the discussion. Surely, everyone here interested enough in this exercise, has seen 'simplified' schematics put up on the chalkboard in engineering school, or in a textbook.
 
Status
Not open for further replies.