John Curl's Blowtorch preamplifier

Status
Not open for further replies.
Jump to Latest
#17,041
Re: Re: Re: Re: Re: Re: Re: VAS fighting

Bob Cordell said:


For BJTs the mismatch will generally be less than 1 mV, while for JFETs it will generally be less than 5 mV. In the example I showed, a 1 mV mismatch in the BJT LTP will result in a VAS bias current change from design nominal of about 7%.
Click to expand...


That is for undegenerated BJT's. The greatest source of offset error is the tollerance of the emitter degeneration resistors, both of for the LTP and the CM (which both contribute to an input offset error). You are running a (low) tail current of 2mA. That gives 225mV across each emitter degeneration resistor - a 1% error returns an additional 2.25mV.

A 5mA tail current would be preferabe, both for input stage linearity and slew rate with the miller compensation. Input offset will increase proportionally.


Cheers,
Glen
 
#17,043
scott wurcer said:
LP or CD?

I was wondering if the CD was 16/44.1, since that is one of the great inconsistencies of the audiophile community.
Click to expand...

I know of no CD that isn't 16/44.1 (maybe a 14 bit encode in fact but. . .) they just don't play. However there are now a number of good recordings at higher sample rates and greater bit depth, just not on CD. Most are only available as download.
 
#17,045
scott wurcer said:


LP or CD?

I was wondering if the CD was 16/44.1, since that is one of the great inconsistencies of the audiophile community. Even the most diehard subjectivists have raved about at least one ordinary CD but when asked they give you lines like, "if you can stand them you should find another hobby".
Click to expand...

Did you Scott take Earthworks' demo CD on AES last year?
I can't believe differences between 50 KHz and 20 KHz mics are audible. But they are.
 
#17,050
 
#17,052
 
#17,053
john curl said:
See what it is like for me, Wavebourn? When people criticize me without understanding.: geezer:
Click to expand...

John,

From your comments placement and the fact I commented on your post, I'm assuming this was directed at me... There was no criticism only commentary. For those who have never really listened to a properly functioning Ampzilla it's easy to throw stones. To those who have, they have trounced many a later design in the important/musical aspects; even thirty years later.

An yes, I know you're commenting on the electrolytics bypassing the differential emitter resistors, but this was 1975, but also look at the choice of the original diff transistors...

There are subtle lessons to be learned but you have to look at the whole execution not just the simulation and damping factor. Mr. B was ahead of is time as were others we know and love...

Cheers, Mike.
 
#17,056
Re: Re: Re: Re: Re: Re: Re: Re: VAS fighting

G.Kleinschmidt said:



That is for undegenerated BJT's. The greatest source of offset error is the tollerance of the emitter degeneration resistors, both of for the LTP and the CM (which both contribute to an input offset error). You are running a (low) tail current of 2mA. That gives 225mV across each emitter degeneration resistor - a 1% error returns an additional 2.25mV.

A 5mA tail current would be preferabe, both for input stage linearity and slew rate with the miller compensation. Input offset will increase proportionally.


Cheers,
Glen
Click to expand...


Guess I'll just have to use 0.1% resistors; they are still a lot cheaper than high-quality caps 🙂.

For those who still have a problem, they can with my circuit cripple the VAS as much as they like; they always have that option.

Cheers,
Bob
 
#17,058
Re: Re: Re: Re: Re: Re: Re: Re: Re: VAS fighting

Bob Cordell said:



Guess I'll just have to use 0.1% resistors; they are still a lot cheaper than high-quality caps 🙂.

For those who still have a problem, they can with my circuit cripple the VAS as much as they like; they always have that option.

Cheers,
Bob
Click to expand...


NPO's are a dime per dozen and nearly everyone has a half decent capacitance meter these days in their digital multimeters.

Anyway, my main point was that in full-diff amps with high LTP gain the importance of accurately matching the Miller caps is no more important than accurately matching just about everything else.

As for the option of "crippling the VAS", if you want to make a design tolerant to component value spreads then that may be a perfectly valid option.
I've already shown how one such input/VAS design with 1k LTP loads (shock! horror!) can distort well below that of a class A BJT OPS (TMC gave a ~5 fold decrease in THD, proving that the OPS non-linearity was well and truly dominant over the input stage and VAS non-linearity).

The net linearity benefit of further "un-crippling" the VAS in such a design would be next to zero.

Cheers,
Glen
 
Status
Not open for further replies.
Top Bottom