John Curl's Blowtorch preamplifier part II

Status
Not open for further replies.
OK. The point is further made re OPS distortion under low Z loads... and showing that it is cross-over distortion. BTW - wouldnt FFT show this also?

Richard, of course it would. Measurements with fundamental notched of 61dB. FFT is far best tool to show this.

I assume this might be audible. Our imagination is not restricted to a single sine ;)

BTW, this is the opamp + BUF634T. So even a monolithic buffer is not a complete cure.

Regards,
 

Attachments

  • op_unloaded.PNG
    op_unloaded.PNG
    78 KB · Views: 277
  • op_loaded.PNG
    op_loaded.PNG
    84.1 KB · Views: 275
What else about OPS of opamps?

Follower topologies of the output stages of said opamps.

I have a book on IC design from the mid 80's in my possesion, whose author expressed his dislike of the complementary EF topology that is so popular here and is also incorporated at the output of thousands of opamps these days.
He talked about the condition where one of the input transistors of such EF becomes saturated (even if for a very short time) and what happens when this EF is inside the FB loop.
Since I'm moving and can't locate the book right now, I won't comment any further at this time.
But I hope that the others will chime in. Especially Scott .. ;)
 

Attachments

  • buff.jpg
    buff.jpg
    16.2 KB · Views: 266
Last edited:
Follower topologies of the output stages of said opamps.

I have a book on IC design from the mid 80's in my possesion, whose author expressed his dislike of the complementary EF topology that is so popular here and is also incorporated at the output of thousands of opamps these days.
He talked about the condition where one of the input transistors of such EF becomes saturated (even if for a very short time) and what happens when this EF is inside the FB loop.
Since I'm moving and can't locate the book right now, I won't comment any further at this time.
But I hope that the others will chime in. Especially Scott .. ;)

Probably refers to the input followers which might also be starved for current or in the case of the pnp very slow. This would be much less of a problem these days.
 
Run your opamps bootstrapped into class A and with an in loop class A buffer. I did some sims, and younonly get low order harmonics on the rails with this approach,mu like class B which generates harmonics out to 100s of KHZ.

Right - and which opamp has stellar PSRR above 100kHz? Very, very few. I recall Bruno Putzeys pointing out in one of his excellent PPTs that due to Ccomp, PSRR tends towards 0dB at HF, to the rail that Ccomp attaches to.
 
www.hifisonix.com
Joined 2003
Paid Member
LM4562 PSRR.

Very good up to 20 kHz; above 100 kHz it is not spectacular - but then I dont expect discrete solutions are any better (quite the contrary - I expect in most cases they are much worse).

Class A operation and heavily damped (i.e. local RC filtering - I use 22 Ohms and 100uF per rail per opamp) rails fix most of these issues.
 

Attachments

  • LM4562 PSRR.jpg
    LM4562 PSRR.jpg
    138.2 KB · Views: 296
I'd hope a discrete solution could be better - for the simple reason that without the opamp package/compatibility limitations the Ccomp could be brought out to be connected to signal GND, rather than being tied to one or other supply lines. Samuel Groner's discrete opamp does this so I shall investigate if he publishes any PSRR figures for that.

http://weiss.ch/files/downloads/op1-bp/OP1-BP-Datasheet-R1.pdf
 
Last edited:
Don't circuits generally have enough PSRR to be hardly affected by a supply that isn't extremely noisy?

Many messages ago, in this thread, somebody posted 10 or 11 CCS circuits for input stages, showing their PSRR factors. Much the same problems pleague discrete ciruits.

Anyway, their PSRR factors ranged from 10 to 84 dB. The only surprise for me there was the arrangement which uses a BJT, its base fed by a zener diode paralleled by a capacitor, with a resistor and a FET to the ground. Hit the second best in the lineup at -82 dB. A simple, yet obviously effective solution.
 
www.hifisonix.com
Joined 2003
Paid Member
I'd hope a discrete solution could be better - for the simple reason that without the opamp package/compatibility limitations the Ccomp could be brought out to be connected to signal GND, rather than being tied to one or other supply lines. Samuel Groner's discrete opamp does this so I shall investigate if he publishes any PSRR figures for that.

http://weiss.ch/files/downloads/op1-bp/OP1-BP-Datasheet-R1.pdf

Personally, I doubt it.

There are not many discrete solutions that can meet all of the performance points of a good Modern IC opamp. At 100 kHz the 4562 still delivers 50 dB PSSR.
 
Just downloaded and read Ron Quan's latest AES paper, convention paper 9197.

Very briefly, Ron has measured the variation in several types of distortion as a function of the momentary (dynamic) output offset voltage, for instance the variations in high frequency signal distortions superimposed on a slow-varying 'offset' signal in the form of a stair-case or a lf sine wave.

His findings are of interest to anyone serious about audio amplifier design. Not necessarily counter-intuitive, but it is good to see some numbers attached.

Recommended read - thanks Ron!

Jan

But that sounds like a supply impedance problem, not a noise generation problem, so a good regulator would be better even in that case?

I've always asked about the effect four quadrant operation has on output distortion. It isn't very good to ignore the current paths as they change.

jn
 
Status
Not open for further replies.