Improved discrete op-amp

[Mr Evil]

Hi all,

Has the circuit been tried with other transistors? How sensitive is it to variations (i.e. beta mismatch, altogether different set of transistors)? Looks very interesting to try out...

I have tried a few different transistors and it doesn't seem too sensitive to variation except:

• The input transistors need to be matched, as with any LTP. So do the current mirror ones. Ideally they should both be dual monolithic parts for thermal tracking.
• The output JFETs, J5/6, need to be matched, plus their Idss is the output stage quiescent current, so don't use ones that are too high or too low. 5-10mA is a sensible amount.

...I'm curious as to what equipment is being used to characterize the circuits and generate the plots - I seem to pick up that its a sound card on a computer - is that a commercially available "system" - i.e. sound card + test box + sw? If so, what?

Many thanks!

You can see what I used on my website.

 

[clem_o]

Very revealing - thanks for the information - particularly the 2nd harmonic of the OPA2604... :-(

I used to use discrete (differential input) circuits for line-level amplification as well, many years ago, and then shifted to op-amps. Perhaps its high time to go back to discretes with circuits like these posted!

Cheers!

 

[richie00boy]

You didn't mention what software you use on your website Also where do you get those nice cases?

 

[Mr Evil]

...Perhaps its high time to go back to discretes...

It's much more satisfying building this stuff - using ICs is too easy! The fact that the performance is potentially better with discretes is just icing on the cake

You didn't mention what software you use on your website Also where do you get those nice cases?

Cool Edit is what I use. I suppose I should put that on my website.

The cases are made by Hammond and can be purchased from Farnell. They are lovely cases, though I do wish the bezels were conductive so that the end-plates made better contact for shielding purposes.