AD797, how to make the best of it

[nelsonvandal]

In an attempt to do a discrete vs monolithic folded cascode opamp shootout, I've breadboarded an amp for my favourite opamps - AD797 as left/right and AD829 comp pin out as active ground.

I simulated AD797 and AD829 in LTSpice and came up with the schematic below. While a 22 pF cap for the distortion cancelling circuit simulated the best, a 47 pF one as suggested in the data sheet sounds better. I put this cap from the buffers output to pin 8 as proposed by Walt Jung. This cap is absolutely essential to get a good sound from AD797. Without it it sounds worse than NE5534.

Much have been said about AD797, and how cranky it is, but I've had a hard time finding schematics showing a good implementation. So, please tell me how to get the best of this great opamp!

[nelsonvandal]

Ground channel is using a TLE2426 as rail splitter and virtual ground for the small signal returns and this for the headphones return.

[aparatusonitus]

Hi nelson,

Sorry, but I don't get it: you have chosen high impedance node from AD829 for your active ground, as well as LED CCS (again high impedance) for AD797 supply? Could you elaborate please?

Edit: Ah there is another sch

[nelsonvandal]

Originally posted by aparatusonitus
Hi nelson,

Sorry, but I don't get it: you have chosen high impedance node from AD829 for your active ground, as well as LED CCS (again high impedance) for AD797 supply? Could you elaborate please?

Edit: Ah there is another sch

I'm not so clever. I used LED CCS since it was what I had at hand, and it could be squeezed inside the small box I use. I prefer JFETs because of simplicity but i don't have any with that idss, 30 mA something. What CCS would be better? Do you think just using resistors, say 100R, would be better than this CCS?

[juma]

Hi Nelson !

Did you build the circuit with AD797 or your conclusions come from simulations?
Sorry, but it's not clear from your post - you wrote about simulations and you comment on sound in the same time. It would be nice to know about real life circuit because I used AD797 only as a unity gain buffer.

[nelsonvandal]

Originally posted by juma
Hi Nelson !

Did you build the circuit with AD797 or your conclusions come from simulations?
Sorry, but it's not clear from your post - you wrote about simulations and you comment on sound in the same time. It would be nice to know about real life circuit because I used AD797 only as a unity gain buffer.

Yes, I've built it. Most parts are on sockets for tweakability.

[aparatusonitus]

You are using LED based CCS to supplay power to AD797 (pins opa+/-), but intristic property of every CCS is that they represents high impedance output, you need as low as possible impedance from PSU, regulated or not.

[martin clark]

Much have been said about AD797, and how cranky it is, but I've had a hard time finding schematics showing a good implementation

The clues are all in the datasheet; read it closely - it's excellent.

One essential item is the input damping resistors, min. 100ohms , at the 797s signal inputs. Everyone leaves them out, then wonders why the 797 is 'cranky'! It isn't - it's simply a very high performance part, and one cannot expect simply dropping it into conventional 'jellybean opamp' circuits to get the best out of it.

[nelsonvandal]

Originally posted by aparatusonitus
You are using LED based CCS to supplay power to AD797 (pins opa+/-), but intristic property of every CCS is that they represents high impedance output, you need as low as possible impedance from PSU, regulated or not.

So I shouldn't use CCS to the opamps at all? The amp is powered by two 6F22 NiMH batteries in series. I can try to bypass the CCS and see what it does to the sound, but that will be tomorrow since it's getting late now.

[nelsonvandal]

Originally posted by martin clark
The clues are all in the datasheet; read it closely - it's excellent.

One essential item is the input damping resistors, min. 100ohms , at the 797s signal inputs. Everyone leaves them out, then wonders why the 797 is 'cranky'! It isn't - it's simply a very high performance part, and one cannot expect simply dropping it into conventional 'jellybean opamp' circuits to get the best out of it.

I'll try to increase these resistors tomorrow as well.

Thank you for your suggestions.