[NelsonVandal]

Quote:

Originally Posted by majkel /img/forum/go_quote.gif In the AD797 I hear that lower treble glaze, maybe it helps for rock but it's too obvious and pronounced that I perceive it as a residual addition to the sound which I would like to get rid of but I cannot. I don't think it's due to stability issues because no symptoms of that have appeared in my circuit - normal power consuption and no weird behaviors while turning the volume knob or plugging the headphones.

Majkel, are you sure you have a small capacitor (the data sheet recommends 50 pF) from pin 6 to 8 on the AD797? It seems to be mandatory. It changed the sound dramatically in my amps, and there wasn't an increased current draw, noise, clicks, interference, high DC-offset or anything else indicating instability. That cap should probably simply just be there. There's still some sort of coloration in the treble like before but lower by a magnitude.

Since you like the sound of ADA4841-x, you should like AD797. It's a bit drier and lacks some of that colorful sparkle, but I think it's more neutral and does everything better.

 

[Filburt]

I'd like to reiterate that the AD797 is _NOT_ something you can just casually drop into a circuit. It can, and will, incur substantial rise in distortion absent proper implementation, and it will do so to a degree that exceeds that of most other chips. You cannot properly evaluate its performance absent proper implementation, as the overall sound characteristic will be substantially altered otherwise, and there is no "common thread" or what have you that will render an impression, in the presence of improper implementation, as being somehow representative of the performance in the presence of proper implementation.

As for the OPA211, I think it's OK, but the AD797 does seem better overall to me. The 211 has more gain stages and more degeneration, I think, so it gets a better 1KHz THD+N figure, but the 20KHz figure isn't really any better and I do wonder if perhaps TIMD or aharmonic products may be higher due to what may be a less optimal feedback structure and overall topology.

Nelson - "coloured" as compared to what, exactly?

 

[majkel]

No, I wasn't using any capacitors between the pins but I remember such eperiments with other AD op-amps and placing the compensation capacitors was degrading the sound - it was loosing immediacy and speed, leaning towards boring sound while preserving the tonal balance. You have to keep in mind that I've been testing the op-amps in a buffered configuration with a global negative feedback loop. This changes the frequency/phase relationships due to the characteristic of the buffer, and all direct datasheet recomendations become invalid. I can tell you that in my circuit the AD8099 is stable at gain of 1 and produces top notch sound, say amongst the few best. And again, what I am writing about is just a slight efect but my ears seem to be sensitive to this. The same with the OPA637, it's similar to OPA1632, where the latter sounds like a sweet spot between the OPA211 and the OPA637. They are actually close to each other, the OPA637 being the slowest, the most airy and the least precise, the OPA211 a tad too dry with OK speed, the OPA1632... could be more colorful but in this field is still better than the AD744.
Finally, our observations might differ because of different accompanying equipment like sources, headphones and the rest of the amp. Refer to my signature for some info.

 

[NelsonVandal]

Quote:

Originally Posted by Filburt /img/forum/go_quote.gif I'd like to reiterate that the AD797 is _NOT_ something you can just casually drop into a circuit. It can, and will, incur substantial rise in distortion absent proper implementation, and it will do so to a degree that exceeds that of most other chips. You cannot properly evaluate its performance absent proper implementation, as the overall sound characteristic will be substantially altered otherwise, and there is no "common thread" or what have you that will render an impression, in the presence of improper implementation, as being somehow representative of the performance in the presence of proper implementation.

I wish you or just somebody could tell me how it should be implemented. You could also tell me how it should sound, in what way it colors the sound. I wont believe you if you say it's the only amp on earth with a totally pure sound without the slightest imperfections. If you do know a way to get this to sound totally transparent, you ought to tell us so we could stop all this tweaking and upgraditis and just enjoy good music. How have you implemented it, or is it a family secret?

AD797 sounds good in my amps now. It has less coloration than other opamps I've tried.

Are you hinting that fewer active components make a better amp? I also have this feeling, and I've started to explore simple discrete amps with fewer parts and I hope I eventually can leave opamps behind.

Quote:

Nelson - "coloured" as compared to what, exactly?

Compared to nothing. I've already told you so. CDP>amp>amp to be evaluated>phones vs CDP>amp>phones. Or DAP>amp to be evaluated>phones vs DAP>phones. When I speak of neutrality, I don't mean "inner details" or esoteric nuances. I mean tonality and if it sounds soft or hard.

 

[majkel]

I've read the datasheet of the AD797 focusing on pin 8 and it says that you should start using this pin around a gain of 10. I tried at unity gain so it doesn't apply. This is quite possible that in a different design the AD797 will shine. It's all about the synergy and I agree that transparent op-amps don't exist.

 

[Shoewreck]

I've finally assembled a headamp based on AD797. That's a revelation, sincerelly. I hear no colouration, just loads of smooth detail. Of course, I've thoroughly read the datasheet, so Cn was installed before the first listening.
I haven't used any output buffer, that's not good considering my K501 are only 120 Ohms. I'll probably add one later, but for now I'm almost fully content with the amp.
One trick I've done was adding a 470 Ohm resistor inside feedback loop. If you check Fig.19 of the datasheet you'll see that something strange happens to AD797' open-loop output impedance. It grows a lot once the load resistance falls below 200 Ohms. I know that sounds odd, but additional resistor seems to lower closed-loop impedance. I may be wrong here, but at least the resistor should also make driving a headphone cable (capacitance!) easier.

 

[NelsonVandal]

I'm glad you like AD797. Have you read http://www.sg-acoustics.ch/analogue_...distortion.pdf. This confirms the sonics

.

I don't really understand where you put that 470R resistor. What values do you use for feedback resistors etc? What virtual ground concept do you use?

One thing I've found out is that AD797 is happier with low impedance at it's input, but using something like a 1k pot has it's drawbacks loading down the source.

This is what I'm running at the moment. A discrete JFET input, class A output "AD797". Sounds very good.

 

[Shoewreck]

Quote:

Originally Posted by NelsonVandal /img/forum/go_quote.gif I don't really understand where you put that 470R resistor. What values do you use for feedback resistors etc? What virtual ground concept do you use?

The 470R resistor is connected between output pin of the opamp and the point where feedback resistor meets the phone output. I haven't yet moved the Cn - I'm not sure if I should. "Real" ground and dual power supply (+- 19V - just a bit beyound maximum ratings). Non-inverting connection. 470R feedback, 120R to ground (gain is about 4.9).

Quote:

This is what I'm running at the moment. A discrete JFET input, class A output "AD797". Sounds very good.

The output buffer is close to what I'm going to add to my AD797 headamp, except JFET current sources in the driver stage - I'd simply set idle current with resistors.

 

[NelsonVandal]

I use my AD797s with CCS loaded diamond buffers and active ground channel (AD829 or AD825). The amp is battery powered by two 8.4 V batteries. I think AD797 is slightly dry and a bit bright, but it's the best opamp I've ever heard.

 

[majkel]

Quote:

Originally Posted by NelsonVandal /img/forum/go_quote.gif I think AD797 is slightly dry and a bit bright, but it's the best opamp I've ever heard.

Because you're using cheaper ANZ or ARZ series. Try BRZ and you'll lose admiration for discrete voltage stages for some time.

 

[Shoewreck]

Quote:

Originally Posted by majkel /img/forum/go_quote.gif Try BRZ and you'll lose admiration for discrete voltage stages for some time.

I just checked the DS. AD797B has generally higher open-loop gain.

 

[NelsonVandal]

Quote:

Originally Posted by majkel /img/forum/go_quote.gif Because you're using cheaper ANZ or ARZ series. Try BRZ and you'll lose admiration for discrete voltage stages for some time.

I didn't know they sound different. The circuit in A and B have to be the same. It seems to me that the B's are picked out with regards to lower DC offset, and better matched inpupt transistors should give a slightly higher OLG. It's funny the B's PSRR is worse. I don't think I'll loose admiration for discrete amps when they sound as close to perfect as they do, and there's still room for improvements with CFP and/or cascode input stage, better output transistors, better caps, higher supply voltage etc.

 

[NelsonVandal]

Majkel, to start a little discrete vs. monolithic opamp battle - have you ever used SA1016/SC2362. The difference using them vs. SA970/SC2240 or BC550/BC560 is like comparing AD797 to NE5532 or OPA2134.

 

[majkel]

Oops. I know the latter, SA970/SC2240 and SK170/SJ74. Good to know. The AD797BRZ are truly better, also the P vs. PA series gives better sound, or BP against AP, the LT1028ACN8 sound different and better than the CN8 series. I tried it out and I'm telling what I heard. Just another legend revised.

 

[Filburt]

Nelson, where do you get your 2SA1016/2SC2362? bdent has some but some posts on diyaudio seem to indicate they only stock the "F" grade which has the lowest beta. bdent's site doesn't specify what grade they have other than it's the "K" type, where K seems to just specify a different Vcbo and Vceo.