[peranders]

http://www.diyaudio.com/forums/showt...177#post492177

Something to think about!

Morsel and the gang, what do you say about Bruno's opinions?

 

[amb]

Quote:

Originally Posted by peranders Something to think about!

Indeed. There are two issues we're dealing with here, the PSRR of the opamp's supply rails and biasing the output stage into class A (for the purpose of better linearity). Interestingly the two most highly regarded audio opamps have opposite behaviors wrt. PSRR: The AD8610 has better PSRR on the positive rail whereas the OPA627 is just the opposite (see their respective data sheets, there is a graph for PSRR vs. frequency).

So, for the purpose of biasing to class A, and assuming that the opamp has a complementary output stage, it would make sense to tie the current source to the negative rail and turn off the internal PNP transistor. There are some opamps with non-complementary output stages and those are more likely to be NPN-NPN. On the other hand, if maximizing PSRR is the goal then it would depend on which rail of the opamp has better PSRR.

For the PPA (and possibly the M³, as this part of the circuit still being considered), the JFET isolation on the opamp's supply rails should make the PSRR point fairly moot.

 

[ppl]

It appears the good folks over at diyaudio are talking about two different subjects at once Balanced Inputs and Class A biasing of Opamps. Peranders if you cant hear the diffence i would save space if i were you and not use the Jfets. Myself I can hear the difference and with a good supply there is still a difference IMHO your malage may varry,

Now this nonsence about referencing the current source to the Rail with the worst PSRR is just silly as a double darlington complimentry output stage is used in both the 627 and 8610 and this presents an impedance to the VAS stage of several meg ohms, thus its not going to matter what rail has the worst PSRR and the only thing we are diong in the PPA is turning off the slower PNP transistor in the output stage and the 627 has a slow pnp while the 8610 has well matched output devices. The bennifit is greator on the 627 than the 8610 Granted.

Oh and on op amps wih an all NPN output stage then biasing to the Neg rail is a must Look at the schematic of the LT1122 it has an internal CS tied from output to V-

 

[morsel]

AMB and PPL are far wiser than me on these matters, but for what it's worth, I am in favor of empirical evidence for murky situations with multiple variables. Test each opamp biased to each rail, and see which works best, or if it makes a difference.

 

[ppl]

I do not agree with the opion of the good folks over at DIY Audio on this however why dont you get your listeninging pannel together and do subjective evaluations or put the possiblility of providing places on board for ether posistion. also if referenced from the + rail P ch fets are needed for this as if you mearly reverse the n ch devices to operatte on the + Rail you now have a current sink not a current source and this matters sonicaly as a simple singhel ended CS dose not sink as well as it sources YMMV this is why i used complementry Current mirrors on my Buffer so as to obtain a symmectricial slew rate wher a CS aka Jung style has better slew rate in the negitive direction than it dose in the positive and i tested the WJ buffers vs mine and noticed quite a difference in asending and desending riase and fall times in the Jung configuation as opposed to equial in mine.

Somthing to think about Morsel