Opamps v Discrete kit

What are peoples experiences comparing kit using each approach? Can opamps never match fully discrete designs?

Nick.

 

I think a more relevant question is whether fully discrete designs can match op-amps?

 

Perhaps - would you care to say a little more?

I have been experimenting over the last two years with I would guess something like 30 or so opamps (and it might be 50). Almost all of them appeared to be flawed in some respect or had too much "character". This character seemed to hold accross both different components and accross different designs. Most recently I have found some that appear to be exceptional but I am wondering just how much better discrete might be - given the simpler circuit design, less use of feedback, optimal components etc.

Can you give some examples of opamp based kit that can stand up to the best discrete designs?

 

I'd go to say as op=amps maybe old hat these days

 

Beg pardon?

 

Well to a few of us at the moment

 

Ah I see - you mean opamps have been a bit of a passing fad? Well the fact that they all have different flaws makes me suspicious. Simplicity always works best and opamps are very complex with individual parts where size is the overriding concern rather than function. Anyone know of a discrete dip-8 plugin?

 

No I mean, they can be surpassed by properly implimented descret circuitry, and sound so much, much better
However I'd say it takes a while to get it right, lots of rethinking sometimes required.
However the result is quite something.
Meanwhile on the opamp trail, where have you got too?

 

I have reached the end of the opamp road now. I am currently swapping in the ad8067 into all my kit - it is borderline stable in the phono so just needs a bit more tweaking - reconfiguring the gain distribution and adding load resistors. Just left with a bit of hf oscillation hiss. I had to wait 6 weeks for sot23-5 adapters hence the delay.

The ad8067 has no apparent weaknesses and pretty much beats every other opamp all around. Its very fast and clear sounding. The chip is an uncompensated ad8065 which may be a lot of the reason why it sounds so good - only works at a gain of 8+ though, a bit like the opa637.

 

I used to swear by the 8065/6, but found them too bright and lacking bass (even with an N/C across the board) I perfer the 8610, a lot more low end control and eveness across the board, but not quite as fast I'll grant you.
Though my equipment doesn't use op-amps in more in strictest sense, do you have a spec sheet for the 8067 Nick?

 

You may have been hanging too much capacitance off it for the gain to which it was set. I am using the ad8065 currently in the volco and headamp and they dont seem bright or lacking in bass. In my setup it thrashed the ad8610. This is the trouble with opamps. They dont like cap loading.

The ad8067 has better bass and clearer treble. Try the ad847 in your setup - if that works in the bass and treble then you know its a cap problem - the ad847 will drive any capacitiance and its almost as good as the ad8065.

http://www.analog.com/en/prod/0,2877,AD8067,00.html

The ad8067 also has tiny offset. I should be able to dispense with coupling caps in the phono stage if I am lucky.

 

No capacitance of merit was on it at all, it just wasn't up to job in that particular application. I don't need opamps in my set up thanks Nick, I've tried them all except 8067.
may be useful for other non serious jobs. cheers Tony

 

It doesnt take much - a few 100 pF is sufficient and you can get that off a metre or two of cable before you even consider what you are trying to drive. I agree discrete will probably be better but the ad8067/65 are as good as opamps get I am afraid. Which is why I am interested in comparing with discrete.

What were you trying to use it for and at what gain?

Nick.

 

i don't tend to use op-amps for the intended purpose, hence why they show up for what they are, (they are not used to drive anything). Gain is whatever I require for the desired application, the opamp is not making the gain.
Hence why my results tend to conflict with general consenus.I also had iffy results with 40 puff caps too.

 

I assume you had used them in the past for gain according to what you written earlier when you found their limitations?

Quite a few people use opamps for power reg - like some of the PT stuff, but use discrete in the signal path. Hopefully you can help me out with at least one of those goals ;-)

 

Hi,

Most fully discrete designs (especially those that use "op-amp structures") tend to be worse than monolithic op-amp's with truely high performance implemented correctly, both objectively and subjectively.

On the other hand discrete circuitry allows you to break free from "generic circuit design" and implement more interesting stuff, especially open loop designs etc....

Another warning, all the "superfast" Op-Amp's tend to have very limited PSRR (much less than normal "Op-Amp's") due to lower open loop gain among other reasons like circuit topology. This means you get to hear a lot more of the powersupplies than with generic "audio grade" op-amp's.

So, I think the answer is:

"Both monolithic op-amp's and discrete circuitry provide excellent perfomance, which is choosen depends on many factors, not all of which are technical."

Ciao T

 

Yes I would definately agree with that - certainly one of the things I am interested in is discrete distributed feedback designs with low global NFB, a bit like the jung composite opamp. High bandwidth and slew also help reduce intermod. Still with discrete you can tailor the passive components as well. Things work a little differently on silicon and there are more compromises. And much more gain. Most of the better sounding opamps seem to have low open loop gain. I aim to address the PSU issue as well - possibly using some form of isolated super regulated supply. My experience is that this makes a huge difference.

Are you aware of any dip-8 discrete plugins at all?

 

Hi,

Why not dump feedback completely, it's not like it's needed or anything like it?

All this super, mega, hyper, out of this universe and so on regulator stuff is IMHO a load of hype. Non of them are as natural sounding and as unproblematic as essentially passive solutions (like per PSU Line 10R after a generic 3-Pin reg followed by an optimised capacitive decoupling including a few 1,000uF as "bulk" decoupling). But hey, maybe that's just me?

Yes, I am aware of such, but I would not consider them equal to well implemented monolithic Op-Amp's (of course, no doubt some people like how they sound). The physical size invariably forces so many compromises that even very basic more spread out discrete Op-Amp structures are better.

Ciao T

 

Ahem isnt some local feedback required even in valve designs - even the ones that say that dont? How else to achieve linearity?

I previously owned a michell iso which received the lithos upgrade - which is effectively a super regulator - the difference was not subtle.

Do you have any info for the dip-8 plugins? Or anything else that could be mounted in any way as a daughterbaord? Ihad considered the super buffer design or something like it as one of the best options, before I try to go custom.

 

Hi,

Nope. Feedback is not neccesary in either solid state or valve circuitry to achieve linearity. You merely need devices with a flat transconductance vs. current curve. These are rare but do exists.

BTW, I personally dislike the term "Feedback" for degeneration (reducing the devices transconductance but increasing the transconductance's linearity), it is at the very least sloppy language and secondly it suggests that degeneration and looped feedback are the same, which they are not.

Correct choice of active device for the job, correct implementation, including where advisable degeneration?

I am aware of the Lithos. I reverse-engineered them once. I have tested similar circuits (not an exact copy) as well as the current source/shunt reg designs Allan Wright is (quite rightly) so fond of in a variety of design variations (I ended up using simple TL431 Shunts, cascoded by valves or mosfets if needed as best solution with C4S style current sources). In much comparison I still prefer very basic passive filtering, if done right to all the active reguation.

An interesting thing to do is to plot the dynamic impedance of any such "Hyped Up" regulators with transient signals. Very illuminating, especially if no large value capacitor follows the regulator (if it does you might as well go passive anyway).

Depending on circuit I would stick to suitable Op-Amp's. What is suitable/best depends so heavily on the circuit that there are no "blanket" recommendations and no devices that are "the best" everywhere and the best PSU decoupling (symmetric to ground or SE) depends often on internal structures.

If I where to start bothering with custom stuff I'd go open loop. The LC Audio Zap Filter is a good example of how to do this (they have made the schematic available in the public domain BTW), I have one in my CD/DVD Player and it is very good, beat handily my previous J-Fet circuit based around a modification from a design by icl.co.jp....



Ciao T