Discrete OP-Amp vs. Integrated OP-Amp include discret Buffer - e. g. AD797/844+IRF610

[tiefbassuebertr]

The threads "Discrete OP-Amp vs. Integrated OP-Amp without Buffer" are here:
IC's vs Discretes
Discrete opamp vs regular op amps
Descrete jFET versus IC op-amp quality

similar threads are here (integrated OP-Amp):
AD825 module from LC Audio
How good are the AD825 opamp for a TDA1541 dac IV and output ?
AD797, how to make the best of it
Fully complementary AD797?
AD797 in preamp: need help
Using the AD797 with a volume pot...
AD797, OPA134 or discrete buffer stage?
LME49990 Audio Op-Amp - better than AD797?
Discrete AD797 up and running
Power Amplifier Circuit Topology according AD846
AD 846 AN is ther a better sounding opamp out there
AD 846 replacement
Best opamp for I/V conversion? (DAC)
AD8066 vs. LM4562
and here (discrete OP-Amp)
The-simpliest and lowest distortion discrete op-amp
Commercial available discrete OP-AMPs - Overview wanted
(Diyaudio thread overview post #1)
AD797 discrete clone - comparison of seven variations
New Lineup IDEA - Power Follower/Output stage

But there is no thread with the above mentioned topic.

The main question for me is in the moment as follow:
Is there any justify for the use of a discrete OP-Amp, if I implemet to a good integrated OP-Amp an additional output buffer (e. g. a single ended source follower or emitter darlington follower, either with CCS or simple resistor, either inside or outside of the NFB loop) for using both in a line preamp stages and in phono RIAA input and output stages ??

Maybe one of the members have this check out by listening tests with various op amp and buffer versions.
Thank you very much for advices.

 

[JonSnell Electronic]

I don't listen, I watch and measure to make sure my amplifiers are doing what they are designed to do.
Some change op amps and 'like' the difference in sound, some change power transistors making the amplifier almost unstable at HF and think it is an improvement. A good design will have the optimum op amp in it.
My ears are not to be trusted, my oscilloscope and signal generator along with my harmonic distortion meter and a good quality Fluke DVM is all anyone needs. In my opinion.

 

[sq225917]

Jon, do you believe that if they measure the same that they must all sound the same, or do you accept that we don't often measure under enough test conditions.

 

[miro1360]

tiefbassuebertr do you have simplified internal schematic for TPA6120?

 

[Monte McGuire]

There are some good reasons to use a composite amplifier. One is that thermal feedback from the output stage to the input gain stages of the op amp can result in distortion significant in comparison to the electrical linearity of the amplifier.

There's an old National app note that discusses this problem in detail: http://www.ti.com/lit/an/snoa737/snoa737.pdf On page 5 they show the amplifier's thermal response with its electrical response, and show that the sum closely resembles the observed Vout/Vin transfer curve of many op amps.

So, while you have increased complexity, an op amp used only for loop gain driving a high impedance buffer will not have these thermal problems. While it's hard to measure, I suspect that these problems also happen with transient signals, not just sine waves, so these thermal feedback errors might have a greater effect on musical signals than just adding 3rd harmonic distortion.

I've been working on these kinds of amplifiers and the problem you'll run into is finding a buffer that's fast enough to be stable inside of the feedback loop of an op amp. The AD797 is pretty finicky by itself, and it's a 100MHz amplifier, so you'll have a tough time finding an amplifier fast enough to stick in the feedback loop. I cannot imagine the AD844 working inside of the 797's feedback loop. The 844 is only a 60MHz amplifier, and I'm pretty sure it's not stable as a unity gain follower. If you use it as an inverter, you defeat the point entirely, as the 'loop' op amp ought not drive anything tougher than a bare op amp input terminal, certainly not a 1kΩ input resistor. You can shunt feedback away from the 844 buffer at high frequencies with an extra capacitor, but this also defeats the purpose, and adds needless complexity to what needs to be a very compact layout.

The LME49710 and its family are a lot more forgiving (it's only a 50MHz amp), and much easier to use with a buffer. If you dig around, you'll find around 20-30 high speed op amps suitable for use as a buffers of various cost and package types. Someone mentioned the TPA6120: this seems to be a potentially suitable ~100MHz amplifier, but TI does not provide a SPICE model for it, and I don't have the spare time to make a PCB just to test it when there are a handful of other amplifiers that should be as good or better.

 

[diyralf]

Most op amps are not suitable for audio. It's pretty stupid to exchange an op amp for another type. Most of them start to swing. Even for professionals, it is difficult to keep oscillation under control.

 

[miro1360]

.... Someone mentioned the TPA6120: this seems to be a potentially suitable ~100MHz amplifier, but TI does not provide a SPICE model for it, and I don't have the spare time to make a PCB just to test it when there are a handful of other amplifiers that should be as good or better.

TI provided spice model for it ... just download it, unpack into folder and create own sym component in ltspice with reference on this library: THS6012.LIB

can you please share which amplifiers sounds better as TPA6120? thanks

 

[Monte McGuire]

I just compared the specs to the THS6012, and the TPA6120 does seem to be the same chip - thanks! That helps me a lot.

Not sure about which 'sound better', but for my use as a line level amp possibly driving a long cable, I do not need more than 100-200mA of output current, so I can include amps that might not work so well as a headphone amp - not sure what your application is. Right now, some of the as good or better performing amps (re. stability and low HF distortion) are the THS3001HV, THS4031, THS4631 and the THS4062. However, there are other potentially interesting amps, and I haven't finished looking around.

Now that I have a proper data sheet for the TPA6120, I can see that it is a good amplifier - the TPA6120 data sheet makes it hard to compare to other amplifiers, and provides specs on performance that isn't important to me - all of the audio band performance will be dominated by the LME49710, and what I need are specs about the HF region so I can see how it will work in a feedback loop.

Thanks again for the pointer to the SPICE model - that helps a lot! I like the RGY package that they offer for it, and now I can see if the amp will work in my circuit.

 

[DPH]

Tief -- you'll get plenty of nonsensical answers like the latter, but it's really important to have a goal in mind and work backwards to a schematic that fulfills those objectives.

How much do you want to invent is the other question?

Line-level stuff, especially if you're not driving long cables is going to not load down any number of modern opamps any more than setting up their feedback networks for low noise will. Composites are cool, but definitely tricky to keep stable, and really of questionable necessity in audio (except for their pursuit just for the fun of it).

RIAA has a definite place for hybrid/composite discrete/IC setups as you may want more voltage swing along the way (for overload) than your typical +/- 15V can provide. Then again, with something like the 44V LME49860 (same exact thing as a 49720 except 44V validation vs 36V), how much do you really need? We so rarely need high gain in audio, and line-level is very much in the wheelhouse of IC's.

Jim Williams wrote dozens of articles/datasheets (Linear Technology) on discrete + IC hybrids as did Walt Jung. Really worth your while looking through them and gleaning wisdom.

 

[miro1360]

Monte, you can try create simple buffer after TPA6120 and you can turn headphone amp into amp for speakers, like I did in my easyAMP, this is simplified version with biaspot:

this one test version goes to small PCB 5x5cm

the sound of easyAMP is perfect, I never heard something so good in speakers ...

 

[MarcelvdG]

The main question for me is in the moment as follow:
Is there any justify for the use of a discrete OP-Amp, if I implemet to a good integrated OP-Amp an additional output buffer (e. g. a single ended source follower or emitter darlington follower, either with CCS or simple resistor, either inside or outside of the NFB loop) for using both in a line preamp stages and in phono RIAA input and output stages ??

Maybe one of the members have this check out by listening tests with various op amp and buffer versions.
Thank you very much for advices.

Well, this forum is about DIY audio, that is, about audio circuits that people build because they like to build them. If someone likes building discrete circuitry more than using integrated op-amps, that in itself is a perfectly valid justification for using discrete circuitry.

By the way, there are some feedback configurations that you can make with discrete transistors or valves but not with op-amps (unless you add floating supplies). This includes a nice configuration for RIAA amplifiers invented by Hoeffelman and Meys in the late 1970's, and an interesting but not very well-known tone control designed in the late 1940's by L. V. Viddeleer.

 

[tiefbassuebertr]

Tief -- you'll get plenty of nonsensical answers like the latter, but it's really important to have a goal in mind and work backwards to a schematic that fulfills those objectives.

How much do you want to invent is the other question?

Line-level stuff, especially if you're not driving long cables is going to not load down any number of modern opamps any more than setting up their feedback networks for low noise will. Composites are cool, but definitely tricky to keep stable, and really of questionable necessity in audio (except for their pursuit just for the fun of it).

RIAA has a definite place for hybrid/composite discrete/IC setups as you may want more voltage swing along the way (for overload) than your typical +/- 15V can provide. Then again, with something like the 44V LME49860 (same exact thing as a 49720 except 44V validation vs 36V), how much do you really need? We so rarely need high gain in audio, and line-level is very much in the wheelhouse of IC's.

Jim Williams wrote dozens of articles/datasheets (Linear Technology) on discrete + IC hybrids as did Walt Jung. Really worth your while looking through them and gleaning wisdom.

On a very wide range of performed listening tests over the years I still note a great lack in sound quality in all line stages and riaa stages with opamps, because the quiescent current respective the idle current of the integrated output stage inside of the OP-Amp chip is too low, even on devices like NE5534 or AD797. Discrete OP-Amps provide better results mainly through only reason: Higher value of idle current in the output stage. Therefore the question in post #1.

The question does not refer to a special amplifier project, but generally to RIAA and LINE stages in pre-amplifiers and active crossover networks.
P.S.: My name is Andreas and the member name is based on my old, no longer exist web address - go to webarchive under
A. Kirschner, individuelle Basslautsprechersysteme

 

[xrk971]

I built this hybrid by Wayne Kirkwood and it works exceedingly well. One of the lowest district amps I have measured. Very simple and no global feedback. Feedback internal to THAT balanced line driver (used here in unique mode to provide Class A bias current control).

I have extra PCBs if anyone wants to try. Just send PM me, you pay shipping.

 

[MarcelvdG]

On a very wide range of performed listening tests over the years I still note a great lack in sound quality in all line stages and riaa stages with opamps, because the quiescent current respective the idle current of the integrated output stage inside of the OP-Amp chip is too low, even on devices like NE5534 or AD797. Discrete OP-Amps provide better results mainly through only reason: Higher value of idle current in the output stage. Therefore the question in post #1.

The question does not refer to a special amplifier project, but generally to RIAA and LINE stages in pre-amplifiers and active crossover networks.
P.S.: My name is Andreas and the member name is based on my old, no longer exist web address - go to webarchive under
A. Kirschner, individuelle Basslautsprechersysteme

How about forcing an op-amp into single-ended class A by simply adding a current source, or even simply a resistor to a supply rail? That way you can even make an LM324 work without crossover distortion (not that I would recommend using an LM324 for high-quality audio, of course).