The NE5534....misunderstood?

[analog_sa]

Is this amp inverting or non-inverting? What are the input and feedback impedances? 553x amps don't have the best reputation for CMRR (common mode rejection ratio).

A very good question. The 5534 IME sounds incredibly better inverted with equalised source impedances on both inputs.

I have the feeling that most people simply won't put up with the inconvenience of inverted operation.

old Siliconix app note

Could you tell us more about it?

all of these seem dull and uninvolving compared to circuits with tube or discrete transistor input stages.

Indeed they do. The lower the input voltage ranges, the worse the seem to perform.

[Sigmo]

Originally posted by analog_sa


What you say is not without merit but to find the best opamp is not that hard.

Just put together a simple amp with a gain of 10 followed by a resistive divider to equalise the gain back to unity.

Hello all. Please excuse this long diatribe again.

I’m really no expert in Op Amps or audio circuits, and I do believe that you could learn something from an experiment such as you describe. But I disagree about your experiment helping you find “the best Op Amp”. It could only help you choose an Op Amp that you like when used in that or a similar circuit.

Here is just one simplified example:

Let’s say I build a simple gain-of-ten amplifier circuit into which I can plug various Op Amps to be tested. I set this circuit up as a non-inverting amplifier and I decide to use relatively low values for the feedback resistors to keep noise low, but the source impedance happens to be quite high.

Now everything seems ok, especially if I’m going to use JFET input Op Amps because their input resistance is so high that there won’t be any problem with increased offset from the imbalance in the input circuit impedances.

So I begin my testing of Op Amps and I first insert an Op Amp which has input JFETs with very low gate junction capacitances. The circuit will probably sound fairly good.

Next, I insert an Op Amp who’s input JFETs have larger input junction capacitances. This Op Amp will probably display a much higher level of harmonic distortion than the first one did.

The mechanism for this is explained very well by Walt Jung in this article:

http://www.elecdesign.com/Globals/Pl...tent/1518.html

My test might lead me to believe that the first Op Amp (with the low input junction capacitances) is “better” than the second Op Amp (with the larger input junction capacitances). But in reality it is only because the surrounding circuit was not optimized for the Op Amp with the large gate capacitances that it ended up sounding worse than the one with the smaller gate capacitances.

If I were to use the “bad” Op Amp in an inverting circuit, or if I were to balance the source impedances seen by the inverting and non-inverting inputs, I might well prefer the sound of the Op Amp with the larger input device gate capacitances.

And of course, the reverse would also be true. If I designed the circuit to minimize or eliminate the effects of large input JFET gate capacitances, I might end up thinking that the Op Amp with the large input capacitances was “better” than the other one. But again, it wouldn’t be a fair test and certainly wouldn’t prove that one Op Amp was better than the other in every application.

Thus, by using this type of general experiment, I would have fooled myself into believing that one Op Amp was inherently worse than another when in fact, it had just been misapplied – a victim of a circuit which brought out its worst characteristics.

This is just one example. There may be as many examples for how the surrounding circuit might favor one Op Amp over another as there are different Op Amps.

I’m not trying to say that all Op Amps will make good audio circuits. And I, like all of us, want to find the ones that have the most potential for my projects. I certainly won’t be using a 741 or a 1448 in any serious audio circuits. And I would look for faster Op Amps than the 5532 or 5534 for a new high quality design. But those faster amplifiers will require that I do other things in their surrounding circuitry to bring out their best. If I plug a 2500V/uS Op Amp into a socket where I just pulled out a 5534 in some old piece of equipment, I’d lay even money that it will oscillate and sound really nasty. It darn sure won’t perform up to its potential. But again, that doesn’t mean that the 2500V/uS Op Amp is inherently bad. Again, it’s just misapplied.

So my point is that we need to be careful not to draw false and illogical conclusions based on invalid or unfair tests. Op Amps do not behave independently of their surrounding circuitry. Any test circuit can only test the Op Amp’s performance in that test circuit. We can draw inferences about how it’ll behave in similar circuits, but nothing more. You can’t separate the Op Amp from its surrounding circuit – no matter how generic and “ideal” you believe that test circuit to be.

And, as has been pointed out as the basis for this thread, even the much maligned 5534 can sound quite good in the right circuit. Just because I may have heard a circuit where it sounded bad doesn't make the chip inherently bad. Of course, it doesn't make it good either

[TimA]

Sigmo,

I expect you are correct, an op-amp must surely interact with its surrounding circuit. However I have had very similar subjective results from various op-amps in three different circuits, I/V conversion, a balanced to unbalanced converter and a power amplifier voltage gain stage. In these circuits each op-amp displays a strong resemblance to its own sound in the other two circuits. From this I conclude that the circuit complexities within each op-amp impart a sonic signature that dwarfs (at least in the above mentioned circuits) the effects of the surrounding circuit.

I suspect the greatest variable of all is human. Just as our eyesight is different we surely have ears that differ too, leading to pronounced differences in the signals from ear to brain. At first thought this would perhaps explain much about how our perception of good seems to differ, however we experience the same real-world sounds as each other and therefore the brain should ‘know’ what is real and what is not, so perhaps it ‘adjusts’ to the ears own peculiar response unconsciously.

So where does this leave us? I am not convinced that overall system balance is necessarily the overriding influence over what makes one op-amp sound good in one system and bad in another, though it must surely have large significance (I dislike the idea of ‘correcting’ a frequency response anomaly in a loud-speaker – brightness, for example - by selecting the op-amp which sounds least bright. The op-amp probably has a similarly flat response to its counterparts and therefore is probably the wrong 'tool' for the job) and there are certainly large differences between the way different rooms and speakers interact.

So I wonder if the greatest and most significant variable is not the human ear, nor system balance, but human perception - the way we listen to, and what we expect from, music.

Tim.

[JFETglare]

Someone put it best earlier in the thread with "There's no accounting for taste." i.e. personal preferences are not up for debate.

Though the NE5534 doesn't fit my tastes exactly, it obviously suits others just fine. We all hear things differently and certainly focus on different aspects of the sound, no matter how objective we try to be.

I have found many times that an op-amp may perform extremely well in some circuits, and not so well in others. Therefore, I believe it is important to evaluate each situation objectively and not discount one op-amp or another based on previous experience in a completely different circuit.

Happy experimenting! It is good to know that some people actually listen to their circuits and are not exclusively concerned with test specifications.

[analog_sa]

the much maligned 5534 can sound quite good in the right circuit

Sigmo

Again, i mostly agree with your views. My point was slightly different - that opamps, same as everything else, have their own, inherent sound which you simply can't avoid, no matter how well you meet their specific requirements.

Last time i did a straight wire test with a 5534 was almost 20 years ago. Even under ideal conditions it did something bad to the highs - reduced decay, flattening of perspectives, generally lowered subjective resolution. Under less than ideal conditions and especially in non-inerting mode it scored a lot worse.

More recently i've repeated this experiment with a 627 and buf634. Arguably better than the 5534, both these devices, alone or together, add a distinctive sonic imprint on all music passing through them. The 627 may well score better than a 5534 under real world conditions but is just as easy to spot against a 'straight wire' (what a term )

[fmak]

More recently i've repeated this experiment with a 627 and buf634. Arguably better than the 5534, both these devices, alone or together, add a distinctive sonic imprint on all music passing through them. The 627 may well score better than a 5534 under real world conditions but is just as easy to spot against a 'straight wire' (what a term ) [/B][/QUOTE]
-----------------------------------------------------------------------------------
I agree with you about 627/634. Detailed, hifi sound asnd don't understand why people go ga ga about the 627 (I have 12 that I have taken out and replaced with AD825/LM6171/OPA132
etc that sound much better.

[nuvistor]

analog_sa -

I have not tried 5534 in inverting circuit with balanced source impedances, but I have found that solid-state power amps sound cleaner and more detailed with such a circuit as long as the preamp can drive the lower input impedance (I use 1k input). I sometimes wonder how inverting bipolar input op amp circuits with JFET input buffers would sound, I find buffers with cascode NJFET followed by NPN with constant current sink to be essentially transparent if not loaded too heavily.

"Composite Op Amp for High Performance" in the 1986 Silliconix FET databook is the app note I referred to. This is long out of print and I have not see it on line, but it's prettty simple:

5534 pins 2,3 are connected to 4, this turns off its input stage. A 2N5912 dual JFET as differential pair is cascoded by another 2N5912 dual, gate of cascode pair tied to source of diff pair. Drains of cascode pair are connected to 5534 pins 1,8. 1.37k resistors from drains to V+ are in parallel with the 5534's 13k collector resistors, 2.5V is expected across the drain resistor. A 4mA "ring of two" constant current sink with 162 ohm b-e resistor completes the circuit.

In my builds, I used high Vp FETs for the cascodes like 2N4392, they don't need close matching, and for the diff pair, medium gm duals like U401, about 1mA drain current . A modern high gm dual like 2SK369 would probably work well but I haven't tried this. A 2 - 4mA current sink on the output gives a smoother sound.

The Nat Semi app note AN-222 has a MM RIAA preamp using LM394 and LM318 using the same idea of turning off the op amp input stage and connecting an external diff pair, but the Siliconix circuit is a better design.

fmak -

LM6171 were harsh sounding in my first use, a 5mA current sink helped some, and a cap right on the DIP package (I used 18nF NPO) pretty much removes the harshness, these are fast amps and in audio layouts may have stability problems without very close bypassing, my guess is that 100nF isn't much different with close lead dress.


Sigmo -

Agreed, the op amp and its application circuit must be considered together, general statements about superiority of one chip over another may not take this into account.

Tim A -

The multi-bit DAC I/V application is at audio frequencies probably one of the better circuits for 5534 since it is inverting with low source impedances, but bipolar input stages can have slew-related problems due to the step transistions at the DAC current output, this can be mitigated by adding the right cap values between DAC current output and ground and across op amp feedback resistor as described by Scott Wurcer in the AD744 data sheet (Scott where are you? )

[Sigmo]

Hello all,

This is a really great forum. My thanks go out to everyone for all of their insight and willingness to share.

I found this forum while searching for exactly this sort of information. It’s been a while since I built any good audio projects, so I’ve been rooting around for information on what Op Amps I should consider.

Despite the main point of my previous posts to this thread, I really do hope, and fully expect that I will find a couple of Op Amps that sound good to me and which I can use as the basis for some new projects.

Being able to read about which ones sound good to all of you is very helpful in narrowing down the field. The IC manufacturers rarely see high-end audio as a large part of their potential market so you don’t often see high quality audio applications in their data sheets. Of course, there are exceptions. I’ve still got about 20 AD846s that I bought years ago. The ADI databook at that time actually showed an audio preamp schematic as one of the applications in that IC’s datasheet.

From what I’ve read, I really wish that the AD825 were available in a DIP package. It might compromise the operation to put it in a different package, though. But from reading what others have written, and from looking at the data sheet, it seems like a very promising IC. Probably worth the effort to deal with the SO package.

Has anyone played with ADI’s OP275? This is touted as being an “audio op amp” by ADI. It’s a dual, so it might just plug right into many existing circuits for a quick, easy upgrade. I know, I know, I do remember what I’ve said about doing that, but hey, I’m always up for an easy way to make a crummy CD player sound better too!

It does seem as though the semiconductor manufacturers could create, if they wanted, a really great – and somewhat universal – audio Op Amp. The OP275 is the one I see as being ADI’s premier offering in this area. ADI’s data sheet for the OP275 says that its 22V/uS slew rate “is the fastest of any standard audio amplifier”. Now if I recall correctly, the NE5534 is about half this fast, and the gain-bandwidth product of the OP275 is 9MHz. That’s exactly the same as the old NE5532s that I’ve got!

So on the surface, it seems like this state of the art “audio Op Amp” may not be a significant improvement over the NE5532. But I’ll bet that it sounds a whole lot better due to its totally different input architecture. Still, I wonder how it sounds compared to the “non audio” types that we all seem to recruit into audio duty when designing a high-end audio device.

Thanks again for all of your opinions and experiences with this. I’m itching to order some parts and get going with a few new projects here.

[fmak]

fmak -

LM6171 were harsh sounding in my first use, a 5mA current sink helped some, and a cap right on the DIP package (I used 18nF NPO) pretty much removes the harshness, these are fast amps and in audio layouts may have stability problems without very close bypassing, my guess is that 100nF isn't much different with close lead dress.
--------------------------------------------------------------------------------
The 100nF fits exactly on top. You will find that the type nand value of the caps makes a difference to the sound. With 100nF npo, the sound changes tohints of lushness even!

[mlloyd1]

does anyone know if there are any other current dielectric isolation opamps these days, besides the opa627/637?

based on prices I've seen lately, I'm thinking not.

it is possible this has significant contribution to sonic character.