I wish that I could just select a range of 'adequate' IC op amps, and just use them. I am getting too old for listening to subtle differences in virtually equal measuring components, BUT I still hear the differences.
My aim here is to help the newbie.
The thing about opamps is that you arguably need to understand phase margin in negative feedback circuits, in order to have any idea how stable they are. This applies to tube circuits too. You don't want spurious or continuous oscillations. All opamps should have power supply bypass caps within an inch of the opamp, for phase margin consistancy and stability. Most opamps should never try to drive a load less than about 5kohms unless the engineer has picked one that can definitely handle the lower impedance well. One Behringer product that I know of has a standard opamp driving into less than 400 ohms, which caused premature current limit clipping...
I've had good luck with many opamps, but personally I find that the opamps Linkwitz uses (OPA2134) are dam good sounding when used right. I use those and don't really care what someone thinks is better. Linkwitz is a very sharp engineer. If it's good enough for him, it's good enough for me. Ultra-fast opamps are likely to be more sensitive to physical layout, and therefore less stable (more likely to oscillate unless the final circuit is very well tweaked).
Tubes distort more, but usually do it in a way that enhances the perceived sound (unless you're talking about crossover distortion in a push-pull stage). They generate a more natural distortion spectrum, more similar to how the ear-brain mechanism distorts. The dominant distortion product is usually the 2nd harmonic, which is the same note one octave higher, so always musically pleasant. High feedback symmetrical topologies (typical of opamps) will usually create a distortion spectrum that is almost purely odd harmonics, so the 2nd is gone and the 3rd dominates. The 3rd is far less desirable from a musical perspective.
With tubes, the distortion spectrum usually rolls off faster (before clipping), so distortion is less audible and dissonant musically. Tube circuits also have a substantially bigger dynamic range, so overload/clipping is a lot less likely to occur, putting a click sound on loud transients.
Although most recorded music has been run through compressors (easier to listen to and less likely to cause overload/clipping), musical instruments actually have a rather large dynamic range, and need serious headroom if compression is not used.
More esoteric: the negative feedback input stage of an opamp is also a heterodyne mixer to the extent that it is non-linear by itself (feedback not included). It can generate mucho sum and difference frequency spectra, that while very low in level, can amount to noise all over the spectrum. IM distortion is rarely musically enchanting (unless you're distorting a guitar for effect).
Having said all that, opamps used correctly are great, and not likely a significant problem. It always surprises me how many engineers don't know how to use them properly... If you're a newbie, just use the OPA2134 driving >5Kohm load, put a 200 ohm R in series with the output (outside of the feedback path) to reduce susceptability to load reactance (when driving interconnect cables, for example), and use 0.1uF power supply bypass caps within an inch of the chip. Use the star grounding technique with one connection to the chassis. Put a cap across the feedback resistor that rolls off the gain above 100kHZ, and put a passive RC filter at the input that rolls off the signal you're asking it to process at around 50kHZ if you want to be extra careful. Sources, especially digital sources, often have some Rf on the signal that can't be handled well by most amplifier circuits. The passive filter will reduce this.
If you don't ask an opamp to do something it can't handle well, it will work great. Pay attention to dynamic range; will the signal ever get clipped? Use pretty close to the max recommended power supply voltages (usually +/- 15VDC) to get max dynamic range. One way to get a clue about phase margin stability is to drive the circuit with a 10kHZ or higher square wave, and look for ringing at the output with an oscilloscope. Some ringing is OK, but less is always better. Doing this with the expected load connected is even better.
If you put one tube without any negative feedback in a system (signal chain) that is otherwise all transistor, you may get a more natural distortion spectrum delivered to your brain. You might get most of the benefit of "tube sound" without nearly the hassle of going all tubes. Tubes are linear enough to be used without any negative feedback, transistors aren't.
Another trick that some say makes an audible difference is to tie a 20-50Kohm resistor from the output of the opamp to either power supply, thereby forcing the push-pull output stage to effectively run in class A for the smallest signals. The finest details in the music may be less distorted. Any crossover distortion will be more masked by the higher signal level.
Hope this all helps. Any corrections are welcome.
The thing about opamps is that you arguably need to understand phase margin in negative feedback circuits, in order to have any idea how stable they are. This applies to tube circuits too. You don't want spurious or continuous oscillations. All opamps should have power supply bypass caps within an inch of the opamp, for phase margin consistancy and stability. Most opamps should never try to drive a load less than about 5kohms unless the engineer has picked one that can definitely handle the lower impedance well. One Behringer product that I know of has a standard opamp driving into less than 400 ohms, which caused premature current limit clipping...
I've had good luck with many opamps, but personally I find that the opamps Linkwitz uses (OPA2134) are dam good sounding when used right. I use those and don't really care what someone thinks is better. Linkwitz is a very sharp engineer. If it's good enough for him, it's good enough for me. Ultra-fast opamps are likely to be more sensitive to physical layout, and therefore less stable (more likely to oscillate unless the final circuit is very well tweaked).
Tubes distort more, but usually do it in a way that enhances the perceived sound (unless you're talking about crossover distortion in a push-pull stage). They generate a more natural distortion spectrum, more similar to how the ear-brain mechanism distorts. The dominant distortion product is usually the 2nd harmonic, which is the same note one octave higher, so always musically pleasant. High feedback symmetrical topologies (typical of opamps) will usually create a distortion spectrum that is almost purely odd harmonics, so the 2nd is gone and the 3rd dominates. The 3rd is far less desirable from a musical perspective.
With tubes, the distortion spectrum usually rolls off faster (before clipping), so distortion is less audible and dissonant musically. Tube circuits also have a substantially bigger dynamic range, so overload/clipping is a lot less likely to occur, putting a click sound on loud transients.
Although most recorded music has been run through compressors (easier to listen to and less likely to cause overload/clipping), musical instruments actually have a rather large dynamic range, and need serious headroom if compression is not used.
More esoteric: the negative feedback input stage of an opamp is also a heterodyne mixer to the extent that it is non-linear by itself (feedback not included). It can generate mucho sum and difference frequency spectra, that while very low in level, can amount to noise all over the spectrum. IM distortion is rarely musically enchanting (unless you're distorting a guitar for effect).
Having said all that, opamps used correctly are great, and not likely a significant problem. It always surprises me how many engineers don't know how to use them properly... If you're a newbie, just use the OPA2134 driving >5Kohm load, put a 200 ohm R in series with the output (outside of the feedback path) to reduce susceptability to load reactance (when driving interconnect cables, for example), and use 0.1uF power supply bypass caps within an inch of the chip. Use the star grounding technique with one connection to the chassis. Put a cap across the feedback resistor that rolls off the gain above 100kHZ, and put a passive RC filter at the input that rolls off the signal you're asking it to process at around 50kHZ if you want to be extra careful. Sources, especially digital sources, often have some Rf on the signal that can't be handled well by most amplifier circuits. The passive filter will reduce this.
If you don't ask an opamp to do something it can't handle well, it will work great. Pay attention to dynamic range; will the signal ever get clipped? Use pretty close to the max recommended power supply voltages (usually +/- 15VDC) to get max dynamic range. One way to get a clue about phase margin stability is to drive the circuit with a 10kHZ or higher square wave, and look for ringing at the output with an oscilloscope. Some ringing is OK, but less is always better. Doing this with the expected load connected is even better.
If you put one tube without any negative feedback in a system (signal chain) that is otherwise all transistor, you may get a more natural distortion spectrum delivered to your brain. You might get most of the benefit of "tube sound" without nearly the hassle of going all tubes. Tubes are linear enough to be used without any negative feedback, transistors aren't.
Another trick that some say makes an audible difference is to tie a 20-50Kohm resistor from the output of the opamp to either power supply, thereby forcing the push-pull output stage to effectively run in class A for the smallest signals. The finest details in the music may be less distorted. Any crossover distortion will be more masked by the higher signal level.
Hope this all helps. Any corrections are welcome.
Last edited:
Have anyone ever measured the distortion level in the sound from music played on acustic instruments, without any electronics at all, except ofcause for the measuring equipment?🙂
Or maybe record a clean distortion picture of the instruments, and then construct an amplifier to recreate that, canceling the influence of the whole electronic chain out.
(constructive distortion)😀
(constructive distortion)😀
1. Op-amps being "universal" gain blocks, you can not optimise them for the purpose that you want to use them for. For example, you can't change the input stage biasing to noise match your source impedance and when the frequency compensation is internal you can't pick a suitable compensation capacitance for the gain that you use. This disadvantage is only partly solved by the fact that there are many different op-amp types on the market, so you can take the type that suits your application best.
2. An op-amp is a nullor implementation with one of the terminals of its output port connected to the supplies. This connection makes it difficult to apply series feedback at the output.
3. Designing discrete circuits is simply more fun.
The main reason why I often design discrete audio circuits is listed above as number 3.
I for one, already have 100% magnetic distortion in my system, coming from the very powerfull magnets in my PU, and being presented in some very powerfull magnets in my speakers, so no big fuss😀
Are we seriously talking with people who have any knowledge of electronics and/or audio? What is this quatch ?????????????
but that only affects the upper frequency range.
By slowing hearing in the upper frequency range any audible error in the middle frequency range is for you more sensitive as from young guys - so I think.
Other than a few specific posts by knowledgeable practitioners, this thread is a fine example of the utter drivel that surrounds IC opamps. Tragically, audio equipment vendors pander to this garbage, dishing up noisy, expensive, unreliable discrete solid state and tube circuits that in the harsh light of measurements and DBT do not perform as well as the IC opamp, a miracle of modern science and engineering.
Yep, well said.
Marantz and their 'HDAMs'. Burson with their expensive pieces of *deleted* and people claiming to hear 'huge' differences in opamps, so much so, they like to use the term 'rolling' when wasting time, money and effort exchanging one perfectly testing item for another and then convincing themselves it was worth it. The ones that hear 'massive' differences never actually produce actual, meaningful, measured differences do they? Remember, anything you can hear, we can measure can't we?
I use audio test equipment full of opamps, mainly 'ancient' NE5532ANs actually. My distortion analyzer is full of them and it can resolve to 0.001%. Do I trust my ears or my gear?
OpAmps are a true miracle, even from the uA741, early TL0xx etc, right through to all the latest esoteric, overpriced audiophile OpAmps and we are lucky to able to plug in those little 8 pin masterpieces and enjoy their consistency and performance for only a few cents, but to wax lyrical about the differences between one and another is hilarious and only serves to show what little understanding one has for their actual measured performance in the real world.
Last edited:
Yes.
Not so fast....there are deep seated reasons for subtle subjective differences between similarly/same spec'd and performing/measured opamp gain blocks.
In theory an opamp is an opamp is an opamp, in practice and depending on circuit location and surrounding componentry, expect subtle subjective differences between opamps that are spec'd appropriately for the job at hand, ie audio signal current and/or voltage amplification.
Dan.
By all means, when evaluating a device designed to reproduce sound via the human auditory system, NEVER trust your ears.
Similarly, NEVER trust the presence of ice in your glass of water, ALWAYS trust the thermometer.
I HIGHLY doubt that to be the case. It's not that measurement is not useful, but oftentimes we aren't/can't measure the proper things. Bonafide double-blind-test differences should be able to be measured, if we could find the correct parameter to measure.
- Status
- Not open for further replies.