@EC8010 The main benefit of balanced signals is the signal doesn't pass on the shield. The other big one is galvanic isolation using a transformer, converting single ended to floating shielded (over balanced cabling). Magnetic tape and phono is inherently a shielded signal from the source. Its why I chose a fully balanced front end, plus the compatibility with pro studio gear. You also gain 6 dB of SNR if the CMRR is considered. Some people think you can just ground the neg signal side on the output through a series R the same as the source R. While this will function, its not a balanced signal. You also loose 6 dB of level.
The LM4562, OPA1612, OPA1656, and quite a few others will outperform the NE5532 and NE5534 measurably. All of them will drive 600 Ω at low distortion levels. The LM4652 = LME49720 comes in DIP-8 if you want through-hole. All of them are available in SOIC.
That said, the NE5532 and NE5534 were incredible opamps for their time.
Tom
End equipment was specified to be a fairly good 600 Ohm impedance, but the wire pairs are nothing of the sort. At high frequencies the pairs are more like 100 Ohm and at audio frequencies, resistance swamps inductance, so frequency dependent
@profiguy Absolutely. On all your comments.
@davidsrsb Intriguingly, when we measured some local ends in the UK, we found that they did tend towards 600 ohm, which surprised us as telecoms lore is that 600 ohm was the characteristic impedance of parallel wires on telegraph poles, not twisted pairs. But I'll agree that they were certainly not truly resistive. What we didn't measure was any loaded line (deliberate inductance added at intervals). That probably would have looked rather more resistive.
@davidsrsb Intriguingly, when we measured some local ends in the UK, we found that they did tend towards 600 ohm, which surprised us as telecoms lore is that 600 ohm was the characteristic impedance of parallel wires on telegraph poles, not twisted pairs. But I'll agree that they were certainly not truly resistive. What we didn't measure was any loaded line (deliberate inductance added at intervals). That probably would have looked rather more resistive.
Plenty can outperform them measurably for some parameters (not audibly perhaps?), but there are caveats - they have lower current noise than most other bipolar opamps, so in some circumstances they are hard to beat for a preamp. Find my an opamp that outperforms a 5534A in all parameters though... Not so easy.
Also if you have them in an existing PCB and swap to LM4562's for instance, there's a risk the performance drops like a stone. The LM4562 is much fussier about supply decoupling than the NE5532 and if you get it wrong it distorts badly. And the LM4562 isn't screened for noise like the 5534A is.
One of the strengths of the 5532/5534A is that they can drive down to about 500 ohms full swing before distortion starts to rise, which is the topic of this thread. Many opamps are only happy driving 2k (or even 10k) loads at full swing.
The ancient telegraph pairs with around a meter between the wires should be around 600 Ohm characteristic impedance. A twisted pair is much lower, but could be equalised with loading coils. For distances of a few km you are actually measuring the end equipment, in Malayasia this has a 14 dB return loss limit, so pretty close.
Strangely North America used 900 Ohm, not 600.
The OPA 1612 is designed for 2 kohm loads. 600R loads rarely exists in modern pro or consumer equipment .
600R loads were originally used in early balanced lines.
Secondly if you read Douglas Self Crossover Design book electro coupling caps only introduce the above measurement links when there are too small in value. How often do diy people use 22uf?
A voltage drop across the capacitor due to its reactance with the low frequency time constant causes a rising impedance. Values of 340-470uf are appropriate.
All this points to is using a robust approach to design. Calculate, simulate and measure.
In addition a 0.01 Wima Film Foil PP bypass capacitor across the electro NP capacitor can be used to shunt any dielectric or rising impedance with frequency. Passlabs use this approach in most of their consumer equipment. For an extra 0.50 cents there are no restrictions on a diy project.
If you must use a 600R load simply parallel up your chosen Opamp. Douglas Self describes how to do this. The benefit is non correlated noise is also reduced.
Transformer distortion is also covered in the above book.
If his recent two part series of the tests on opamps in Audio Express Self outlines the causes of op amp distortion using a range of legacy and new op amps like the OPA1612.
Aside from his tri modal power amps in Wireless Word Douglas Self was the head engineer for Soundcraft Mixing Consoles for many years. This type of analogue design is definitely in his wheel house. Self has also consulted with Cambridge Audio.
https://www.ti.com/lit/ds/symlink/opa1612.pdf?ts=1743371080061&ref_url=https%3A%2F%2Fnz.mouser.com%2F
You are more likely to hear subtle differences in the spread of different harmonic structures in an amplifier than the from a coupling capacitor.
600R loads were originally used in early balanced lines.
Secondly if you read Douglas Self Crossover Design book electro coupling caps only introduce the above measurement links when there are too small in value. How often do diy people use 22uf?
A voltage drop across the capacitor due to its reactance with the low frequency time constant causes a rising impedance. Values of 340-470uf are appropriate.
All this points to is using a robust approach to design. Calculate, simulate and measure.
In addition a 0.01 Wima Film Foil PP bypass capacitor across the electro NP capacitor can be used to shunt any dielectric or rising impedance with frequency. Passlabs use this approach in most of their consumer equipment. For an extra 0.50 cents there are no restrictions on a diy project.
If you must use a 600R load simply parallel up your chosen Opamp. Douglas Self describes how to do this. The benefit is non correlated noise is also reduced.
Transformer distortion is also covered in the above book.
If his recent two part series of the tests on opamps in Audio Express Self outlines the causes of op amp distortion using a range of legacy and new op amps like the OPA1612.
Aside from his tri modal power amps in Wireless Word Douglas Self was the head engineer for Soundcraft Mixing Consoles for many years. This type of analogue design is definitely in his wheel house. Self has also consulted with Cambridge Audio.
https://www.ti.com/lit/ds/symlink/opa1612.pdf?ts=1743371080061&ref_url=https%3A%2F%2Fnz.mouser.com%2F
You are more likely to hear subtle differences in the spread of different harmonic structures in an amplifier than the from a coupling capacitor.
You allowed to post dis supa dupa circuit or tell us what supa dupa OPA you had to use?
With all the suggestions the focus appears to be on subjective concerns rather than signal to noise ratio.
It would be helpful if the signal lather from the source to the loudspeaker/ headphone head amp was outlined.
In the implementation a discrete Opamp from Sonic Imagery will knock out all the technical concerns on current drive and distortion.
But a more elegant solution is using an unbalanced output stage with ground sensing 3rd wire to measure the ground differential at the input of the unbalanced input. This approach eliminates your ground loops.
Unless you have cable runs of more than four metres the CMRR of a balanced output and input is unnecessary.
The most important consideration not given enough attention is the SNR of the balanced input (often no better than 106 db) versus an unbalanced input (135 db).
Then using an analogue active crossover its residual noise contribution and the residual noise of the balanced input stage will sum. If they are both 100 db the summed residual noise will reduce the SNR to 94 db at the rated output.
An unbalanced input will have negligible effect on the overall SNR.
The wrap here is an expensive transformer and the necessary design for a flat frequency response can be eliminated by adopting a elegant but simple solution.
If monolithic Opamp aren’t sufficient use a SMT discrete Opamp. application notes are supported by Sonic Imagery. Andrew Sparks - Sparko also offers technical support.
Annoying as it maybe most recorded music is via a microphone preamp that typically employs monolithic opamps including the NE5532, the 2604 and others. Is anyone complaining? In TI application notes recommendation are made for layouts ect. Four layer pcbs are not difficult to design with proper ground planes, power supply and sign path planes. The user has put the work in in order to get the results. Then simulate on Tina TI and then measurements which can now be done inexpensively.
Discrete op amps operate in class A and sink and source high output current.
It would be helpful if the signal lather from the source to the loudspeaker/ headphone head amp was outlined.
In the implementation a discrete Opamp from Sonic Imagery will knock out all the technical concerns on current drive and distortion.
But a more elegant solution is using an unbalanced output stage with ground sensing 3rd wire to measure the ground differential at the input of the unbalanced input. This approach eliminates your ground loops.
Unless you have cable runs of more than four metres the CMRR of a balanced output and input is unnecessary.
The most important consideration not given enough attention is the SNR of the balanced input (often no better than 106 db) versus an unbalanced input (135 db).
Then using an analogue active crossover its residual noise contribution and the residual noise of the balanced input stage will sum. If they are both 100 db the summed residual noise will reduce the SNR to 94 db at the rated output.
An unbalanced input will have negligible effect on the overall SNR.
The wrap here is an expensive transformer and the necessary design for a flat frequency response can be eliminated by adopting a elegant but simple solution.
If monolithic Opamp aren’t sufficient use a SMT discrete Opamp. application notes are supported by Sonic Imagery. Andrew Sparks - Sparko also offers technical support.
Annoying as it maybe most recorded music is via a microphone preamp that typically employs monolithic opamps including the NE5532, the 2604 and others. Is anyone complaining? In TI application notes recommendation are made for layouts ect. Four layer pcbs are not difficult to design with proper ground planes, power supply and sign path planes. The user has put the work in in order to get the results. Then simulate on Tina TI and then measurements which can now be done inexpensively.
Discrete op amps operate in class A and sink and source high output current.
Last edited:
I'm very familiar with the typical comparison of electronic components used in the professional recording signal chain as well as in higher end consumer audio gear.
What people don't understand is how and where these famed "professional grade" components are being used in the signal chain. Rather than blamimg or accrediting the sound to specific indivdual components, the entire circuit surrounding it needs to be considered as the affecting entity.
The other aspect is that many times what you're hearing as sounding "better", "detailed" and/or "live" is mainly the result of carefully adding specific types and amounts of pleasant sounding even order HD. The spectral distribution of this added distortion is what makes it sound the way it does, but its in no means lower in HD, neither is the accuracy or fidelity improved.
Music production is just a form of artistic expression. The end product we hear is created by producers and engineers. Its mostly a synthetic product which couldn't be further from an exact "photo copy" of the actual sound coming from the instruments or vocalists. So when people tell me component x or y is also commonly used in professional audio production gear, that just potentially means its characteristic distortion / noise profile is present in the audio, but it doesn't mean that audio is cleaner or more accurate.
What people don't understand is how and where these famed "professional grade" components are being used in the signal chain. Rather than blamimg or accrediting the sound to specific indivdual components, the entire circuit surrounding it needs to be considered as the affecting entity.
The other aspect is that many times what you're hearing as sounding "better", "detailed" and/or "live" is mainly the result of carefully adding specific types and amounts of pleasant sounding even order HD. The spectral distribution of this added distortion is what makes it sound the way it does, but its in no means lower in HD, neither is the accuracy or fidelity improved.
Music production is just a form of artistic expression. The end product we hear is created by producers and engineers. Its mostly a synthetic product which couldn't be further from an exact "photo copy" of the actual sound coming from the instruments or vocalists. So when people tell me component x or y is also commonly used in professional audio production gear, that just potentially means its characteristic distortion / noise profile is present in the audio, but it doesn't mean that audio is cleaner or more accurate.
I have a few of the lower noise Signetics 5532AN which I'm going to try just to listen. If I deal with the input offset accordingly and use the correct lower value gain setting resistor ratios around the peripheral circuit, it should be pretty good. Its just the reduced output current being a potential issue.
The typical 22 - 33 pf compensation resistor usually isn't needed if the traces are kept very short along with using a ground plane. These 5532 don't like resistors with ferrous conductor legs. They also need decent PS decoupling, which I often do with 10R and 4.7uF Oscon caps right on pin 4/8 to a common PS ground (NOT an audio ground).
The typical 22 - 33 pf compensation resistor usually isn't needed if the traces are kept very short along with using a ground plane. These 5532 don't like resistors with ferrous conductor legs. They also need decent PS decoupling, which I often do with 10R and 4.7uF Oscon caps right on pin 4/8 to a common PS ground (NOT an audio ground).
Last edited:
Only the LM4560 and the NE5532 will work with a 600R load without loss of linearity. This is illustrated in the TI application notes. More accurately Douglas Self measured a number of opamps in a two part series published in Audio Express under a number of different circuit conditions.
600R is not the problem. Typically Common Mode distortion
We didn't measure using pulses (as you would an RF transmission line), we drove it with sine waves from a 600 ohm source, terminated the far end with a 600 ohm load. We swept frequency and measured the voltage at the sending end. I think (not certain) we used a local end from Broadcasting House (Oxford Circus) to Television Centre (Shepherds Bush), looped back, so probably about eight miles in total. Early 1980s.
From the days I played with OPAmps, the AD8066 was my absolute favourite at a reasonable price. The NE5534 has great sound on a budget. Both will drive a 600-ohm load.
AD8397 if driving the 600 load (almost) rail to rail, with huge amounts of current, is an absolute priority.
AD8397 if driving the 600 load (almost) rail to rail, with huge amounts of current, is an absolute priority.
Last edited:
But probably difficult to use in many audio circuits due to its high bandwidth - layout will be fussy and it probably works best with a groundplane. Pretty impressive distortion up at 1MHz and above!
@Mark Tillotson @Extreme_Boky Yes, you 100% absolutely need a ground plane PCB design with the AD8066. It will just squeal like a burning pig without it and those dip socket adapters will also cause instability issues. I also separate the signal earth from the PS earth, which also improves low level detail.
@availlyrics The 4556 is one of the most drab sounding op amps (for average impedance levels) I've ever come across. Its ok for low impedance headphone use, but thats about it.
The only worse op amps than the 4556 I've come across are the the LM324, 741, TL062 and 4558. The NE5532/34 AN versions are very good, but limited in current output. I've used 2x NE5532AN paralleled and like the results, especially in the low end. I'd even rank it better than the LM4562 overall. The AD8066 is very clean and accurate, but limited to +/- 12V rails. It sounds great with high impedance (>300 ohm) headphones. I use this op amp with my sennheiser HD650s and the combination has the right balance with any source impedance being a FET input. There is a minimal noise penalty (of course) with lower source impedances but hardly noticeable, even using IEMs. The Etymotic ER4Ps also sound great with the AD8066. I built a portable HP amp running off 2x 9V batteries for this purpose. No complaints at all with this setup. If you want ridiculous current reserves for driving HPs, the 4580 won't disappoint. Under specific conditions, its even better than the OPA1612.
@availlyrics The 4556 is one of the most drab sounding op amps (for average impedance levels) I've ever come across. Its ok for low impedance headphone use, but thats about it.
The only worse op amps than the 4556 I've come across are the the LM324, 741, TL062 and 4558. The NE5532/34 AN versions are very good, but limited in current output. I've used 2x NE5532AN paralleled and like the results, especially in the low end. I'd even rank it better than the LM4562 overall. The AD8066 is very clean and accurate, but limited to +/- 12V rails. It sounds great with high impedance (>300 ohm) headphones. I use this op amp with my sennheiser HD650s and the combination has the right balance with any source impedance being a FET input. There is a minimal noise penalty (of course) with lower source impedances but hardly noticeable, even using IEMs. The Etymotic ER4Ps also sound great with the AD8066. I built a portable HP amp running off 2x 9V batteries for this purpose. No complaints at all with this setup. If you want ridiculous current reserves for driving HPs, the 4580 won't disappoint. Under specific conditions, its even better than the OPA1612.
Some other options require a 2520/990 footprint, but the 5534, 5532, and LME49710 are available in DOA-990 hybrid kits with either ME171/181 or BD139/BD140 outputs. Any of these will drive over 200ma.
The 5534 and LME49710 are available from Capi-gear.com
The 5532 is the same footprint with the second section can be enabled as a servo function. It's available from JLM Audio. Output can be biased for Class A or Class AB. Or "none" can be selected for the opamp to roll what ever you can put into a DIP8 socket.
The 5534 and LME49710 are available from Capi-gear.com
The 5532 is the same footprint with the second section can be enabled as a servo function. It's available from JLM Audio. Output can be biased for Class A or Class AB. Or "none" can be selected for the opamp to roll what ever you can put into a DIP8 socket.