[DIY/Review] Three IC Op Amp Performance Metrics and Counting

Product: CD Players
Manufacturer: Various
Manufacturing cost: None
Reviewer: Paul Hunting - TNT UK
Reviewed: October, 2022

Introduction

This piece looks at three integrated circuit op amp performance metrics with reference to six CD player and op amp combinations (Table 1). Performance metrics covered are noise, slew rate and common mode and power supply rejection ratio. The piece looks at what these metrics mean and what they imply for sound reproduction. Four of the op amp/CD player combinations use the same CD player or similar copies: the Philips CD753. The four players are fed by the same DAC (TDA1549T signal to noise ration: 110dB) and build. The Marantz CD5000 is a CD753 clone. It is important that the same CD player is used so that we pick up differences between op amp and not also between CD player build and DAC differences. Two dissimilar op amp/CD players are also included to provide context.

Noise

The equivalent input voltage noise (uVrms) is voltage input terminal noise and this is amplified by the op amp gain stage and appears in the op amp output. From the figures and listening the Equivalent voltage noise value is a good measure of op amp noise performance. Track 10 Music for 18 Instruments by Steve Reich (Nonesuch) features maracas shakers that keep pace and add tension. With both the NE5532/CD550c (0.7uVrms, DAC S/N Ratio 117dB) and the NJM4562/CD753 (0.6uVrms), in a Philips CD753 (DAC S/N R 110dB), the maracas can be clearly heard against the ensemble and this, for me, added to the listening experience. With the OP275/CD753 (0.85uVrms), also in a Philips CD753, you had a sense of the maracas shakers but they were not as clearly defined or as clearly heard. The maracas shakers could not be heard with the AD711 fitted CD4SE player, though the shakers broke through at musically quiet periods. This comparison shows how much noise the OP275 had and that a noise value of 0.85uVrms can be heard compared to a value of 0.6uVrms. Note the higher S/N ratio (117dB) of the DAC (Wolfson WM8740) in the CD550c will mean that op amp noise value, in this player, could go slightly higher and you would still hear the maracas shakers.

Noise gets in the way of what you are listening to and noise will squeeze out background instruments played in an ensemble. If noise goes higher than 1uVrms, like with the NJM4560 (1.2+uVrms) fitted in the Marantz CD5000, a Philips CD753 copy, groups of instruments, in large ensembles, will sound melded together and instrument detail will be lost. If large ensembles are not part of your listening, then you might be fine with a higher noise op amp.

1/f Bass Frequency Noise

Noise can rise sharply in the bass frequencies below 1kHz, this is called 1/f noise. Taking a noise (uVrms) reading from the plot of noise by frequency at 100Hz or at 50Hz gives an indication of likely 1/f noise.

The NE5532 IC op amp, used in the CD550c, has noise of 1.1uVrms at 100Hz (Figure 1 page 6 datasheet). That is an increase in noise, at 100Hz, of about 50% over its average figure of 0.7uVrms and noise continues to rise as frequency falls. Bass instruments, with the NE5532/CD550c, could not be made out so well and were a bit lost in large ensemble pieces. There were, however, no noticeable issues with bass, with this op amp and player, playing solo instruments or small ensemble pieces and bass sounded clear and musical. The Exposure 2010S2 uses two NE5532s and WhatHiFi? said of this player that the bass suffers from thinness and a lack of resolution. 1/f noise?

It needs to be said that you probably would not notice 1/f noise, if your system does not go much below 100Hz or if your system has a higher noise floor than the op amps or if you only listen to rhymically straight forward or single instrument bass.

1/f Noise and Pace

What 1/f noise means is that noise is not linear over the audio frequency range. And the bass, due to the 1/f noise, is heard, with some material, differently to the rest of the frequency band. This audio difference is thought to be interpreted as pace errors, but only with complex material.

Recently I have been experimenting with the 1/f compromised OPA1602 and this op amp had pace problems for about 5% of CDs played even though it has a highish slew (20V/us) value. For the OPA1602 noise increased from 0.27uVrms at 1kHz to 0.32 at 100Hz (18% increase) and to about 0.41uVrms (52% increase) at 50Hz (Figure 1 page 5 datasheet). However there is some doubt as to the gain compatibility of the OPA1602 in the host CD player (CD753) and this will have a performance impact. And there may well be other compatibility factors. Further, pace issues were identified in a shunt resistor based pre amplifiers, see another article here on TNT-Audio, and it is, perhaps, telling that resistors also have 1/f noise. The link between 1/f noise and percieved pace is, though, a hypothesis and needs further research and observation.

Noise - Different Measures

Op amp datasheets may quote the noise figure in nV/square root(Hz) units. This is uVrms expressed in nano volts (nV) divided by the square root of the frequency range (square root Hz). The audio frequency range is usually taken to be 20Hz to 20kHz or 19,980Hz, the square root of this figure is 141. So to move from uVrms to nV/square root(Hz) units or vice versa just divide or multiply by 141 and move the decimal point.

Slew Rate

Slew rate is measured by V/us as the rate of change in the output voltage caused by a step change at the input. High slew rates mean that the op amp can output an accurate recreation of the input signal or, at most, as accurate as the input signal.

Players with high slew op amps have been cited as being musical. HiFi Choice, issue 163, labeled the Cambridge Audio CD4SE as musically convincing and HiFi News, January 1997 said of the Audio Research CD1 that it had inherent correctness. Both players have high slew op amps. The CA CD4SE is fitted with the AD711 (20V/us) while the AR CD1 is fitted with the SSM2131 (50V/us). Other highish slew op amps, like the OP275 (22V/us), are regularly cited by commentators as being musical. But other factors may also be important.

Slew Added Distortion

Low slew rate players have wider output to input time delays compared to high slew rate players. These time delays add error odd-harmonics not present in the original signal according to Bullard and Bullard Laws of Harmonics. The harmonic errors are defined by the slew rate value. It is hypothesized that these time-related added odd-harmonic errors result partly in the differences between what is heard between op amps with different slew values. The greater odd harmonics of low slew op amps give a bit of an edge and a cleaner sound or call it less tone-full. High slew op amps will sound less edgy and less clean and will be perceived as having more tone. And yes this is another hypothesis that needs further research. An alternative model says that lower slew rates result in higher jitter and that jitter defines op amp sound differences. But both models say much the same thing, undoubtedly output to input time delays result in jitter but jitter is a much more broader term.

I listened to the players in Table 1 and there was a noticeable relationship between slew rank and instrument tone. Low slew value players (NJM5462/Philips CD753, 3V/us) sounded a little tone washed out while higher values were associated with more tone. l could hear wood tones to the string instruments with the CD4SE (20V/us), more so than with the OP275/Philips CD753 (22V/us), which implies there are additional factors to consider here. In general: lower slew rate op amps seemed to have more loss of instrument tone and higher slew rate op amps less tone loss.

Slew and Pace

High slew rate values should mean better tracking of the input signal and to better playback pace accuracy. However in general I did not find from listening that lowish (3V/us) slew rate players lack pace. There are instances but these events could be linked to other factors. The fast paced, complex and congested piece Lick (Julia Wolfe, Bang on a Can) is accurately presented using the high slew OP275/CD753 (22V/us) but seems a bit muddled and had a perceived pace issue with the NE5532/CD550c (9V/us), though the rise in 1/f bass noise of the NE5532 op amp, and the hypothesised effect of 1/f on noise linearity over frequency and on perceived pace, may have implications here. The OP275, in comparison, has no rise in 1/f noise. The NE5532/CD550c (9V/us) excelled in playing rhythmically and bass simple but pacey opera.

The NJM4562/CD753 (3V/us) played the slow Feldman piece Piano And String Quartet (Kronos Quartet) at a seemingly fast pace. I was more aware of the slow, drifting, unwinding and meditative pace of this piece with either the NE5532/CD550c(9V/us), AD711/CD4SE(20V/us) or OP275/CD753 (22V/us) in place. And this is how it should be heard. However output current or other factors of the NJM4562/CD753 may have had an impact here rather than the low slew rate.

Common Mode Rejection Ratio and Power Supply Rejection Ratio

Common mode and power supply rejection ratios indicate the ability of the op amp to suppress noise signals common on both inputs. It is a measure of the op amp after the CMRR/PSRR iteration. A high CMRR/PSRR rate means better noise rejection. CMRR is the more important measure here.

The ability of CMRR to reject common mode noise usually declines as frequency increases. The CMRR figure of the OP275 (TPC. 7 page 4) is a flat 106dB up to 3kHz and thereafter falls reaching a value of 90dB at 10kHz and 85dB at 20kHz. Most op amp datasheets in Table 1 do not chart CMRR against frequency but they do give the typical figure and a minimum, top octave, figure, though manufacturers might not have the same definition of top octave.

CMRR and the Top Octave

CMRR values are compared between the OP275/CD753 (typical CMRR=106dB, min. 80dB) and the NJM4562/CD753 (typical CMRR=90dB, min. 70dB). Both op amps are here DIY fitted in the same CD player model (Philips CD753) and are fed by the same DAC (TDA1549T) and build. The NJM4562/CD753 is the more detailed player but is notably as having more top octave noise heard as an unwelcome brittleness to instruments played. It is thought that at this top octave frequency, the NJM4562/CD753, with its lowish 70dB CMRR value at 20kHz, just lets through that much more noise and pushes instruments played at this frequency into brittleness. The OP275/CD753 with its higher 80dB CMRR top octave value means less noise at high frequencies and top octave hardness and instrument brittleness is not heard so much.

Considering the more recent OPA1602 op amp, also trialed in a CD753, this has a CMRR figure of 114dB at 20kHz and the sound of the high treble was very good: realistic, smooth and detailed and with no hint of treble harshness whatsoever.

If your top octave is rolled off or the tweeters are well damped or you just do not listen to instruments played in the top octave then a high top octave CMRR value might not be so important to your listening pleasure. Also what you hear in the top octave will be impacted on by DAC filters and noise shaping. The CD753 implements a relatively weak noise shaping (2nd order) so will need a high CMRR value to temper top octave noise. In contrast, the Marantz CD53 has a stronger 4th order noise shaping and a highish CMRR value at the top octave might not be needed for this player.

Conclusion

CMRR and slew are relevant performance metrics and it is probably best to avoid 1/f noise compromised op amps. Yes, noise is an important metric but not necessarily the most important. After listening to various op amps I do not think that I can hear that much difference between CD players with op amp noise values below about 0.6uVrms. But the user needs to decide their own preferred op amp combination of noise, slew and CMRR values. I know from forums that some users prefer an edgy and clean sound (low slew) while others distrust a detailed sound as being less true to a concert hall experience and these users might prefer a high slew but a relaxed on noise op amp. I like a detailed sound with a good sounding top end and no perceived pace issues so for me I would like a lowish noise op amp that has a high top octave CMRR, a middling slew value and low 1/f noise).

Op amp rolling is not an automatic upgrade solution as the realisation of an op amps performance metrics is, in part, dependent on the op amps compatibility in the host CD player and the players surrounding circuitry. Further, overall op amp performance may be impacted on by downstream components like the preamp, amplifier and speakers and upstream by filtering and the DAC. In addition you may wish to maximise other op amp features like output current (NJM4556) or a high open loop gain figure.

There are so many parts to CD reproduction that it is at worst spurious to link any measure to a performance gain or loss. However, we should not give up trying. Clearly we can not judge by listening every DAC, op amp, or build variation of CD players, playing every type of music. And metrics like noise are measurable and fixed. Although CD player circuitry and DACs will have an impact on final sound, fitting a lower noise and compatible op amp, all other things equal, should yield a lower noise floor. You might like this you might not. Understanding metrics gives a little power back to the buyer, DIYer and user.

DISCLAIMER. TNT-Audio is a 100% independent magazine that neither accepts advertising from companies nor requires readers to register or pay for subscriptions. If you wish, you can support our independent reviews via a PayPal donation. After publication of reviews, the authors do not retain samples other than on long-term loan for further evaluation or comparison with later-received gear. Hence, all contents are written free of any “editorial” or “advertising” influence, and all reviews in this publication, positive or negative, reflect the independent opinions of their respective authors. TNT-Audio will publish all manufacturer responses, subject to the reviewer's right to reply in turn.