RMAA Testing of Audio IC Op Amps

March 08, 2009

This article presents basic audio performance test results of analog audio IC operational amplifiers as measured using:

The information is provided to assist those wishing to design basic analog audio IC gain blocks for various purposes (e.g. boosting the output of MP3 players or other audio devices which don't have sufficient voltage output to drive the RMAA tests).

Walter G. Jung's Audio IC Op-Amp Applications 3rd Edn. 1987 SAMS provides a great deal of information and insight into op-amp parameters important for good audio design and the test circuits used here are basically those described in that book.

The initial op amps tested here include the recent high-performance (wide bandwidth/ultra-low distortion) LM4562, and two medium-quality economical and readily available devices: NE5532 and LM833N.
Under the test conditions here, the peak output voltage of the op amps (required to drive the RMAA test) is about 2.6Vpeak. All the op amps above have sufficient slew-rate margin at 20 kHz (by at least a factor of 5) to guarantee that slew-induced distortion will be negligible compared to other op-amp distortions (like THD and IMD). [See the reference by Jung for more details on this type of distortion in audio op-amp ICs.]

Since the Audigy 2ZS Platinum Pro card is not a premium-quality reference card, clearly some results will be masked by the noise/distortion of this sound card itself.

The basic test setup is shown below.

The test-circuit is implemented in a circuit proto-board and is built inside an aluminum box for shielding. A resistor-attenuator box allows compensating for the op-amp gain (20 dB or 60 dB) in adjusting the RMAA signal levels. Good quality bypass capacitors and metal-film resistors help to reduce component noise. The proto-board has proven to be adequate for this type of shielded audio testing up to several-hundred kilohertz. The proto-board also is a great way to quickly modify circuit layout and evaluate differences in components and the sensitivity to their physical placement.

A closeup view of the proto-board showing component layout is shown below.

The test-bed layout shown is for standard 8 pin DIP dual op amps. All measurements were taken with op-amp dual power supplies Vcc = -Vee = 6.0 VDC.

RMAA Testing Method

All RMAA 6.1.2 tests used a 24 bit/96 kHz sampling rate in the loop-back configuration:
      Line Out 1 --> Circuit-Under-Test ---> Line In 3 RMAA was run in mono-mode except for the Audigy 2ZS Platinum Pro Reference run in which case Line Out 1 was connected directly to Line In 3.
For the op-amp tests, Line Out 1 (from Platinum Pro card) was connected to the attenuator box with the settings -20 dB attenuation for the 20 dB gain amp or 0 dB for the 60 dB gain amp (since the *signal* gain for the 60 dB noise-gain circuit is unity). The attenuator output was connected to either amplifier 1 or 2 and the amplifier output was connected to the Line In 3 on the rear of the Audigy Platinum Pro external I/O hub, as prescribed in Creative's "Audigy 2 ZS Series Testing Methodology .." When the level settings are adjusted using the Creative Mixer console with Master = 100% and WAV = 90%, the Line Out 1 level is 2.60 Vpeak or 1.8 VRMS and therefore is well below the rail-voltage (6V) of the op amp. The output load driven by the op-amp circuits is the input impedance of the Line In 3 input which was measured (using a simple 10 kohm voltage divider network at 1 kHz) to be about 14 kohm.

Mixer settings and the RMAA level adjustment and spectrum display are shown below:

20 dB Audio Amplifier Results

The detailed results of RMAA testing with the 20 dB (x10) op-amp circuit reveals very little circuit-performance differences between op amps. The results are summarized below.

Summary of 20 dB op-amp RMAA test with Audigy 2ZS Platinum Pro Card

TestLM4562NE5532LM833NAudigy 2ZS Platinum Pro Ref
Frequency response (from 40 Hz to 15 kHz), dB: +0.02, -0.08+0.02, -0.08+0.02, -0.08+0.01, -0.06
Noise level, dB (A): -98.3-98.3-98.3-105.6
Dynamic range, dB (A): 98.597.998.1105.5
THD, %: 0.00260.00250.00250.0023
IMD + Noise, %: 0.00410.00420.00400.0028
Stereo crosstalk, dB: -50.4-49.8-49.7-103.7

In this low gain configuration, the op amp has a great deal of feedback and this lowers the op-amp contribution to the THD to a point masked by the sound card at ~ 0.0025%. The circuit however degrades the noise characteristic due to input resistances and op-amp noise voltage. In this circuit, the input resistance is about 2kohm (1k from the attenuator box divider and 1k input series resistor Rs) which contributes thermal noise. The bandwidth results are essentially identical for all op amps and is limited by the Audigy 2ZS Platinum Pro ~ 40 kHz rolloff as appropriate for the 24 bit/96 kHz sampling mode used here. Therefore the noise-bandwidth will be the same for all the op amps and a similar noise figure, for the tested bandwidth is expected. In fact, the expected signal to noise ratio (for an input signal voltage of 1.8VRMS/10) is about 103 dB for the op amp alone. When combined with the sound-card noise (~ 105 dB) the measured result of 98 dB is reasonable for this 20 dB amplifier circuit. [Using a higher performance reference sound card for testing with lower THD would reveal differences in THD between op amps, but the noise results would be similar to those measured here as they are limited by the amplifier input.]

Bandwidth and slew-rate differences, for this 20 dB gain circuit can however be seen at much higher frequencies, which of course would not impact the hi-fi audio performance. The scope traces below show the 20 dB amplifier response for a 320 kHz 300 mVpeak input signal (with the output amplified signal being about 3 Vp). At this high frequency, the LM4562 is clearly a better performer with essentially identical input/output waveforms with no sign of any slewing. Both the LM833N and NE5532 are at the slew-rate limit even though the frequency is still only one-third of the 3dB bandwidth for these devices. The slewing is clearly evident in the amplifier output shape distortion (output curves are phase-delayed to the right as expected). For comparison purposes, the result for an older general purpose op amp, MC4558 with a GBW=2.5 MHz and SLR= 1.6 V/us is shown, indicating very strong slew limiting.

60 dB Audio Amplifier Results

The detailed results of RMAA testing with the 60 dB (x1000) op-amp circuit start to reveal real differences between op amps. The results are summarized below. Two devices of each type were tested. The results below show different LM4562 chips, demonstrating how uniform the characteristics are (similar uniformity for the NE5532 and LM833N was observed):

Summary of 60 dB op-amp RMAA test with Audigy 2ZS Platinum Pro Card

TestLM4562LM4562 2NE5532LM833N
Frequency response (from 40 Hz to 15 kHz), dB: +0.05, -0.38+0.05, -0.38+0.17, -1.41+0.28, -2.13
Noise level, dB (A): -80.1-80.1-73.9-77.3
Dynamic range, dB (A):
THD, %: 0.00520.00530.0140.015
IMD + Noise, %: 0.0270.0270.0550.038
Stereo crosstalk, dB: -43.0-42.3-39.5-41.3

First, the frequency-response curves, for this gain of 1000 circuit, allows us to easily measure the gain-bandwidth product. The plots show the superior performance of the premium LM4562 op amp, with a 55 MHz GBW. Because the LM4562 has considerably more available gain (loop-gain) in the audio-spectrum compared to the NE5532 and LM833N, this translates into lower THD. The noise level is again degraded due to the high noise gain but unity signal gain. The LM4562 is still notably better in these performance tests than the other op amps.

Other tests that would be of interest include performance under op-amp output loading, say with a standard 2 kohm load and higher gain circuits (80 dB).