3MHz LM4562 Unity-Gain Follower with Discrete Push-Pull Current Boost Output

RightMark Audio Analyzer test

Testing chain:
Sound Card: Audigy2 ZS Notebook
External loopback (sound card line-out - LM4562+Push-Pull amp - sound card line-in)
Sampling mode: 24-bit, 96 kHz

LM4562 and Push-Pull output with no bias on push-pull transistors
Re(Loads): 22 ohm and 10 ohm Vcc/Vee = 6V MJE172/182GOS Transistors
Mixer Setup: Vol=100%; WAV=100%; Line-In=56% (+0.7dB).
RMAA Signal Level: -1.1 dB ; Measured Vin = 2.65Vp (1.87 VRMS) @ 1 kHz


Frequency response (from 40 Hz to 15 kHz), dB: +0.02, -0.14+0.02, -0.14
Noise level, dB (A): -96.1-96.2
Dynamic range, dB (A): 95.795.6
THD, %: 0.00650.0081
IMD + Noise, %: 0.00610.0078
Stereo crosstalk, dB: -50.2-50.7

Schematic Diagram

100 kHz Response

100 kHz Response with RL=10 ohm showing input and output

Circuit Layout

Frequency response

Spectrum graph

Noise level

Spectrum graph

Dynamic range

Spectrum graph

THD + Noise (at -3 dB FS)

Spectrum graph

Intermodulation distortion

Spectrum graph

Stereo crosstalk

Spectrum graph

This report was generated by RightMark Audio Analyzer 6.0

Biased Push-Pull Output Transistors

The above results operate the output push-pull transistors at zero bias and rely on the high slew rate of the opamp to reduce crossover distortion. While the results above are quite good, improved distortion can be obtained by biasing the output transistors as shown below:

Double 1N4148 diodes are used to provide a better match to the power transistors. The quiescent collector bias with the components shown is about 15 mA. Biasing the output push-pull transistors lowers the THD from 0.0081% to 0.0026% and the IMD+Noise from 0.0078% to 0.0057% for the 10 ohm load case. (Results for the LM4562 opamp follower alone are THD 0.0018% and IMD+Noise 0.0055%, limited by the noise and distortion floor of the sound card used here). The THD spectrum shown below for the biased output case clearly shows the reduced harmonic distortion products compared to the simpler unbiased results above:

Bandwidth and Slew Rate Discussion

The above results measure the LM4562 "high fidelity" audio opamp with a discrete transistor output power booster stage over the audio bandwidth of ~ 10 Hz to 20 kHz at an input signal level of about 1.8 Vrms (7.3 dBu) and a 10 ohm load (load power ~ 320 dBm). While the emphasis in the RMAA measurements above is obviously on audio spectrum performance, it is interesting to consider the frequency and signal limitations for this simple circuit. The unity gain follower opamp alone has a 3 dB bandwidth of about 50 MHz (the GBW of the LM4562 alone). The power transistors used here have an ft of 50 MHz. The achieveable bandwidth of the push-pull emitter follower stage is determined by the input RC lag network consisting of the power transistor input and feedback capacitances, transistor beta and the input source resistance and emitter load resistance. An analysis of this emitter follower stage predicts a 3 dB bandwidth for the output stage of about 8 MHz for a 10 ohm emitter load and 29 MHz for a 50 ohm emitter load. Since the slew rate of the LM4562 is 20 V/us, for Vp = 2.65V, slew limiting will occur at f ~ 1 MHz as confirmed by viewing the ouput directly. At a lower input level of 0.5Vp, the output frequency response was well beyond 3 MHz, consistent with the predicted 8 MHz bandwidth, as shown below confirming the high bandwidth available with these power transistors:

400 kHz response LM4562 + biased push-pull current booster; 10 ohm load; 0.5 Vp input

3.0 MHz response LM4562 + biased push-pull current booster; 10 ohm load; 0.5 Vp input