ACCE3 CFA - current feedback power
amplifier
This project is dated of 2016. I started it for the reason that I
was interested in the current feedback topology behavior. The
current feedback concept is well described in Renesas company
application notes. For those who are interested, I recommend to
download the application notes that I have linked below. I could
hardly say anything more.
https://www.renesas.cn/cn/zh/document/apn/an9420-current-feedback-amplifier-theory-and-applications
https://www.renesas.cn/cn/zh/document/apn/an1993-voltage-feedback-versus-current-feedback-operational-amplifiers
Amplifier schematics
is as follows. Not shown is the DC servo that keeps amplifier
output DC voltage close to zero. The circuit was inspired by
Accuphase topology used in E-306 and E-406 amplifiers.
ACCE3 CFA
Accuphase topology:
The goal was to design a simple and robust amplifier that would
be able to drive difficult loads, would be stable and would have
balanced time and frequency domain characteristics. The lowest
nonlinear distortion was not the design goal, just to keep it
inaudible.
The amplifier has now been reliably operated for 9 years, without
a single problem.
Output devices are robust OnSemi MJL21194/93. Drivers are MJE15030/03.
Similarly as in case of my PA4 amplifier, the rugged and reliable
amplifier was my goal.
The amplifier is strictly dual-mono, channels do not share any
common ground point, so no ground loops are created.
Photos of the amplifier
Front view (silver panel)
Rear view
Teardown
Basic specifications
Distortion measurements
1. THD+N vs. output power into 4ohm at various frequencies, @BW=20kHz
The amplifier distortion is almost independent of signal
frequency and the maximum power does not change with signal
frequency.
2. THD vs. frequency into 4ohm load at 3.5W - 55W
THD slowly rises above 4kHz. The profile remains almost unchanged
with output power.
The amplifier is fast, with nice aperiodic step response and
short rise time, which are the benefits of the CFA topology.
It was also suggested to post distortion
residual instead of distortion spectrum. So, I am posting both.
Distortion residual is obtained when fundamental test frequency (1kHz
here) is notched (filtered out) from the amplifier output. The
method has been usual since old days of audio and still sometimes
used in Stereophile. I am posting both methods below.
3. Distortion residual shape in time domain
4. Distortion residual in frequency domain
5. Complete spectrum with 1kHz test signal
Test with capacitive load
Another test I always perform is a THD vs. power and THD vs.
frequency with capacitive load 2.2uF added in parallel to the
load resistor. I am showing results with 4.7ohm load compared to
4.7ohm//2.2uF load below.
6. THD vs. output voltage at 1kHz
The plots are identical, there is no difference if the capacitor
is added or not.
7. THD vs. frequency at 7.41Vrms output voltage
We can see some small rise of distortion above 5kHz when the
capacitor is added.
These results are excellent and confirm stability with complex
load and overall robustness of this topology.
This is a very different story when we compare it to results
obtained with class D top performers Hypex and Purifi. These are
the results measured previously with Hzpex NC252MP and Purifi 1ET400A.
These amplifiers have big problems with capacitive load, as I
have shown many times at ASR. This applies to most of class D
amplifiers and to me it is an important reason to keep my
preferences of class AB amplifiers. Side by side with continuous
output power and long term reliability.
FYI, Accuphase E-306V datasheet:
https://www.accuphase.com/cat/e-306ven.pdf
@PMA 2016 - 2025