The amplifier employs an all-discrete, direct-coupled,
fully-complementary
topology featuring a JFET differential input stage with cascodes and
constant current sources. A voltage amplification stage
using two paralleled pairs of BJTs follows, and four parallel
pairs of power BJTs serves as the push-pull output stage.
A DC servo keeps output DC offset in check.
All amplifier transistors are from Toshiba. The differential
stage JFETs are hand-matched for Idss
and the complementary BJT pairs are hand-matched for
hfe.
A companion two-stage power supply (also designed by Kevin Gilmore)
is used. Two LM338T voltage regulators
supply a pair of OPA541AP power opamps, which then drives the
positive and negative power rails for the amplifier. The positive
rail is referenced to a REF02 chip and the negative rail is
configured to track the positive rail.
The power supply and each amplifier channel dissipate approximately
20 watts of heat at all times. There is also some heat from the
power transformer.
This unit uses the printed circuit boards designed by Kevin Gilmore.
Separate cases are used for the power supply and the amplifier,
to prevent the power transformer from inducing hum and noise into
the amplifier.
Overkill-size finned heatsinks are used in both units,
allowing all devices operate warm but not scalding hot.
The heatsinks form the sidewalls of each case, which are custom
home-made with an aluminum base plate,
acrylic front and rear panels finished in matte black (but with
frosted/translucent edges), and a transparent acrylic top cover.
The volume control potentiometer is centrally-mounted between the
two amplifier boards, so that the wiring length from the potentiometer
to the boards are kept to a minimum. A shaft-extender and panel bearing
are used to link the potentiometer to the front panel knob.
This test provides data and graphs of frequency response, noise, dynamic
range, total harmonic distortion, intermodulation distortion and
stereo crosstalk performance.
The same measurement methodology is employed for all of my RMAA tests.
Namely, the amplifier-under-test has its volume control set to maximum
position, and the test levels are adjusted using the sound card software.
The amplifier outputs are loaded with a custom switchable dummy load box.
RMAA test result comments
The Dynahi measured well, producing excellent THD and very good IMD
results. The THD and IMD are largely unaffected by the load, recording
virtually identical numbers and graphs for 330Ω and 33Ω loads,
and deteriorating somewhat with 8Ω load.
Thanks to the separate PSU chassis construction, the noise spectrum
is devoid of any spikes that would indicate AC hum interference.
The broadband noise floor is low, but not the lowest of all the amplifiers
I've tested.
The stereo crosstalk is excellent with 330Ω load, but impaired by
some 8dB or more above 1KHz when loaded with 33Ω, and deteriorating
an additional 10dB over much of the audible spectrum with 8Ω load.
This is not surprising for a passive-ground topology. Nevertheless
the 33Ω result is still very good thanks to careful headphone
jack ground return wiring (See
my discussion on this subject at the
Headwize DIY Workshop).
Other Test Results
These were measured using a Wavetek 188 4MHz sweep function generator, a
Fluke 95 50MHz digital ScopeMeter and a Protek 6510 100MHz oscilloscope.
Maximum output voltage
(prior to onset of clipping)
17.8Vrms (50.3Vp-p) into 330Ω
15.0Vrms (42.4Vp-p) into 33Ω
12.4Vrms (35.1Vp-p) into 8Ω*
Maximum output power
(prior to onset of clipping)
0.96Wrms into 330Ω
6.8Wrms into 33Ω
19Wrms into 8Ω*
Frequency Response
(at 1Vrms output, sine wave)
0.4Hz to > 4MHz; +0, -3dB
Note: The frequency response begins to rise above 500KHz,
reaching +3dB at 2.3MHz and continues to +4.6dB at 4MHz (the
upper limit of the function generator).
Rise time
(1KHz square wave, at maximum output voltage, 10% to 90%)
0.44µS
Slew rate
(1KHz square wave, at maximum output voltage)
72V/µS
Output impedance
0.4Ω at 1KHz
* Sustained maximum output into 8Ω is not recommended due to
PSU regulator heat dissipation concerns.
Oscilloscope waveforms
The following shows the waveform response of the Dynahi amplifier.
In all graphs except the Lissajous waveform, the top trace is the input
and the bottom is the output. These were measured using a Wavetek 188 4MHz
sweep function generator and a Protek 6510 100MHz oscilloscope.
The tests were done with the amplifier volume control set to maximum,
and the output level is adjusted to slightly below the threshold of
clipping using the function generator's amplitude control.
The graphs show that the amplifier preserves absolute phase.
There is a very slight overshoot at the leading edge of the square wave
response. The rising edge of the square wave has a peculiar uneven slope.
When the output amplitude is reduced to one half of maximum, the
slope shape straightens out, but there is pronounced ringing at the
edge. These are illustrated in the expanded 100KHz graphs shown below.
These characteristics are not unique to this build. Another
Dynahi (built by dgardner) also exhibited similar behavior.
The ringing and the rising frequency response above 500KHz suggests
that the 33pF compensation capacitor is not sufficient. Increasing
the value of the capacitor would stabilize the amplifier at the expense
of slew rate and bandwidth.
There is minimal phase shift even at 100KHz, as shown in the Lissajous
graph.
The output waveforms are somewhat affected by the load. The graphs
below are taken with 330Ω loads connected to the output.
There is a noticeable rounding of the square wave edge, and a slight
reduction of voltage amplitude when a 33Ω load is switched in.