- Power consumption.
By setting the vacuum tube-based unit to a flame
height of 1cm, it draw 114 watts of power (corrected to exclude filament
power, fan power, and power supply losses). Our FET-based project drew
only 67watts for the same 1cm flame. This is a power savings of 40% for
the same output.
- EMI testing.
We had two facilities lined up to perform this sensitive test, but both
opportunities fell through, so we were unable to measure this
quantitatively to see if it met FCC regulation 15.223. We demonstrated
how our Faraday shield worked by lighting a fluorescent lamp a few inches
from the coil and then showing that the tube could not be lit above the
faraday shield, so it effectively blocked the electric field. we also
made a coil shield using a short section of PVC pipe and an aluminum
gutter screen.
- Flame stability.
The corona flame is very stable in a breeze and cannot be blown out.
Blowing on it produces a hissing sound, however.
- SPL output.
The output level was somewhat dissapointing using the power rail
modulation technique. This was because we only used a +/-15v power supply
to drive the toroidal audio transformer. This provided a peak AC
signal of ~15volts on top of the 150v power rail. We think that Pulse
Width Modulation would be the ideal way to get the sound into the flame,
since the PWM would drive the gate driver directly (very high impedance),
so it would take virtually no extra power, whereas the power amp draws
around 25watts to drive the power rail. Using a sine wave (which has a
crest factor of CF = Vpeak / VRMS = 1.414, we can
get an output of ~85dB ad 2". Using music, which has a typical crest
value of CF=6, the output is around 70dB. Typical listening would require
around 100dB of dynamic range.
- Frequency Response.
The Audio Contorl subsystem limits the signal to 4kHz to 40kHz, but we
believe the actual output would extend down to ~2kHz, if a sufficiently
large flame was used.
- Harmonic Distortion.
This test was not performed due to lack of equipment. We think that the
distortion would have been unacceptably high, however (~5-10%). The
distortion figure depends on the steady-state flame height as well.
- Subjective testing.
The treble was grainy as restrained in comparison to the vacuum-tube
version. We attribute this to the fact that our project uses the FET as
an on-off switch to maximize power efficiency, whereas the vacuum
tube-based unit operates in a near class-A method, which lends itself to
audio.
|
