Full Size Corona
Before progressing with the corona wind I could see a need for
"more power". I had reached the limits of what my power supply could provide and amplifier could
sink. After false starts with several (expensive!) high voltage supplies, usually too underrated for
current, I took the decision to fork out for a proper piece of kit. I purchased a digitally controlled
1.5kW 30kV DC power supply. It provided a pure DC output with a short circuit protected 50mA current
output. This still need an additional RC network to balance its response to the striking load of plasmas
and prevent it responding to the audio load, but it meant I had the maximum available supply - two plugged in
for stereo would be the maximum from a UK mains socket. It was programmable for voltage or current
limit giving me maximum flexibility.
To cope with the new power possibilities I needed an amp to match. The basic amplifier design had not
changed, I had gradually extended the same basic stacked FET single ended design. I just need it to be able
to cope with the power so I built a 12kV (12 FET) design with nice large heatsinks). The 19" rack kit that
resulted is pictured. Amp is at the top, the control module in the middle and the white 30kV supply at the
I went through a number of design processes whilst considering the best way of scaling up the corona wind
design. It was out of the question to wire up each pin to an individual resistor as I was looking at
thousands of pins. So I decided on a design using many bars with pins on, with a resistor to each
beam. Another array of beams would then be placed at 90 degrees facing that one so each pin had its
own current path.
One effect that the small 100 pin version had was that of beaming. Because the sound was generated
uniformly across the flat array it caused the sound to be produced in a very directional way. I realised that
a larger array would beam even more and may be an unwanted effect. I designed the electrodes of my new array
to be curved so that the wave front would be less directional. However, once designed I realised I would be
spending a lot on an unproven method so went for a flat array thinking I would do the curved version once
proven. Rather than a bespoke set of array electrodes I found an alternative, bow saw blades.
I think even with a picture it is still difficult to describe. Here you see the
basic assembled corona loudspeaker. The grey square frame is over 1 metre square and there are two
frames one behind the other. The front one has vertical blades with the points facing away from
us. The back frame has horizontal blades with the point facing towards us. There is a gap of
around 1cm between the two sets of points. The orange band on the right is the limiting resistors,
that set of resistors got replaced eventually with more expensive HT ones. The black bands are tie
wraps - used to hold each blade in place and tension them. The grey frame is very strong and insulating
fibre glass square tubing.
The front array was attached to 30kV, the rear attached to the amplifier - one loud speaker. I spent a
long time with this and worked on it for probably around 2 years in the end. For a start it worked. It
played music at a reasonable sound level - loud talking at best - with a reasonable frequency
response. It beamed very strongly and you had to sit right in front (at any distance) to hear the full
frequency range and volume. It produced ozone gas very efficiently in large quantities and had a background
hiss and crackle with constant issues of arcs forming - which as it was corona wind were very unwanted. The
arcs stress the balancing resistors too much, encouraging smoke and flames.
The ozone problem was a big one. Within 5 minutes of being switched on, the
room in which the speaker ran was filled with ozone beyond safe levels. Running the speaker with the
window open was not a good option and after trying various extraction methods I found some really useful
material. A company in Taiwan makes activated carbon cloth, mainly for use in biochemical suits.
This material is a polyamide cloth that has been burnt in an oxygen free atmosphere. It is quite
delicate but very efficient at adsorbing ozone. As it was a thin material I could drape it around the
speaker without blocking any sound.
Right is a picture of the full assembly, in a chipboard box with the activated carbon cloth front and
back. The amp pictured is a slightly earlier version of the one above which used slow fans for cooling (not
ideal). This took up most of my living room for quite a while.
As I say, I spent a long time on this build. Various problems were sorted but the biggest one was the
background noise. Being larger with more electrode pins just meant that the noise was amplified. Even
though the amplifier has an active feedback effect that helped damp a lot of noise electrically the noise of the
ions driving the air was always noticeable. When clean and with insulated coatings on the blades the noise
would be reasonable but over time the high voltage breaks down most insulation, attracts dust and encourages
oxidation of the tips. This all goes to increase the noise over time. Also however much voltage and
power I threw at it I couldn't get this huge speaker to go much louder than normal speech. Partly due to my
use of the wasteful single ended design but mainly due to the limits of the corona wind technology.I
eventually, and very reluctantly, called it a day on this build. I was failing to get the volume up and
noticed that any plasma arcs that started inadvertently were much louder. Therefore I changed my focus to
plasmas. I didn't want to build just another plasma tweeter and my main intention was to try and build a full
frequency range plasma speaker.