DuKane Ionovac
The History of a Brilliant Failure
Copyright March 2004
In 1954 the DuKane Corporation in
Siegfried
Klein, the inventor of a practical plasma tweeter, wrote to Mr. William Torn, DuKane’s Chief Engineer for Commercial Sound, and made a
proposal for DuKane to manufacture and sell what he
called an Ionophone. Torn was interested and invited
Klein to present a demonstration. On his first interview with Torn he carried a
sample of his speaker in a bowling bag. Torn was impressed and convinced the
management at DuKane that this would be a marketable
product. DuKane named the product “Ionovac” and set up a new division to develop and
manufacture it. Torn was so excited about the project he asked to be assigned
as Division Manager. Some of the project engineers under him were Mr. Charles
Swisher, Mr. Wilbur Ogden, Mr. George Mavrogenes, Mr.
Charles “Ben” Jewison and Mr. Art Block who joined
the division when
Siegfried
Klein was contracted as a consultant to develop the Ionovac
and to assist in other projects. Klein was a rather colorful character with
long red hair and the ‘absent minded professor’ manner of an inventor. The DuKane people nicknamed him “Fiti”
(pronounced feetee). He did much of the design work
on the initial Ionovac. Klein’s demonstration speaker
had a number of problems. It had too much distortion,
it was noisy and tended to be unreliable in initiating the plasma when first
turned on. George Mavrogenes worked with Klein to develop the inventor’s
model into a saleable product. After his two year contract was up Klein
returned to his home in
The
first cell design evolved into a cylindrical hollow chamber made of a high
temperature glass called Vycor®. Vycor® is a 96% silica glass used for heater sheath,
incandescent lamp envelopes, defroster tubing and other high temperature
applications. The cell was made by starting with Vycor® tubing cut to length and
placing the electrode inside. The tubing was heated and a neck was formed into
the central portion to clamp down on the electrode and approximate the first
section of the acoustic horn on the open end. Wire was wrapped around the neck
portion to make the outside return electrode. These cells were made by hand and
no two were alike.
Along
with the cell design, an electronic circuit was needed to make it function. To
provide a high voltage signal to create the plasma, a radio frequency power
oscillator was used. 27.12 megahertz was chosen as the operating frequency.
This frequency was high enough such that the tuning components were relatively
small and it didn’t require exceptionally tight shielding. At that time
operation of this kind was allowed in the 27 megahertz citizen’s band with
looser radio frequency leakage requirements than at other frequencies. Any
interference was limited to remote radio controlled devices and shielding was
not likely to be a major problem.
The first production model Ionovac had the power supply and oscillator in one metal enclosed chassis. The speaker horn and cell were in a separate package connected to the power unit by a two conductor shielded cable several feet long with a connector on it for easy disconnection. Radio frequency energy from the power oscillator was transformer coupled to the shielded twisted pair cable and then transformer coupled again in the reproducer unit to maintain a true differential drive over the cable. This helped to minimize losses over the cable run from the oscillator to the reproducer and allowed the horn unit to be tuned for maximum efficiency.
Initial
designs showed a difficulty in keeping the operating frequency in tune. The
cell was the capacitive part of a parallel tuned transformer circuit in the
horn. Once the plasma initiated, an event called “striking” or “igniting”,
there was an initial shift in the cell coupling due to the change in impedance.
Furthermore heating tended to change the capacity of the cell and that detuned
the frequency of the cell/transformer circuit at the horn. As the tune of cell
frequency shifted with heat, the oscillator frequency, remaining at a fixed
value, became mismatched to the cell resonant frequency so it did not function
well and distortion was introduced. Tuning the oscillator to match the cell
after warmup caused the speaker to be unreliable or
fail to start at all when it was cold.
To
remedy this a device was used to slowly adjust the
oscillator frequency along with the change in the cell as it heated up. A
thermally sensitive bimetal strip was mechanically coupled to the tuning
capacitor that determined the oscillator frequency. A heater powered by the
filament transformer was wrapped around the bimetal compensator. As the bimetal
strip heated up from a cold start it deflected an actuator adjusting the tuning
of the capacitor in the oscillator.
When
the speaker was first turned on the tubes warmed up to operating temperature
and the oscillator began to produce radio frequency energy. At this point the
oscillator frequency is tuned to match the cold cell. When sufficient RF energy
was available, the air in the cell ionized and heated to a
hot plasma causing the cell to heat up. As the cell heated it changed
capacitance and shifted the resonant frequency. The slowly warming bimetal
compensator adjusted the oscillator frequency to compensate for the changing
cell capacitance as it warmed up. This method worked better but it was still
not the best solution.
There
were problems with the cell design. It still had startup problems and generated
noise after deposits formed on it from ablation of the inner electrode. If the
plasma did not ignite within a short time after the oscillator began to develop
radio frequency energy it would never start. This is because the compensating
heater was adjusting the oscillator to match the cell, but since the cold cell
was not changing, the oscillator was being pulled farther away instead of
tracking the cell. Startup could only be tried again after shutting off the
power and allowing the heater to cool down to room temperature again.
This
first model used a type 6146 beam power tube as the power oscillator. A 12AU7
dual triode was used as a two stage audio amplifier and modulator. The first
triode stage was an amplifier/DC level shifter with the plate direct coupled to
the second triode grid. The second triode was a series pass voltage
amplifier/modulator with the cathode direct coupled to the screen grid of the
6146 oscillator. DC supply voltages were derived from a power transformer and
5U4GA full wave rectifier tube.
As part of the research in
developing the Ionovac, special test equipment was
purchased and set up in a testing lab. Special prototypes were tested out to 400KHz using ultrasonic microphones capable of response out
that far. After testing and evaluation the speaker design was specified at a
crossover frequency of 3500 hertz so that the horn could be kept at a
convenient size. Also above 3500 hertz there is only about 3% of the total
energy in the typical program source. Since the Ionovac
was not capable of high power levels it was left to conventional midrange and
low frequency drivers to do the heavy work and let the Ionovac
reproduce what it did best at frequencies that would not overload it. Because the production Ionovac
used a transformer to couple the audio into the speaker electronics the upper
frequency response was limited to what the transformer could reasonably pass.
In
mid 1957 a splash of magazine articles announced the new plasma speaker. A
production run of 500 of these speakers were made. The first of the DuKane
production Ionovac speakers were marketed at $147.00
each under a cooperative agreement with Electro-Voice. It was called the model
T-3500 because it was a Tweeter and used a crossover frequency of 3500 hertz.
Electro-Voice coupled it to their top end Patrician speaker as the tweeter in
the system. It sounded great when it worked but problems remained. Cell
failures were excessive and problems remained with keeping the oscillator in
tune. A July 1958 consumer magazine article raved about the sonic quality but
rejected the T-3500 on the basis of the unreliable operation.
A
second experimental cell design was proposed to replace the first one. The new
cell used a ground sleeve with a removable electrode inside. Apparently the new
design suffered similar problems with electrode erosion and cell deposits. A
consumer magazine article dated May of 1959 described the improvement as “still
a brilliant failure”. The T-3500 was dropped a short time after that.
After
the poor reliability of the T-3500, DuKane continued
with an intense research effort to eliminate the problems and make
manufacturing easier and cheaper. One of the first problems was to redesign the
electrode and cell so that they would work efficiently with a reasonable life. DuKane partnered with the Illinois Institute of Technology
Armor College of Engineering, Materials Engineering Department to find more
reliable cell and electrode materials.
DuKane engineer Ben Jewison did testing on the cell
and electrode material system. Ben had a room full of Ionovac
units running burn-in tests on all kinds of metallic alloys. Considerable
research was done on the shape of the plasma chamber inside the cell. Small
changes in the size of the chamber, the slope of the inner walls and length had
audible effects on the reproduced sound. One goal was to produce as flat a
frequency response as possible from the crossover frequency to the upper limit.
Plus or minus three decibels was about as close as they could get. Ideally an
exponential shape from the tip of the electrode to the exit opening would be
desired. In practice it was found that a segmented wall of two straight conical
sections with a break point in the middle of the chamber was acceptable for
good frequency response and flatness.
High
purity alumina was tried as a cell material and it worked fairly well for a
short time. It was easy to form and did not have the disadvantage of requiring
expensive machining. It had a problem though that caused it to be noisy. As
long as the cell was clean it worked well. After a few hours of operating time
deposits formed on the inner wall of the cell from vaporization of the center
electrode causing the plasma to deform. These deformations created hot spots
leading to a breakdown in the cell wall and seriously affected the performance.
After
additional research high purity fused quartz was selected for the cell
material. Quartz has an expansion with temperature of almost zero. It is strong
and it resists heat and temperature cycling extremely well. It has a good
dielectric characteristic and a high breakdown voltage. The main drawback is
that it is hard and brittle and had to be ground to shape rather than molded
like alumina ceramics. Grinding quartz to precise dimensions was difficult and
time consuming by conventional methods. To solve this problem William Torn
designed and patented a means of grinding quartz blanks inexpensively using
frequently replaced threaded ultrasonic grinding bits.
The cell design was changed considerably. A quartz rod was centerless ground to three tenths of an inch in diameter and cut to length to make a blank. Each blank was placed in an ultrasonic grinding fixture. The blank was flooded with abrasive slurry and a grinding bit was pressed onto the end. Ultrasonic energy applied to the chuck holding the bit caused an abrasive action to form and the bit would grind its way into the quartz. The bits were manufactured on a screw machine, which is a type of lathe used for small precision parts, to the dimensions that were desired for the inner surface of the quartz tube. As the bit completed its travel, the inner surface of the quartz took on the exact shape of the outer surface of the bit. The actual production of the cells was contracted to an outside supplier.
The
new electrode material was determined by trial and error. The ideal material
would stand up to the high temperatures and adverse conditions of exposure to
the plasma without evaporating and causing deposits to foul the inner wall of
the cell. It should also be relatively inexpensive and easy to form into the
desired shape. Jewison tried many materials and many
hours were spent in operating tests and examining samples under a stereo
microscope. One alloy that seemed promising belonged to a family that was
popularly used in automotive cigarette lighters and as the center electrode in
spark plugs. After testing various alloys to see which one
would give the best results an alloy consisting of 80.7% iron, 15% chromium and
4.3% aluminum called Alkrothal 14® was chosen.
The
physical design of the electrode needed a sharp point to initiate the plasma
inside the cell and also must not carry away too much heat from the tip. High
temperatures were necessary to maintain a proper plasma or instability and
spontaneous noise generation would result. A sharp edged point helped to
increase the voltage gradient between the electrode and the outer metallic
sleeve around the cell to greatly enhance initialization of plasma. A conical
end point helped to prevent the plasma from walking around the tip and spread
it out evenly.
The new electrode had a sharp edged
conical tip blunted at the point. The tip section was thermally isolated from the
main electrode body by a narrow stem. This helped to maintain the tip at a high
operating temperature. The electrode and cell were assembled in two separate
pieces and remained separable for easy cleaning and replacement. Electrode
manufacture was contracted out to a local machine shop in
The
holder that the cell was supported in at the narrow end of the horn had to
withstand the high temperatures that the cell operated at. It had to be a good
insulator of heat so that the cell would maintain the high temperature and also
a good electrical insulator for the outside electrode. Several ceramic
materials were tried and a suitable ceramic was chosen.
The
cell was retained in place using a spring loaded bar made of a glass bonded
mica material called Supramica®. The bar was made very long and narrow to help minimize heat loss and
capacitive coupling to the radio frequency circuit. A metal cup in the center
of the bar made contact with the electrode holding the electrode in place and
completing the electrical circuit. The cell was easily removed for service by
drawing the bar back with one hand and taking the electrode and cell out of the
holder with the other hand.
The new production design used a type 6DQ6 beam power tube as the power oscillator. The 12AU7 amplifier/modulator was replaced with a modulation transformer connected directly to the screen grid of the oscillator. The power transformer with the vacuum tube rectifier was replaced with a silicon rectifier voltage doubler supply operating directly off of the AC power line. The filament transformer was retained.
The 6DQ6 tube was chosen because it was a common type used in black and white television sets as a horizontal output amplifier. It was easy to drive, rugged and inexpensive. Screen grid modulation of the oscillator was used as it was in the T-3500. The secondary winding of the modulation transformer was connected in series with the screen voltage supply. In this way the audio either added to or subtracted from the screen voltage causing the radio frequency amplitude on the plate to vary in direct proportion. The turns ratio of the transformer was such that the voltage applied to the screen grid would remain within the linear region of oscillator operation with an input voltage of .75 volts RMS or about 1.06 volts peak.
The oscillator circuit was simplified. The power supply and power oscillator were split into two separate units connected by a four conductor cable and ground strap to supply the DC operating voltages to the oscillator. Since the cell and horn were part of the oscillator chassis, the radio frequency generated by the tube was connected directly to the cell without the need for any intervening umbilical cable or transformer coupling. This allowed the problem of tuning shift as the cell heated up to be solved by making the cell capacitance part of the frequency determining circuit in the oscillator. The frequency shifted as the cell warmed up but now it did not matter since the capacitance of the cell altered the oscillator frequency directly and did not affect the efficiency of coupling the oscillator energy into the cell as it did before.
Although
DuKane advertising claimed a guaranteed operating
life of 1200 hours or one year whichever came first on the new cells, Mr. Jewison does not recall ever performing long term life
tests on electrodes for this purpose. Although service life was a factor,
reliable ignition and clean quiet operation was of more interest. The
speculation is that the advertising people needed a number to quote and an
average 100 hours a month sounded good. For the reader’s reference, I life
tested samples of electrodes I had specially made for experimental purposes and
they operated for over 2500 hours. As Mr. Jewison put
it 2500 hours was “doing pretty well”.
Once
a working design was accepted, Jewison performed
radio frequency leakage tests on the new package. Special test equipment was
bought and set up to measure the leakage in order to qualify for FCC approval.
Some of the tests were set up on the roof of the DuKane
factory and some were done out in an open free field area so that interference
from reflections off of metal objects and structures could be avoided. Acoustic
testing for the Ionovac and associated full range
systems was done in special acoustically quiet rooms, called anechoic chambers,
at DuKane, Quincy Speaker in
Two
models of the new Ionovac were built beginning in
late 1960. Both of them used identical circuitry and hardware with a few minor
differences. The 14A430, called the basic model, had the power supply totally
enclosed in a metal box. The oscillator horn chassis had a six foot connecting
cable with a plug on the end of it that plugged into a mating socket on the
power supply. The 14A435, called the system model, had the power supply mounted
on an open face metal plate. A sixteen inch permanently attached connecting
cable bridged the oscillator horn chassis to the power supply. The 14A435 had a
power switch and the 14A430 did not.
Later
production models designated 14A430A and 14A435A were identical in every
respect to the earlier versions with the exception that a radio frequency choke
was added to the screen grid circuit in the oscillator chassis to reduce RF
leakage out of the oscillator. A letter dated
The
new design was first demonstrated at the New York Hi-Fi
show in September of 1960. As Ben Jewison put it, the
new Ionovac was a big hit. It was far more reliable,
it had a clean pure sound that impressed the attendees and got good reviews
from the audio publications.
Research
continued to try and develop a higher power Ionovac
that could be used down to 800Hz. Larger cells and horns
were experimented with but no products came out of it. When the cell design was
scaled up problems developed with initiating and maintaining a
stable plasma. Larger electrodes and cells did not confine the plasma as
in the smaller cells and stability was poor. A few special design Ionovacs were built under contract for medical research
using ultrasonic sound waves. Charles Swisher spent some time following up on
the inventors claim that the plasma speaker could be used as a high quality
microphone. Experimental results were very disappointing and the project was
dropped.
While
development was going on with the new Ionovac design,
William Torn headed up a design effort to match the plasma tweeter with
conventional midrange and woofer speakers. Torn designed midrange and woofers
that were sent to
§ DUK-5 A basic model 14A430A Ionovac tweeter. Frequency response was claimed at 3500Hz to 30Khz +-3db and the entire unit weighed 4 pounds. It sold for $69.50.
§ DUK-10 A system model 14A435A Ionovac unit packaged in a wood cabinet with a plastic waffle grille. Frequency response was claimed at 3500Hz to 30Khz +-3db. It measured 13.5 inches high, 5.25 inches wide by 11 inches deep and weighed 7.5 pounds. It sold for $79.50.
§ DUK-15 The same as a DUK-10 with the addition of a 9A775 crossover kit factory installed in the back of the cabinet. It sold for $99.50.
§ DUK-20 DuKane’s larger “bookshelf” full range system that could be used alone or placed horizontally free standing on the floor with the optional 22 inch high leg kit. It was rather big for the typical bookshelf at 30 inches high by 14 inches wide by 13 inches deep. This one was a three way using a 14A435A Ionovac tweeter, two 3.5 inch closed back midrange and one 12 inch woofer. Frequency response was claimed at 40Hz to 30Khz with no limits of deviation given. It sold for $187.50. The optional legs were available for $15.00.
§ DUK-30 The mid size of three column systems available. The speaker lineup was the same as the DUK-20 put into a larger corner fitting column. The 12 inch woofer and two 3.5 inch closed back midrange drivers were mounted in a sealed ducted port lower section. A 14A435A Ionovac was in a separate open upper section with the characteristic waffle grille in front. A 9A695 two way crossover network with an adjustable L-Pad on the tweeter feed was mounted in the upper chamber on the back panel. The 9A695 crossover had a similar construction to the 14A435A power supply. It was assembled on an open face plate that was mounted flush in an opening of the back of the upper section. It had a terminal strip on the back for connection to the amplifier and pigtail wires for connections to the speakers. A separate high pass network was connected to the midrange speakers in the lower section with a crossover of 800Hz. Frequency response was claimed at 40Hz to 30Khz with no limits of deviation given. The cabinet was 48 inches high, 15.375 inches wide by 11 inches deep and weighed 45 pounds. It sold for $199.50.
§ DUK-40 This was the smaller tower system. The construction was similar to the DUK-30 but it was only two way. It used a 14A435A Ionovac above and a single 8 inch woofer below. Frequency response was claimed at 55Hz to 30Khz with no limits of deviation given. It measured 37 inches high, 12.5 inches wide by 9.5 inches deep in a corner fitting column. It sold for $149.50.
§ DUK-50 The smaller two way bookshelf system with an 8 inch woofer and a 14A435A Ionovac. Frequency response was claimed at 55Hz to 30Khz with no limits of deviation given. It was far more suited for bookshelf placement measuring 11.5 inches high by 24 inches wide by 12 inches deep and weighing 30 pounds. It sold for $139.50
§ DUK-60 This was the big daddy top of the line tower system similar to the DUK-30 but with a larger midrange driver and an unusual compliment of two woofers. It used a 14A435A Ionovac up above in the open section. The lower section used one 8 inch midrange with it’s own level adjustment. Also down below were an 8 inch woofer and a second 12 inch woofer to round out the response. The frequency response was claimed to be 40 Hz to 35KHz +-4db with crossovers at 3500 and 800Hz. The cabinet measured 51 inches high, 21 inches wide by 13 inches deep and weighed 68 pounds. It sold for $246.00
§ 438-37 Replacement Cell kits sold for $9.00
§ 9A775 Crossover Network kit was a two way crossover for adding an Ionovac tweeter to an existing system. The kit consisted of a 9A770 3500Hz crossover and a 99A235 8 ohm L-Pad. The 9A770 crossover was totally enclosed in a metal box and had a set of screw terminals to attach the speakers and amplifier. The kit sold for $20.00.
§ 438-44 Custom Wood Grille kit was offered for those who wanted to build their own enclosure, as many did in those days. It consisted of the characteristic white plastic waffle grille and an attractive wood frame to hold it. The kit was designed to mount in the rough opening of a home built enclosure or room wall. This was a nice addition to the product line because it allowed the Hi-Fi owner to add an Ionovac to a custom speaker system and maintain a professional finished off appearance. The grille sold for $7.50.
The
tower systems were shaped with sloped sides in order to fit into a corner.
William Torn had heard that new apartments were smaller and decided that a
speaker design utilizing the otherwise wasted space in a corner would be more
attractive to modern city dwellers. The corner placement saved space and
enhanced the bass response of the systems. Advertising brochures and magazine
ads showcased the speaker placed in corner settings and pointed out the
efficient use of corner space that the DUK towers could provide.
All
of the cabinets were available in several types of genuine hardwood veneers.
The choices included: walnut, cherry, mahogany, blond
and (except for the DUK-60) unfinished birch. A gold sticker prominently
applied to the cabinet proclaimed “Fine Hardwoods Association Official
Certification - Genuine Hardwoods” signed E. H. Gatewood
secretary. The finish was described in the sales brochures as “modern decorator
tones” whatever that meant. The dealer merchandising kit proclaimed
“Ear-thrilling for the men, eye appealing for the ladies”. Try that one on the
public today and see what happens.
The last production run of the new Ionovac design was made in August of 1961. It is unknown how many were made but serial numbers up into the 1600’s have been seen on surviving units.
The first Ionovac
speaker sold for $147.00 in 1958. For the time that was a lot of money. Owners
were very pleased with the sound but the price, reliability and maintenance
requirements dampened initial public acceptance. Even after improvements made
the Ionovac more reliable DuKane
was disappointed with sales and hesitant to continue with production. Rather
than continue development to lower costs and improve reliability they decided
to dissolve the Ionovac Division in late1962.
Remaining Ionovac systems were sold off direct to the
public from the
Wilbur Ogden transferred out of the
Ionovac Division in 1960 to head the DuKane Special Products Division, working on specialty and
contract work. One of those contracts involved the design and development of
smoke detectors.
Ben Jewison was with the Ionovac division from 1957 until it closed in 1962. He transferred to the Audio Visual division and retired from DuKane in 1997.
The author would like to extend his sincere thanks to Mr. Wilbur Ogden and Mr. Ben Jewison for the original information they provided in personal interviews and DuKane documentation related to Ionovac history.
Alkrothal 14 is a registered trademark of
References: