Spark Gap Generator – Part1

In the early days of my research, and before we built the spark gap generator, it was unclear to me which parts of the electrical system were most directly responsible for generating unusual electrical phenomena, whether it be the generator or high voltage source, the types and arrangements of the various coils, or a combination of these elements setup and arranged in a specific manner, tuned in a specific way, and operated in a specific method.

The complete spark gap generator (SGG), including the diathermy replica (DR), and the MMC capacitor bank unit, is shown in Figures 1 below, and mounted on top of, and connected to the high voltage supply:

Far more consideration is normally given to the experimental components (e.g. coils), their construction, dimensions, materials, and the results that they yield, and much less on the generators that produce the high voltages and currents that are used to power the experimental apparatus.

Over time this appears to have led to an “air of mystery” surrounding the generators that are used in these types of experiments, quite besides a good generator is a complex and involved process to design and build, and can take much more time than any other system component to “get right”. Certainly when I started out by watching the experimental work of Dollard et al.[1,2], I was very much left with the impression that many of the unusual results obtained were mostly a product of the special generator and components used, and the experimental coils allowed these effects to be transformed, observed, and experimented with.

I have not been alone in these impressions, as I have received very similar comments from others in the field that have not actually built a working experimental system for themselves, but rather still feel that “air of mystery” that surrounds the generator and specialised components and materials in the systems construction. Only by building or contributing to a working experimental system, (including the generator), is it possible to really dispel this “air of mystery”, as it becomes possible to understand and characterise how the generator is producing the types of voltages and currents specific to its type.

In addition to this, many of the components referred to in important works such as Dollard et al.[1,2,3], may well have been more readily available in the 80s and early 90s, but are now quite scarce, and often command high prices for working items, or “new old stock” components. For example a 1920s H.G.Fischer diathermy machine used as the primary generator in the experiments of [1] are very rare, and when they very occasionally are available, they are expensive. Without understanding what is inside a generator such as this, and how it is working, it is very difficult to know how to build a comparable generator, or whether one will be able to gain the same types of unusual electrical phenomena demonstrated in works such as [1,3].

This was certainly the place I found myself in the early days where I wanted to begin by reproducing and confirming for myself the unusual measurements and results obtained by others, before using this as an established foundation to advance further in exploring my own ideas and insights regarding electricity, and the displacement and transference of electric power. When I started out there were no good examples of a working diathermy machine currently available, (especially in the UK), so I decided to design and build a diathermy replica for myself, and using readily available materials and components. If this generator could be used to explore unusual electrical phenomena then it would certainly for me increase my understanding enormously of how such generators are designed and constructed, whilst also dispelling the “air of mystery”, and providing a generator design that could also easily be used by others in the field. Later I did finally acquire an original 1920s H.G.Fischer diathermy machine  and have been able to make a characterisation and comparison between the replica and the original.

The posts reporting the Spark Gap Generator – Parts 1 and 2 are the result of building a working high power diathermy replica, which is now routinely used in my daily experiments, and has contributed significantly in bringing me to a core understanding, that all parts of the electrical system play a specific role in the generation of unusual electrical phenomena. In the exploration into the displacement and transference of electric power each part of the system apparatus must first be measured and characterised carefully to establish a well tuned and balanced overall system, and where the electric and magnetic fields of induction are balanced and in a state of dynamic equilibrium.

From this point it is possible to experiment directly into the properties of electricity, and come to an understanding that the unusual phenomena observed are a product of the inner workings of electricity itself, where the generator and experimental apparatus are necessary to set up the conditions and boundaries required to explore these inner properties.  The properties of electricity are there to be revealed rather than “generated”, which also assists greatly in dispelling the “air of mystery” that the generator is the source of the unusual phenomena, but rather, the instrument that provides the necessary tension to trigger the imbalance of the electric and magnetic fields of induction, and hence to observe electrical phenomena within the system. The differences here are subtle but hugely important to the overall understanding of electricity and particularly displacement of electric power.

The circuit diagram for the SGG and peripherals is shown in Figure 2 below, or click here to view the high-resolution version.

In the early days in order to replicate the experiments and results of Dollard et al.[1] , it was considered that the diathermy replica (DR) should be as close as possible to the original, which of course posed a challenge when there was no original from which to take measurements, dimensions, construction methods etc. To overcome this a range of references from the internet were studied, along with available circuit diagrams. The most useful references proved to be a combination of material from [3,4,5], and allowed key dimensions and some circuit component values to be extracted from the images and videos.

Click on the following links to view circuit diagrams for various original H.G. Fischer (HGF) diathermy machines, the Model G2, the Model H, the Model CDC, and the Model A[4].

In any Tesla or resonant coil design it is first necessary to define the desired properties of the secondary coil. With this defined the primary and other components of the system can be designed around the secondary properties. In the case of the DR, and in the absence of any good measurement data, e.g. the resonant frequency of the HGF secondary, the DR secondary was designed according to the known and available dimensions of the HGF secondary, the wire size and type, and the number of turns, (mostly gained from [5]). The dimensions and wire were then adjusted for readily available materials and then key parameters extracted using the software Tccad 2.0[6]. The DR secondary coil properties were adjusted primarily to match F0 the primary resonant frequency of the secondary coil, while keeping the physical dimensions as close as possible,  whilst using readily available material types and sizes. The original HGF and the designed and adjusted DR properties are shown in the following table:

Original H.G. Fischer Model G Diathermy:

From reference pictures and video[5]:

Turns: 90

Wire: Solid copper 20 AWG cotton-clad, total diameter 1.0mm

Secondary Former (OD): 25/8“, 65.7mm


Secondary Coil length: 92mm

Primary Former (OD): 31/2“, 89.6mm

Primary Coil: 6 turns 3/16” copper tube

Oudin extended coil: 11 turns 12 AWG solid (2mm) magnet wire

From Tccad 2.0:

L = 340uH

C = 2.87pF

F0 = 5066 kc/s

Fλ/4 = 3978 kc/s

Additional top load Cλ/4 = 1.78pF

Wire length (WL) = 18.9m

Diathermy Replica:

Adjusted for available materials to match F0:

Turns: 98

Wire: 19/32 20 AWG Kynar (PTFE coated), total diameter 1.2mm

Secondary Former (OD): 63mm (standard UK polypropylene pipe)

Secondary Former OD + PTFETAPE + PTFEWIRE = 63.40mm

Secondary Coil length: 120mm (the PTFE wire  coating adds additional length)

Primary Former (OD): 90mm (standard UK polypropylene pipe)

Primary Coil: 6 turns 3/16″ annealed copper tube

Oudin extended coil: 11 turns 14 AWG solid (1.6mm) magnet wire

From Tccad 2.0:

L = 350uH

C = 2.84pF

F0 = 5065 kc/s

Fλ/4 = 3806 kc/s

Additional top load Cλ/4 = 2.03pF

Wire length (WL) = 19.4m

Figures 3 below show how the Tesla/Oudin unit was constructed, and the types of materials used in a simple open structure that can be easily adjusted and modified according to the experimental requirements. Components are mounted on a mdf wooden base, and conductors insulated from the mdf using PTFE and Nylon 66 mounts, bolts, and nuts.

The formers of the primary and secondary coils were first coated with PTFE tape to improve the thermal barrier between the coil and the polypropylene former material when running at high output powers. In addition a small low voltage fan was located under the primary coil to help keep the primary cool at sustained high power outputs. The Tesla EHT output and the HT output taps are all mounted on PTFE insulators both for electrical isolation, but also for good resistance to melting and burning which can occur when drawing discharges from these terminals. Nylon 66 can also be used here, but has lower thermal resistance to discharges, but has the benefit of being a much cheaper material than PTFE.

It is important to note that the secondary coil is located in the primary at the opposite end to the Oudin extended coil. In an early version of the DR the secondary was incorrectly located at the same end as the Oudin extension, which because of the increased tension in both coils easily causes breakdown between the two coils, and led to burn-out of the first secondary coil. The tank capacitor banks are made from 3 series connected Cornell Dubiler 941C03 series 3kV polypropylene film capacitors each of value 47nF, which combined gives 15.7n 9kV for each bank. These tank capacitors have been proven to be long-lasting, and robust, and have never been changed,  even with input powers in excess of 1.5kW and even up to 2.2kW for short bursts of power.

The tank capacitors are force air-cooled, and mounted on insulated conductors which allow for easy connection and adaption to the circuit under test. The capacitance of the tank was initially higher at around 47n to match more closely the HGF schematic of the Model A, but was reduced to its current value after the real HGF was acquired and measured. At 47n the available output power was quite a bit less than with 15.7n per tank, as the primary resonance is pushed lower and further away from that of the secondary, hence reducing the primary currents, and hence the power coupled from the primary to the secondary.

Figures 4 below show how the static spark gap (SSG) unit was constructed.

The SSG unit was at the time the most difficult unit to build when only a large pillar drill, large vice, and bench sander were available in the workshop for mechanical construction. Each electrode of the spark gap is made from 1/4” diameter 99.9% pure tungsten rod 1″ long, (not to be confused with tungsten carbide rods), which were pressed using the vice, into the centre of a drilled A2 stainless steel fine pitch bolt. The bolt had a 6.2mm hole centre drilled, by mounting the bolt in the drill chuck and clamping the drill stationary on the stage, and opposite to how a drill press is normally used. This arrangement made a very rudimentary “lathe” and made it easier to drill a centralised 1.5″ hole down the centre of the bolt. The tungsten electrode was then pressed into the bolt leaving 5mm externally for the spark electrode. Alternating stainless steel washers (large and small) were then threaded onto the bolt and finally tightened with a thin stainless steel nut to form the cooling fins of the electrode body.

Each pair of electrodes were then mounted in threaded aluminium blocks, locked in place with a nut on one side, and with a threaded bakelite handle on the other, to allow adjustment of the spark gap space by winding the bolt in or out of the aluminium block. The aluminium blocks were arranged and mounted to form a series connection of all four spark gaps, and also allowing for tapping from any of the 4 stages for experiments using a single gap, all the way up to 4 series gaps, or 4 parallel gaps with shorting shunts. The adjustable electrode was tensioned in the alumium thread by a small compressed nylon rod, ( from an M3 nylon bolt), which was inserted in a vertical hole drilled above the thread, and then tensioned using a screw locked in the correct place by a nut.

The aluminium spark gap blocks are mounted onto PTFE insulators and then mounted to a wooden base where each gap is suspended above force cooling provided by a pair of low voltage plastic fans mounted into the base of the unit. Overall the SSG unit is robustly constructed and can withstand very large powers in constant use. Tuning of the gaps for optimal running can be made carefully during operation via the four insulated adjustment handles, and is demonstrated in the operation video in Part 2.

Figures 5 below show the MMC tank capacitor bank (TCB) which was used to remove one of the tuned primary stages, and hence increase the efficiency and optimisation of the generator driving, for example, the tuned primary of a TMT (Tesla magnifying transformer).

The SGG used with the DR as a generator is most commonly connected with one of the L, M, or H outputs to the primary stage of a tuned TMT, of which the primary stage would typically consist of a coil with a parallel tuning capacitor. In this arrangement the DR, which in itself is already a tuned primary stage, is now connected to another tuned primary stage. Whilst this was useful for preliminary experiments in confirming the key experiments and results of Dollard et al.[1,2] and keeping the generator as close as possible to an original HGF, it has been shown to be more efficient to eliminate the double tuned primary stage by removing the DR and placing the TCB in its place.

The TCB is simply a standalone capacitor unit exactly the same design as used in the input to the DR. Here the TCB is shown with two pcb MMC banks, but the Cornell Dubilier 941C03 series banks could equally as well be mounted in the same way. Figures 1.6 and 1.7 show how the TCB is connected directly to the SSG, and then in turn the output of the TCB is connected directly to the input of, for example, the tuned primary of a TMT, or some other experiment or load. This creates a resonant drive circuit with the TMT coil in series resonance with the TCB capacitors, and is exactly the same arrangement as internally for the DR, and the HGF. Use of the TCB in replacement for the DR has allowed for an easier and more accurate resonant match between the SSG unit and the TMT load, and with greater power transfer between the two units. Impedance measurements are also simplified by removing one resonant circuit from the generator chain. A very wide range of capacitors can be mounted on the TCB, and for higher power experiments up to 2.5kW output power, force cooling is available via the low voltage fan in the base of the unit.

Overall the spark gap generator whether used with the DR unit or the TCB unit at the output, has proven to be a robust and reliable generator for a wide range of experiments. It has enabled the replication of the key experiments as presented by Dollard et al.[1,2,3,4], as well as forming a flexible and powerful tuned static spark generator for my own experiments in the displacement and transference of electric power, as well as telluric transmission experiments.

Click here to continue to part 2 of the spark gap generator where the operating characteristics are measured both in frequency and time, as well as a short video to show its general operation and running.

1. Dollard, E. & Lindemann, P. & Brown, T., Tesla’s Longitudinal Electricity, Borderland Sciences Video, 1988.

2. Dollard, E. & Brown, T., Transverse and Longitudinal Electric Waves, Borderland Sciences Video, 1988.

3. Dollard, E. & Mackay, M., Tesla Radiant Energy Experiment, Bedini-Lindemann Conference, June 29-30, 2013.

4. Mackay, M. & Dollard, E., Tesla’s Radiant Matter Replication, 2013, Gestalt Reality

5. Sergey Z., Fischer Diathermy Narrating and Exploring a 1920’s Tesla Coil, May 2014, Youtube

6. Chapman R., Tccad 2.0 for Windows, 2000.


1920s H.G. Fischer Diathermy

Later in the research, (and after the replica diathermy unit had been designed and built), I was lucky enough to come across a real 1920s H.G. Fischer diathermy unit (HGF), which although being sold untested, and in unknown condition, looked suprisingly good from the pictures. It survived the shipping from the USA to the UK all in one piece, and on closer inspection proved to be in good physical and working condition, including the thermo-ammeter, and the original fuse. The only part missing was the 6V power indicator bulb.

The unit is an early model GP, which is actually a model G in a compact portable cabinet, with a robust carry handle, and metal cabinet corner protectors. Being from the USA the ac line voltage input is specified at 110V 60Hz with a maximum power input of 500W. In the UK this line voltage requires a step-down transformer, a suitable auto-adjusting 1kW transformer was found to be adequate for long-term operation. This post presents detailed pictures, measurements, and an analysis of the HGF as a spark gap generator, and a comparison to the replica diathermy already built and reported in the Spark Gap Generator Part 1 and Part 2.

Figures 1 below show the front-panel of the unit, and in detail, the construction and configuration of the various inner components of the HGF:

The actual circuit diagram for the HGF, which was obtained by opening and measuring the unit, is shown in figure 2 below, or click here to view the high-resolution version. Click on the following link to view the original circuit diagram which appears the closest match to this unit model, and comes from the original Model G2.

The line input is connected through a switch, (also with outputs for a foot switch), to the input of a multi-tap choke. The choke is designed to restrict the current, based on the output tap setting, through the primary of the high voltage transformer, and hence control the output voltage of the transformer and the strength of the spark discharge. The model G has 5 power settings which control the input power measured in the range 75 – 440W @ 120V, (line output of the auto-adjust step-down transformer). The choke is constructed from multiple layers of windings using AWG 18 solid magnet wire, wound over thin cardboard layer separators, and coated in a Shellac type resin to hold all the windings together. The core of the choke is made from many thin mild steel laminations and mounted each end to the front panel by wooden spacer blocks.

The output of the tap selector from the choke is connected to the primary of the high voltage transformer. The primary is mounted on thicker mild steel laminated core that is rectangular in shape, with the primary at one end, and the two secondary coils at the other. The primary is wound again like the choke with multiple layers of, cardboard separated, AWG 16 solid magnet wire, held in place with the same resin, and then externally coated with paper tape. The two secondary coils are again made from many cardboard separated layers, with ~25 windings per layer of AWG 22 solid magnet wire, and structurally held together by thin wooden layers compressed by insulating fibre threaded rod, and plastic threaded round nuts.

The combination of the primary and secondary coils on the laminated steel core forms an early transformer arrangement. The efficiency of this type of basic transformer is not as high as would be expected for a typical modern design, constructed with modern materials. Increased losses in the laminated core and materials lead to more power dissipation and heating during running, although the size and bulk of the construction results in a robust transformer, that can be run all day long at maximum power without overheating. This is further evidenced that the transformer is still functioning correctly, and is not far short of 100 years old!

The two high voltage transformers are connected in series, and with the outer ends of the two coils connected to the outer ends of the 3 series connected spark gaps, via coiled cotton-clad stranded wire. The spark gaps appear to be a tungsten tipped centre electrode, clad with a machined copper heat sink fin. The other end of the centre electrode is very fine threaded, and with a bakelite adjustment handle at one end to fine adjust the gap spacing of each electrode pair. The outer ends of the 3 series connected spark gaps are connected both to the output of the secondary coils, and the inputs to the series capacitors, in the first stage of a typical Tesla[1] “hairpin” arrangement. Further detail regarding the Tesla “hairpin” circuit and its history is given in a summary presented by Kraakman[2].

The two series tank capacitors, ~15nF each, are constructed from alternating thin copper conducting sheets with thin insulating mica sheets. These are sandwiched together between two cotton webbed sheets, insulating them from the outer wooden blocks that compress the capacitor sheets together. The wooden end blocks are compressed together by insulating fibre threaded rod, and plastic threaded round nuts. In fig. 1.6 it is important to note that on the left hand tank capacitor there is another smaller capacitor formed on the outside of the wooden block. This is the floating ground connection capacitor important for diathermy use when connected to a patient. This capacitor prevents the final outputs of the diathermy unit to accumulate to very high potentials relative to ground, which would present a considerable discharge danger to the patient from their body to ground. It is somewhat disconcerting to imagine that the safety of the patient, when connected to this type of high tension generator, was only really ensured by the two series tank capacitors and the floating ground connection capacitor. Failure of any one of these three capacitors would effectively connect the patient to around 6kV @ 100mA at the input line frequency of 50/60Hz!

The two tank capacitors are connected to each end of the primary coil, which is 6 turns of 3/16” copper tube, and then extended by an 11 turn  AWG 12 solid copper wire Oudin coil extension. The low, medium, and high output taps are then derived from the primary and Oudin extension as shown in the schematic diagram of figure 2. The primary and Oudin extension are wound around a 31/2” primary former which appears to be a resin/mica composite material. The primary turns are separated from each other by a cotton woven thread which matches the diameter of the copper tube and wire. Connection between the spark gaps and the tank capacitors, the primary coil, and the output taps, are via 3/16” x 1/16” solid copper flat bar.

The secondary coil is a 90 turn 20 AWG cotton coated magnet wire on a 25/8” cardboard former, and the turns retained in place by the same Shellac type resin. The secondary coil is retained in the centre of the primary former by two wooden end caps, themselves compressed together by another insulating fibre threaded rod, with plastic threaded round nuts. The inner end of the secondary is connected to the common connection as shown in the schematic, and the other outer “hot” end of the secondary is fed through the centre of a bakelite stand-off/insulator through the front-panel and to the high tension Tesla terminal, (ball connector with 4mm socket hole). It is important to note that the secondary coil is orientated within the primary former at the opposite end to the Oudin coil extension, which prevents the very high tensions at the top-end of both the primary and secondary from breaking through the primary former forming a spark discharge path between primary and secondary coils.

The primary output taps are connected through a manually positioned thermo-ammeter shunt that shows rf output current on two ranges 0-1A, and 0-4A. The manual shunt can be positioned to remove the meter from the circuit, or to connect it in high and low range positions. The low range is protected by a 1A slow-blow fuse that shunts a 22Ω 50W wire-wound power resistor. When the fuse blows the shunt across the resistor is removed and the meter is protected by the series power resistor. The overall low-side output of the manual meter/shunt circuit, is connected to the indifferent (ground) terminal on the front-side of the diathermy main panel. The measured component parameters combined with their important physical attributes are shown on the schematic in figure 2.

Figures 3 below show the small signal impedance measurements for Z11 up to 10Mc/s at the output of the spark gaps for both the original HGF and the diathermy replica (DR) unit:

To view the large images in a new window whilst reading the explanations click on the figure numbers below, and for a more detailed explanation of the mathematical symbols used in the analysis of the results click here.

Fig 3.1. Shows the fundamental resonant frequency Fat M1 of the primary circuit, and the fundamental for the Tesla secondary coil FS at M2, and its second harmonic FS2 at M4. FP at 1120kc/s is the series resonant frequency formed by the combination of the two series tank capacitors CP connected together by the spark discharge, in series with the primary coil inductance LP. LPCP forms a series resonant circuit where the reactance of Land CP cancel each other out at resonance, leaving only the series resistance of the primary RP at M1, which in this case is 0.49Ω.

M2 shows the fundamental resonant frequency of the secondary FS = 2900kc/s, and M3 the frequency at which a 180° phase change takes place FØ180 = 3180kc/s. As is normal for a secondary coil where there is considerable distributed resistance across the coil end, points FS and FØ180 do not occur at the same frequency, and the parallel resonance formed between LS and the distributed capacitance CS set the fundamental resonance of the coil at M2. When electrical energy is coupled to the secondary from the primary the coil will resonate at the frequency indicated at M2. It is also to be noted that the Q of the resonance at M2 is considerably lower than expected, showing that losses in the secondary coil, materials, and mountings are considerable, where rf energy is being both dissipated in RS, and leaking out of the circuit formed by LSCS through parasitics to the surrounding medium. The low Q of the secondary considerably impacts the energy stored in the system, and will reduce considerably the rf oscillating currents in the secondary, which can be seen in figures 3.

The second harmonic of the secondary FS2 occurs at M4 and M5. If the primary is tuned closer to M4 then the secondary coil will resonate at FS2 = 8140kc/s which represents the second odd harmonic of the secondary wire length, 3λ/4. The parallel resonance at M4 is noted also to be very low Q, and similar in this case to the fundamental. The low Q of the secondary can most likely be attributed to, firstly, the cardboard former of the secondary coil, which over considerable time will have absorbed moisture, and presents a considerable leakage or parasitic resistance to the windings of the secondary coil. Secondly, the windings in themselves are only cotton-clad un-insulated (bare) magnet wire, which also presents a significant leakage path to a moisture impregnated cardboard former. Thirdly, the cardboard former is mounted to the primary via wooden end boards which themselves can absorb moisture, and when combined with moisture in cardboard former of the secondary could also form a significant leakage between the windings of the primary (bare copper) and the secondary coils.  It is conjectured that the Q of the secondary when the HGF was new, or much younger, would have been better than now measured, but still due to the nature of the materials used, would still present a much lower Q than that which can be obtained by using plastic formers, and with magnet wire either PTFE coated, or high temperature varnish coated.

Fig 3.2. For comparison the same small signal impedance measurement is shown from the spark gap generator diathermy replica (DR). Due to the slightly different geometric sizes, of the readily available materials used, for the primary and secondary, and the modern equivalent of the tank capacitors, the key resonant frequencies of the primary and secondary are at slightly different points. However the general characteristics of the frequency markers remains very similar on the horizontal scale.  The series resistance of the primary circuit at resonance for the DR is less than half that for the HGF, showing that larger primary currents can be generated in the DR providing stronger output currents in the low, medium, and high primary taps.

The big difference between the HGF and DR is in the Q of the secondary coil, which is very much larger, and well-defined, in the case of the DR. This shows the difference largely between the types of materials used to construct the secondary, which in the case of the DR is a plastic polypropylene former, with PTFE coated Kynar secondary windings, insulated from the plastic polypropylene former of the primary, by nylon 66 connecting bolts. All these plastic insulating mediums do not suffer with moisture absorption over time, do not degrade significantly over normal time spans, and present a very high impedance between the primary and secondary coils, which reduces any leakage currents in the secondary to very low values. Hence the Q of the secondary circuit is very sharp and well-defined.

Figures 4 below show the large signal time domain waveforms of the HGF as measured from the indicated output taps, and illustrate the different stages of the spark discharge burst both in the primary and secondary coils of the generator. The HGF was being run at an input power of ~ 300W, (monitored using a Yokogawa WT200), which was kept constant throughout the measurement. Output waveforms were measured using a Pintek DP-50 high voltage differential probe, (max. 6.5kV up to 50Mc/s), which was connected to a HP 54542C oscilloscope to observe and record the output waveforms.

To view the large images in a new window whilst reading the explanations click on the figure numbers below. For a detailed description of the different sections of the spark discharge refer to the analysis of figures 3 and 4 in Part 2 of the Spark Gap Generator post.

Fig 4.1. Shows the burst waveform measured at the low output tap of the HGF. The vertical amplitude scale is 1kV/div, and the horizontal timebase is 5µs/div. This figure shows clearly the  second section, a ring-down of specific frequency based on an exponentially decaying oscillation in the primary coil of the HGF. The third section, a ring-down of another specific frequency on an exponentially decaying oscillation in the secondary coil of the HGF is too small to be observed on this vertical scale. The maximum amplitude of the burst is ~ 5kVpk-pk, and lasts for about 7.5µs before decaying to less than 1% of its initial amplitude.

Fig 4.2. Here the horizontal timebase has been reduced to 500ns/div and the waveform buffer delay adjusted so that section two dominated with the primary oscillations fills the entire trace. The monitored average frequency of trace 1 can be seen to be 1077kc/s which is FP, the fundamental resonant frequency of the primary circuit, which corresponds well to that measured in the small signal impedance measurements at M1, of 1120kc/s.

Fig 4.3. Here the vertical amplitude and horizontal timebase have been reduced from fig. 4.1, in order to show the very small third section of discharge burst, that occurs from the secondary ring-down reflected into the primary circuit. The secondary ring-down is barely 100Vpk-pk and reflects the very low Q of the secondary fundamental resonant frequency FS, as shown in fig. 3.1. Due to the low Q from the high leakage through the materials used, the third section ring-down has a low amplitude and decays away very quickly, only lasting in this case ~ 4µs after the primary ring-down in section 2.

Fig 4.4. Here the third section has been magnified and the monitored average frequency of trace 1 can be seen to be 2929kc/s, which again corresponds well to that measured in the small signal impedance measurements at M2, of 2900kc/s.

Fig 4.5. The low tap burst, again on the original scales, for amplitude comparison with the following two figures.

Fig 4.6. Shows the burst waveform measured at the medium output tap of the HGF, and on the same scale as before. The inital burst amplitude has increased to ~ 8kVpk-pk.

Fig 4.7. Shows the burst waveform measured at the high Oudin output tap of the HGF, and on the same scale as before. The inital burst amplitude has increased to almost 12kVpk-pk.

A comparison of some of the key characteristics of the HGF and the replica diathermy already built and reported in the Spark Gap Generator Part 1 and Part 2, are shown in the following table:

H.G. Fischer Model GP Diathermy:

Primary Coil: 6 turns 3/16″ copper tube

Primary Former (OD): 31/2“, 89.6mm

Oudin extended coil: 11 turns 12 AWG solid (2mm) magnet wire

Seconday Turns: 90 turns, 20 AWG cotton-clad magnet wire

Secondary Former (OD): 25/8“, 65.7mm

From figures 3. Z11:

FP @ M1: 1120kc/s

RS @ M1: 0.49Ω

FS @ M2: 2900kc/s

FS2 @ M4: 8140kc/s

From figures 4 (large signal time domain, LSTD @ 300W):

FP: 1077kc/s

FS: 2929kc/s

VL (pk-pk): 5kV

VM (pk-pk): 8kV

VH (pk-pk): 12kV

Diathermy Replica (SGG Parts 1 and 2):

Primary Coil: 6 turns 3/16″ annealed copper tube

Primary Former (OD): 90mm (standard UK polypropylene pipe)

Oudin extended coil: 11 turns 14 AWG solid (1.6mm) magnet wire

Secondary Coil: 98 turns, 19/32 20 AWG Kynar wire (PTFE coated)

Secondary Former (OD): 63mm (standard UK polypropylene pipe)

From figures 2. Z11 (Spark Gap Generator – Part 2):

FP @ M1: 950kc/s

RS @ M1: 0.22Ω

FS @ M2: 3180kc/s

FS2 @ M4: 8390kc/s

From figures 3 LSTD @ 300W (Spark Gap Generator – Part 2):

FP: 895kc/s

FS: 3214kc/s

VL (pk-pk): 4kV

VM (pk-pk): 8kV

VH (pk-pk): 11kV

In summary, the original H.G. Fischer medical diathermy unit explored in this post is a robust, self-contained, and easy to use generator suitable for some preliminary experiments and replications in the field of Tesla and electricity research, where high tension oscillating currents are required, e.g. with a TMT experiment. The HGF’s lower overall performance, in comparison to the replica diathermy unit, largely results from age related wear and tear, component degrading, and generally lower quality, or less suitable, materials for high voltage applications. Having said this, the HGF is almost 100 years old, and still working more than adequately as a high voltage generator, which is in itself an impressive accomplishment from another time. Without access to an original HGF, it has been shown in other posts, that a good high performance generator with very similar yet improved characteristics, can be constructed using readily available materials at a very affordable cost.

Click here to continue to part 1 of the spark gap generator where the diathermy replica is designed and constructed.

1. Tesla, N., Experiments with alternate currents of very high frequency and their application to methods of artificial illumination,  American Institute of Electrical Engineers, Columbia College, N.Y., May 20, 1891.

2. Kraakman, N., A Brief History of the Tesla Hairpin Circuit, December 7, 2017, Waveguide


Multiwave Oscillator Impedance – Part 2

In part 2 of the multiwave oscillator impedance (MWO) measurements we take a look at the high frequency impedance characteristics Z11 for the MWO resonator rings (top-load). The MWO Tesla style drive coil measured in part 1 needs to be separated from the MWO top-load for this type of measurement, because the impedance and load of the drive coil masks out any resonant features at higher frequencies, making it impossible to measure the MWO when connected to the top of the secondary coil and measured at the input to the primary coil.

Experiments, operation, and measurements in Tesla coil based systems, and indeed in my own research regarding the displacement and transference of electric power, are normally limited to experimental apparatus which operates at maximum in the HF band < 30Mc/s of the electromagnetic spectrum. At 30Mc/s the corresponding wavelength λ is 10m. As a standard rule of thumb in high frequency or RF measurements the feature size of the element under measurement should be < λ/10 to be considered as a discrete lumped element in the system, and therefore not subject to impedance transformations through transmission line characteristics or other distributed network characteristics. At the upper end of the HF band this means that active measurement regions < 1m in size or length can be considered to have only small, if any, transformation and change on the impedance outside of that attributed with its discrete lumped element network.

As we progress to higher frequency measurements in the VHF band (30 – 300Mc/s) and even above in the UHF band (300Mc/s – 3Gc/s) very great care needs to be taken regarding the length and size of cabling and connections, the feature sizes of components within the measurement network, the impedance match or mismatch between transmission line sections and discrete components, and any other stray, parasitic, or proximity effects arising from boundary conditions that are both metallic or dielectric in nature. Accordingly, much greater care and attention needs to be taken to calibrate as accurately as possible directly to the plane of measurement, and then to normalise out measurement effects between the calibration plane and measurement apparatus.

In this part we are concerned with measurement up to ~1Gc/s (λ = 30cm), and where the λ/10 length is only 3cm. At λ/4 quarter-wavelength we would expect to see a complete impedance transformation along a transmission line from a very low impedance (short-circuit) at the output of the source generator to a very high impedance (open-circuit) at the end of a wire or measurement plane. Accordingly we need to keep connections small and efficient which represents in itself a unique challenge when connecting to the MWO rings, which at UHF frequencies represent a large distributed network of inter-connected transmission line elements, with complex boundary conditions. Subsequent parts will look at the MWO rings as an antenna with corresponding radiation resistance and defined radiation pattern and particularly in relation to the transmitter – receiver configuration of the complete MWO system.

For this part, and to measure as accurately as possible Z11 for the MWO rings directly, and up to frequencies in the order of ~1Gc/s, it was necessary to make a connecting jig that allowed for extension of the reference plane from the VNA-SDR directly to the mounting spheres of the outer driven ring of the MWO. Figures 1 below show the measurement setup, signal feed adapter, MWO mounting jig, calibration elements, and overall arrangement of the high frequency impedance measurement apparatus.

To view the large images in a new window whilst reading the explanations click on the figure numbers below.

Fig 1.1. Shows the overall measurement setup with the VNA-SDR connected to the MWO mounting jig by a short SMA terminated RG316 cable, and the other end to the signal feed adapter (SFA), and standard calibration elements.

Fig 1.2. The SFA mounts directly to high voltage terminals, that are themselves connected to the brass mounting bolts used to hold the MWO rings to the jig. The jig is constructed predominantly in nylon 66 with a wooden internal support for strength inside the main jig supprt column. The high frequency path length and connection effects are normalised out by extending the calibration reference plane right to the brass bolt mounting points for the MWO outer ring. The SFA in this case is used with shorted series capacitance feed, and is balanced with equal weights of copper on the positive and negative feed paths, which includes the SMA connectors and the RG316 cable to the VNA-SDR. Equal weights of copper, or more accurately at high frequencies (> 3Mc/s), equal volumes of copper assist in balancing the boundary conditions for the electric and magnetic fields of induction and so reducing impedance mis-matches in the signal feed path. As for part 1 an SFA with 3Gc/s balun was also tried but found to be largely unsuccessful in the small signal measurements due to the masking of high frequency detail, which results from the balun coupling transformer dominating the frequency characteristics in the mid to high-band frequencies.

Fig 1.3. The calibration elements fit directly onto the end of the MWO brass mounting bolts. With no element attached the mounts are calibrated for open-circuit, a single-sided pcb element is used for the short-circuit, and a single-sided pcb with 50Ω surface mount resistor in the centre for the calibrated load.

Fig 1.4. Shows the MWO rings mounted to the measurement jig.

Fig 1.5. An improvement to the calibration procedure removed the SMA connecting cable by mounting the VNA-SDR directly to the SFA by SMA coupler. In this configuration some additional resonant features were removed from the normalised calibration results.

Fig 1.6. The final measurement setup for the high frequency MWO impedance measurements with VNA-SDR directly mounted to the SFA, which is in turn directly mounted to the jig, and where the calibration plane is at the point where the MWO outer ring brass balls meet the brass mounting bolts.

Figures 2 below show the Z11 impedance characteristics of the calibrated reference plane and the lower high frequency measurement band 100kc/s – 500Mc/s. The entire high frequency band was split into two in order to best comply with the calibration bands within the VNA-SDR instrument. At 500Mc/s the instrument internally changes reference method and level, and sensitive measurements cross-band are not recommended. Therefore the measurements are calibrated into a lower band 100kc/s-500Mc/s and an upper band 500Mc/s – 1.3Gc/s.

Fig 2.1. Calibration for the 50Ω reference element over the band. Markers M1, M3, and M5 show the consistent calibration across most parts of the band. At M2 at 138Mc/s and M4 at 377Mc/s there are resonances within the calibrated signal path which are too significant to normalise out. These points will need to be factored into the measurement results to remove features of interest, if any, that correspond at these frequencies. At these 2 points features in the results are considered measurement artifacts that arise due to the measurement system and inter-connection between the instrument and the MWO rings at the calibration plane.

Fig 2.2. Shows the rich range of resonant features that result from the MWO rings and their inter-action across the band. In this first result markers M1 – M12 are used to identify the lower 12 set of features, and in the next set of results the upper 12 set of features. There is no corresponding feature at 138Mc/s as seen in the calibration scan, and hence this point does not need to to be removed from the results. The first 250Mc/s of the results show a number of strong resonant points indicated by the large phase changes occurring at M4, M6, M8, and M9, and most likely attributed to resonances in the outer few rings of the MWO. There are also smaller phase inversions across the first half of the results occurring at M1, M2, M5, M7, M10, and M11 which most likely arise due to resonant inter-actions between the outer few rings of the MWO. At M12 287Mc/s  a large peak change in |Z11| occurs for a very small phase inversion, which would likely indicate where the outer drive ring of the MWO becomes self-resonant, in a similar way to that of the primary in the drive coil which becomes self-resonant at ~38MC/s. The point at M12 may also correspond to the 2nd harmonic of this point with the fundamental occurring at M8, this however being less likely due to the very low impedance swing at M8 to ~6.9Ω indicating that M8 is more likely to be associated with the fundamental resonance of one of the few outer rings, but not the driven ring.

Fig 2.3. Shows the markers on the upper 12 set of features. At M10 375Mc/s there is a large resonance, the only in the upper 250Mc/s of the results, which also very closely corresponds to the calibration feature noted in Fig 2.1. at M4 at 377Mc/s. The close correspondence of these two points means that the feature at M10 needs to be attributed to a strong resonance in the calibration path which could not effectively be normalised out by the calibration procedure. Otherwise in the upper part of the reults there are some small phase inversions at M11 and M12 which could be tentatively linked to the fundamental resonant frequencies of some the rings further into the MWO centre. In the upper part of the results we start to see a general masking, smearing, and suppression of further detail, not necessarily because the detail is not there, but rather the coupling between rings is not sufficiently loose to allow for the natural resonance of each ring to be reflected in the results, and combined with rapidly increasing losses, large reductions in Q, and decreasing amplitude of oscillation as the results move towards the centre of the MWO. This could be further tested and measured by decreasing the coupling factor between each of the rings by changing the structure of the MWO from a flat set of rings to a cone configuration, where each of the rings retain the same diameter but are separated further apart vertically on the cone yielding a three-dimensional MWO structure. The other way to measure and assess this further is to remove every other ring in the flat MWO so reducing the coupling between rings and also reducing the large number of inter-active resonances. The process of starting with one ring inside the driven ring and subsequently adding ring by ring taking an impedance scan with each ring, would also assist in clarifying which fundamentals occur from which rings, and which features occur due to cross-coupling between the rings.

Figures 3 below show the Z11 impedance characteristics of the calibrated reference plane and the upper high frequency measurement band 500Mc/s – 1.3Gc/s.

Fig 3.1. The calibration of the upper high frequency band shows very good consistency and results for the 50Ω reference load up to M3 at 1096Mc/s. Above this point the signal attenuation is such that the returned measurement currents to the VNA-SDR are so small as to be falling outside the lower dynamic range of the instrument, and hence the results become progressively more inaccurate and noisy. It is to be noted that the base 50Ω reference load can still be extrapolated visually to the end point ar 1300Mc/s. However, above ~1Gc/s the results on this measurement apparatus and setup are considered to not be of further use as can be more clearly seen in the next result.

Fig 3.2. At the lower end of this band markers M1 – M5 continue the upper band of the previous results in most likely showing fundamental resonances and maybe even some inter-actions between some of the more inner rings of the MWO e.g. rings 4-7. It is unlikely that any details can easily be discerned for the inner rings 8-11 where the signal reflections are very small, and stronger coupling between rings has largely smeared out the detail and suppressed the Q of the response.

Overall the high frequency impedance measurements have presented a wealth of detail particularly in the lower part of the band and with discernible results being contributed to by as many as the first 6-8 rings inside the outer driven ring. It is also clear that cross-coupling between rings will contribute to very high frequency components to be emitted from the MWO, even though such detail is largely outside of the scope and dynamic range of the measurement technique used here. A useful comparison measurement would be to look at the range of frequency components generated during normal high tension operation of the complete MWO system, using a spectrum analyser instrument with capabilities into the Gc/s range. The high frequency impedance characteristics still need to be measured for the complete MWO system where both transmitter and receiver are connected to the system, which will be reported in a subsequent part.


Multiwave Oscillator Impedance – Part 1

The original Lahkovsky Multiwave Oscillator (MWO) apparatus combines two Tesla style drive coils in a transmitter and receiver configuration, each consisting of a primary and secondary coil cylindrically mounted on axis. The top-load for both transmitter and receiver is a complex combination of concentric half-wave resonators. The impedance characteristics of even a single drive coil with top-load represents a complex measurement challenge with results that can span over a very wide frequency range, in the order of 100kc/s – > 1Gc/s.

In this first part the small signal impedance characteristics, Z11 (magnitude and phase) with frequency as seen by the generator, are measured for a single drive coil both with and without the MWO top-load over the lower frequency range of 100kc/s – 20Mc/s. In this measurement the impedance characteristics are dominated by the drive coil which will mask any higher frequency measurements pertaining to the MWO top-load. For this reason the top-load is measured individually in part 2 in the frequency range 100kc/s – 1.3Gc/s. Subsequent parts will look at the overall MWO system impedance characteristics when both transmitter and receiver are combined together in the original Lahkovsky arrangement, and later in an optimised and balanced drive arrangement as designed and presented by Dollard[1].

In this first part the following measurements are presented:

1. Z11 (magnitude and phase) with frequency for a drive coil without MWO top-load in the range 100kc/s – 20Mc/s

2. Z11 for the drive coil combined with MWO top-load, and over the same frequency band.

3. Primary tuning measurements to match the resonant frequency of the primary coil, (with series loaded Cp), to the secondary coil at the fundamental, second, and third harmonics.

The SDR-Kits Vector Network Analyser 3E (VNA-SDR) was used to make all Z11 measurements, and the apparatus and method of measurement is shown in Figures 1 below.

To view the large images in a new window whilst reading the explanations click on the figure numbers below.

Fig 1.1. Shows the overall measurement setup with the VNA-SDR connected to the drive coil by a short SMA terminated RG316 cable, and the other end to the coil feed adapter, and standard calibration modules.

Fig 1.2. The drive coil used in these measurements has 4.5 primary turns of AWG 10 ~2.5mm diameter magnet wire, and the secondary has 248 turns of AWG 24 ~0.51mm diameter magnet wire. The top-end of the primary coil is connected directly to the bottom-end of the secondary coil and forms the negative or ground terminal. The positive or drive terminal is connected to the bottom-end of the primary coil, and the top-end of the secondary can be seen emerging from the coil through a brass bolt which attaches to the driven top-load. Both coils are wound anti-clockwise on the former from the base.

Fig 1.3. In order to make an accurate measurement of Z11 it is necessary to calibrate the VNA-SDR as close to the drive coil as possible. In this case the calibration plane is extended to the input of the primary coil terminals, (two 4mm high voltage shielded terminals), via a signal feed adapter pcb (SFA).  The SFA can be removed from the drive coil by drawing the two-pronged 4mm probes out of the drive coil terminals. With the SFA disconnected the VNA-SDR can be calibrated by fitting an open, short and 50Ω standard load to the end of the SFA. The effective calibration plane then becomes the input to the drive coil, and spurious impedance effects due to any cables and the SFA itself can be removed from the final results. When calibrated, a frequency scan of the SFA with the 50Ω standard load will show a flat impedance line for |Z|, (magnitude of the impedance). The phase of this scan will swing repeatedly between ±180° indicating the near perfect match between the calibration plane and the 50Ω standard load.

Fig 1.4. Shows a close-up of the SFA connected directly to the two high voltage drive terminals in the base of the drive coil. The SFA has an SMA input feed and is then connected via equal weights of copper to the series connection point. The black terminal indicates the negative or ground point where both primary and secondary are connected together, and the red terminal the high voltage feed end of the primary. The series connection point in the positive terminal allows for a calibrated capacitance box to be connected in the primary circuit for tuning measurements. In this picture the series connection is not being used and is terminated with a SMA short. In tuning measurements, when series capacitance CS is added, the SFA must first be calibrated with the capacitance box connected to the SFA with a nominal 1µF set at its output. The 1µF series capacitance has a very low impedance at the measurement band of interest and acts effectively as a short-circuit of the series connector during the calibration procedure. During tuning measurements the capacitance of the box is reduced in the range 100pF – ~ 50nF. The capacitance box itself uses surface mount standard capacitance values and can be reasonably used with SMA connection up to ~ 100Mc/s. The SFA is an unbalanced feed adapter and takes an unbalanced coaxial cable input directly to a balanced two terminal output without any compensation for this change in balance. A similar SFA was also tried which incorporated an RF (upto ~ 3Gc/s) balun in order to effect the transformation between the unbalanced and balanced connections. However, even with calibration the balun SFA proved to be less effective to measure Z11 accurately and cleanly, as it dominated the impedance changes in the frequency band masking changes due to the drive coil further downstream. The standard SFA was therefore used with careful calibration up to the intended reference plane at the input terminals to the coil.

Fig 1.5. Shows the calibration procedure where the SFA is connected in turn to the standard calibration modules, here connected to the short circuit module. In this case the series capacitance terminal is not being used and is shunted with an SMA short.

Fig 1.6. Shows the SFA in measurement configuration and after calibration, where a 1nF series capacitance is added to the primary feed. The calibrated capacitance box can be adjusted in 1pF increments in the range ~30.5pF –  ~10µF, where the parasitic capacitance of the box with SMA cable and when set to zero is ~30.5pF.

Fig 1.7. Shows the overall measurement setup with the top-load connected, and then calibrated and measured in the same procedure as previously discussed. The outer ring of the top-load forms an effective wire extension and capacitive load to the top of the secondary coil, which will both reduce considerably the fundamental resonant frequency and Q-factor of the drive coil. When compared with the drive coil without top-load the outer ring forms a λ/4 driven ring, which is coupling to 11 other λ/2 resonating rings. Given the size of the outer driven ring, and the proximity of the inner rings the coupling factor is considered to be quite high between the driven ring and the inner resonators, which will tend to constrain the free resonation of the inner rings and considerably quench the Q and hence measurements at much higher frequency. This is considered more closely in part 2.

Fig 1.8. Shows in more detail the mounting points at the top-end of the secondary and the MWO top-load.

Figures 2 below show the Z11 impedance characteristics of the calibrated reference plane, the drive coil, and primary and secondary tuning with series added capacitance, of the drive coil only.

Fig 2.1. Shows the calibrated reference plane of the SFA when connected to the 50Ω standard load over the wideband range 100kc/s – 20Mc/s. The calibration is as expected and accurate over the band, with a slight phase variation between ~ 3 – 3.5Mc/s indicating a resonance in the signal path between the calibration plane and the VNA-SDR which has been normalised out by the calibration procedure.

Fig 2.2. The impedance characteristics of the drive coil with the series capacitance Cs = 1µF, showing a wealth of resonant peaks and corresponding phase changes from the fundamental resonant frequency FS of the secondary coil at 1.28Mc/s up to the 12th harmonic FS12 (M12) at 16.51Mc/s. It is interesting to note that the 2nd harmonic FS2 at M2 is a stronger resonance than that of the fundamental at M1. This is further demonstrated in subsequent impedance results. The VNA-SDR allows for only 12 concurrent markers which is why the final two harmonics on the results are not marked. At frequencies above 20Mc/s the rapidly increasing inductive impedance of the primary masks out any further harmonic points of interest, and the impedance curve rises rapidly until the primary coil becomes self-resonant at ~38Mc/s.

Fig 2.3. Shows the calibrated reference plane of the SFA when connected to the 50Ω standard load over the narrowband range 100kc/s – 5Mc/s. With more accurate calibration in the narrowband it can be noted that the phase variation seen in Fig. 2.1 has now been completely normalised out of the reference plane.

Fig 2.4. Shows the narrowband impedance characteristics for the fundamental FS and the first two harmonics FS2 and FS3. Here the difference in magnitude of the resonance between FS and FS2 can be very clearly observed, although the Q of the coil remains very similar for both frequency points. The magnitude of the impedance |Z11| ~ 28.5Ω at FS, and ~ 49.7Ω at FS2 would significantly impede currents developed in the primary from the generator, coupling very little power to the secondary and ultimately to the MWO top-load. The primary circuit will need to be brought into resonance with the secondary at the corresponding frequency in order to considerably reduce the driving impedance seen by the generator, and hence maximise the primary currents, and the amount of power transferred from the generator through the drive coils to the top-load. On the very far left of the scan at 100kc/s the 180° phase change FØ180 of the primary in resonance with the series capacitance CS can just be observed. This will be adjusted by varying CS in subsequent results in order to tune the primary resonant frequency to that of the secondary. Given the difference in magnitude between the fundamental at FS and the 2nd harmonic at FS2 it may be more optimal to arrange the primary to resonate at FS2 rather than the fundamental for normal experimental operation. This will be tested during experiments and reported in subsequent parts.

Fig 2.5. Here the series capacitance CS has been reduced from 1µF to 400pF which has increased FØ180 (the resonant frequency of the primary to M4 at 4.93Mc/s. At this point the impedance seen by the generator has reduced drastically to ~0.72Ω allowing for much larger primary currents. In this case FØ180 does not correspond to a resonant frequency in the secondary so reduced power would be transferred between the two coils. At lower frequencies in the measurement band, the impedance of CS is higher and has the effect of attenuating currents in the primary reducing considerably the magnitude of the lower resonant points, in this case considerably suppressing the fundamental resonance at FS. This scan clearly shows the effect of including a series resistance in the primary and making it resonant within the measurement band of interest.

Fig 2.6. Here CS has been increased to 540pF which has brought the resonant frequency of the primary equal to the 3rd harmonic resonance of the secondary at Mat 4.31Mc/s. As always with coupled resonant circuits, mixing of the signals causes beat frequencies and hence frequency splitting into two sideband frequencies where the impedance minimum points are at both the lower and upper resonant frequencies FL3 (M3) and FU3 (M5) respectively. If driven by a generator with CS = 540pF FL3 would be the best driving point since proximity coupling to the secondary top-load, or discharge of stored energy on the top-load, would momentarily increase the loading on the coil reducing slightly its secondary harmonic frequency towards FL3, and hence maximising transference of electric power between the primary and secondary coils.

Fig 2.7. Shows CS increased further to 1200pF which now tunes the primary resonance to the 2nd harmonic FS2 at M3 a frequency of 2.95Mc/s. Again frequency splitting occurs and the impedance of FL2 and FU2 are very low ~0.5 – 1Ω facilitating high current drive from the generator.

Fig 2.8. Shows CS finally increased to 6500pF which tunes the primary to the fundamental resonant frequency of the secondary FS at M2 at 1.28Mc/s. At FL (M1) the lowest impedance drive point is obtained of just 0.34Ω at 1.17Mc/s. Given the very similar minimum |Z| characteristics of both FS and FS2, when tuned respectively to the primary with corresponding values of CS, either the fundamental or 2nd harmonic could be used to operate the coil in experiments to explore MWO operation and effects.

Figures 3 below show the Z11 impedance characteristics of the calibrated reference plane, the drive coil, and the primary and secondary tuning with series added capacitance, of the drive coil with the MWO top-load.

Fig 3.1. Shows the wideband scan for the drive coil with MWO top-load on the same horizontal and vertical scaling as in Figures 2. The top-load at lower frequencies represents a considerable capacitive load at the top of the coil, very similar capacitively to a toroidal, cylindrical, spherical, or sheet metal top-load added to a conventional Tesla coil apparatus. The frequency of this scan is too low to show any of the high frequency features that result from the resonant rings of the MWO, (see part 2 for these features), but the capacitive loading of the MWO top-load could be clearly seen in this lower frequency scan. The biggest effect is noted at M1 at FS1 the fundamental resonance of the secondary which has reduced from 1.28Mc/s to 650kc/s, a 49.2% reduction . The 2nd and 3rd harmonics at FS2 and FS3 respectively have also reduced but in a much smaller range than the fundamental. FS2 has reduced from 2.96Mc/s to 2.84Mc/s a 4.1% reduction, and FS3 from 4.32Mc/s to 4.21Mc/s a 2.5% reduction. Higher harmonics also reduce with progressively reducing amounts up to FS12 at M12 which reduces from 16.51Mc/s to 16.46Mc/s a 0.3% reduction. It can be seen for this arrangement of drive coil the fundamental resonant frequency would be very sensitive and easily effected by metal loading to the top end of the secondary, the close proximity of metallic structures, or even to the proximity of partially conducting mediums such as the human body.

Fig 3.2. The narrowband scan shows the large change in the fundamental in relation to the 2nd harmonic both in frequency shift and in the magnitude of the impedance and phase change. The MWO top-load has considerably quenched the fundamental of the drive coil, whilst leaving the 2nd harmonic very similar to the coil only results (Figures 2) both in frequency and the magnitude of the impedance and phase.

Fig 3.3. Series capacitance CS = 400pF as before shows the primary resonance in the band, but in this case has completely quenched the already weak fundamental resonance. It can be seen that driving the MWO at the 2nd harmonic during experiments and operation may lead to considerably more stable and reliable system characteristics.

Fig 3.4. Shows the primary tuned at FS3 at M4 at 4.20Mc/s and a slightly increased series capacitance required to tune, increased from 540pF to 570pF.

Fig 3.5. Shows the primary tuned at FS2 at M3 at 2.83Mc/s and again a slightly increased series capacitance required to tune, increased from 1200pF to 1300pF. It can be noted that the fundamental resonance FS can now just be discerned at M1 at 650kc/s.

Fig 3.6. Shows the primary tuned at FS at M2 at 0.65Mc/s with FL = 610kc/s and FU = 700kc/s. The primary is tuned when CS = 26000pF (26nF) which is a very large increase on the 6500pF required without the MWO top-load. At FL (M1) the impedance seen by the generator in the primary is now 0.61Ω which is no-longer the lowest driving impedance. With CS = 1300pF and tuned to FS2 the driving impedance at FL2 is 0.47Ω which indicates that better power transfer from the generator to the secondary could be effected by driving the MWO at the 2nd harmonic. At this point the system would also be more stable and less effected by proximity of metal structures and other partially conductive mediums including the human body.

The lower frequency impedance measurements have shown a wealth of frequency harmonics associated with the Tesla style drive coil, and can be tuned progressively with the primary of the drive coil via series capacitance to make use of the system at different designed resonant frequency points. The MWO top-load at lower frequencies introduces a considerable capacitive load on the drive coil which has a large quenching effect on the fundamental resonant frequency of the system, and suggests a shift of optimal setup away from being driven at the fundamental and towards the 2nd harmonic where greater transference of electric power should be possible, and the system would be more stable and tolerant of loading conditions on the drive coils, the MWO top-load directly, and also from proximity of other conducting and partially conducting mediums. It will be interesting to determine if this changes when both the transmitter and receiver are combined in the full Lahkovksy MWO system, and also whether these effects can be directly demonstrated through experiments in the operation of the complete MWO system.

Click here to continue to part 2 of the multiwave oscillator impedance measurements.

1. Dollard, E., Design and presentation of an optimised and balanced MWO power supply and drive coils., Energy, Science, and Technology Conference (ESTC), 2018.


Vacuum Tube Generator (811A) – Part1

The vacuum tube generator (VTG) mainly used for experiments in the displacement and transference of electric power is based around a pair of 811A power triode vacuum tubes, of either RCA or Russian origin, and with electrical characteristics generally as defined in the RCA 811A datasheet. The 811A’s have demonstrated to be highly flexible, with high reliability, and good overall medium power performance, versus cost and availability, when used in a variety of different configurations. The final output of this generator with these tubes can provide a maximum sustained RF output power of ~600W, and peak output for short bursts (up to 10s) of 900W, and over a wide frequency band up to ~5Mc/s. In addition, the design and implementation of this generator has been arranged in such a way that it can be used in a variety of different configurations, including:

1. A tuned plate class-C Armstrong oscillator which derives automatic feedback from a pick-up coil placed close to the secondary coil.

2. A variable frequency Hartley power oscillator when combined with an independent oscillator drive module.

3. The output stage of a linear power amplifier, when fed with a suitable drive waveform in grounded grid, grid biased, or cathode follower configurations.

4. A modulator stage with suitable cathode keying via a mechanical or semiconductor switching circuit.

5. CW, burst, and modulated oscillation modes when used with or without a dc smoothing capacitor in the plate HV supply.

Other VTGs using the RCA 833A/C and Eimac 4-400A/C are more suited to higher power experiments including, telluric transmission and strong non-linear impulses for displacement experiments, and will be reported in subsequent posts.

Figures 2 below shows an overview of the VTG apparatus, complete with the high voltage supply, the independent Hartley power oscillator module (HPO), the high voltage bridge rectifier (HVBR), and the supply dc smoothing capacitor.

The circuit diagram for the VTG and peripherals is shown in Figure 2 below, or click here to view the high-resolution version.

The high voltage supply VHT (described here), is connected to the input of the HVBR. The rectifier consists of 4 sets of 2 x HVP20 20kV 750mA high voltage diodes in parallel and arranged in a bridge configuration. The output terminal of the bridge B is connected to a tank capacitor 60µF 4kV k75-40a Russian pulse capacitor which results in a smoothed dc output suitable for rapid charging and discharging through a primary, or as a stabilised dc supply for the VTG plate circuit.

A dc meter is also so arranged on the output B to read from 0-4kV in configurable steps based on where the base terminal of the meter is connected to a resistor divider. The high voltage dc meter is a simple arrangement using a 1mA fsd (full-scale deflection) ammeter connected to a resistor divider with 4 x 1MΩ resistors. With all four resistors in series 4kV will result in a 1mA passing through the meter moving it to fsd, with one resistor in series 1kV will result in meter fsd, and accordingly for the 2 and 3kV scales. The meter face has been recalibrated to indicate needle reading in kV.

Before considering each of the configurations listed above in detail, it is necessary to cover the basics of the VTG design. The plate supply is provided by the HVBR output B+ and is a variable 0-4kV peak supply either smoothed or unsmoothed based on whether the rectifier tank capacitor is connected or not. The plate supply B+ is connected directly to the driven load LPCP, which in this case is the primary coil in parallel with the primary tuning capacitor. In this arrangement the primary load (parallel resonant circuit) would present the highest impedance to the vacuum tubes when at resonance. When the primary capacitor is so adjusted to set the secondary resonance frequency it is optimal for the impedance of the primary load to be equal to the output impedance of the dual 811A vacuum tubes in parallel. In this way maximum power is developed in the primary load or through the primary of the coil.

It is important to note that when using a VTG the primary circuit, (in this case the load LPCP), is NOT arranged to resonate at the same frequency as the secondary circuit, which avoids large primary currents being developed in the VTG which would lead to excess plate dissipation in the vacuum tubes, and rapid degrading or destruction of the tubes. Rather the impedance of the primary load at the resonant frequency of the secondary coil defines the driving impedance presented to the vacuum tubes, which in turn are adjusted by the grid bias or feedback to match their internal impedance to the driving impedance, leading to maximum power being developed in the primary circuit whilst not overloading the vacuum tubes beyond their maximum ratings.

Hence the VTG drives the coil in a linear sinusoidal CW mode with optimally arranged load impedance matching, as compared with for example, a spark gap generator where the primary and secondary resonance frequencies are arranged to be equal to allow maximum power transfer from the maximised primary currents during the ring-down phase of the primary tank discharge.

The load of the primary is connected to the plates of the vacuum tubes via high frequency chokes L1R1 and L2R2. These chokes quench very high frequency oscillations which can be generated by discharges within the vacuum tubes during overload conditions, and in turn help prevent very high frequency oscillation run-away conditions which can lead to rapid tube destruction.

The cathodes of the vacuum tubes are bridged by RF bypass capacitors to minimise the impedance of the high frequency signal path, and then connected via a sequence of jumpers J1-J3 to allow for different configuration modes. The cathodes are further connected to the heater power supply which provides a constant bias to the cathode combined heater element of the 811A. The heater power supply is a mains step-down transformer so arranged to provide an ac supply of 6.3VRMS @ 8ARMS for two 811A heater circuits in parallel. The heater supply transformer is fine adjusted in this case by R5 which reduces the voltage across the heater elements and has an RF bypass capacitor C5 to again minimise the impedance of the high frequency signal path. The ac voltage across the tube heater elements is monitored using a calibrated 0-10VRMS ac voltmeter, and R5 adjusted, (in this case 4 x 0R1 25W series connected resistors), to provide an optimal 6.3VRMS +- 5%.

When the vacuum tube generator is used as a linear power amplifier in grounded grid (cathode driven) configuration it is necessary to prevent the RF input signal from feeding back into the heater supply and being dissipated in the low impedance power supply stage. To prevent this bifilar high frequency chokes can be used between the tube cathodes and the heater power supply. At low-frequency, (50Hz for the heater supply), current can pass through the choke from the heater supply to the tube cathodes, but the RF signal at the cathode is prevented from feeding back into the heater supply. Modulation and switching via cathode keying can also be arranged via mechanical or semiconductor switches, and allows for a range of “switched” experiments important in the comparisons of electrical phenomena experienced in the exploration of the displacement and transference of electric power.

The grid of the VTG is arranged with jumpers J4-J6 to allow for different configuration modes, and a current limit resistor R3 to restrict the maximum grid current below the nominally rated for the 811A, and according to the configuration it is used in. In addition both 811A’s are forced cooled by the fan driven by the auxiliary 15V supply. Force cooling allows for higher sustained power in CW modes, and affords an additional protection during peak power overloads. It is quite normal for the 811A plate to glow slightly red under higher sustained powers, and in peak power for short periods 0-10s for it to glow intensely red. Sustained high peak powers > ~650W will lead to plate dissipation overload combined with flash-overs between the plate and grid causing rapid grid damage. In most experiments in the displacement and transference of electric power I have found a comfortable sustained power between 400-600W, which leads to long vacuum tube life, and well sustained tube characteristics according to the nominal data.

Construction of the VTG and HVBR modules are shown in detail in Figures 3 below. For ease and simplicity of experimentation the VTG is open assembled on a simple board with simple insulated mountings and a combination of metal, plastic, and wooden mounts for the various components. Whilst this does lead to a very quick and flexibly modifiable prototype generator, there are significant EMC, interference, temperature, and stability benefits to housing the entire VTG build within a screened metal case, with directed cooling inlet and outlets, and with careful consideration to connection of high current and tension paths with minimal inductance copper bars etc.

Figures 4 show the construction of the Hartley power oscillator module which is considered further in configuration option 2 below.

The different configurations of the VTG module are now considered in more detail:

1. Tuned plate class-C Armstrong oscillator

This configuration is very well suited to investigations at the upper and lower resonant frequency of the coil being driven (FU and FL). In this case the VTG is a series-fed version of the Armstrong oscillator deriving the grid bias feedback from a pickup coil placed in proximity to the secondary coil. In the case of the flat coil this is a small 10 turn cylindrical coil (diameter 100mm) mounted behind the secondary coil and on axis with the coil centres. Oscillation via the pickup coil feedback automatically keeps the VTG on the resonant frequency being explored and adjusts automatically to maintain the resonant frequency when loading is applied to the secondary coil outputs.

In this configuration J1-J3 are left open and J4 is connected. Circuit operation is as follows. When first turned on the impulse current from the tank circuit B+ conducting through the vacuum tube causes a ringing oscillation (ping) in the primary circuit LpCp. This oscillation couples to the secondary circuit LsCs which is further coupled by the pickup coil to the grid bias leakage circuit. When the phasing of the pickup coil is the correct way round for positive feedback, the grid bias leakage capacitor C1 becomes negatively charged during the positive half cycle in the primary circuit, pushing the grid voltage down and progressively restricting conduction in the vacuum tubes towards the off state with much reduced plate current. The negatively charged grid leakage capacitor C1 then discharges through the grid leakage resistor R4. As this happens the grid voltage on the vacuum tube starts to rise progressively towards 0 volts turning on the vacuum tubes with an increasing plate current. The plate current through the primary circuit LpCp again is coupled to the secondary LsCs and the cycle repeats. With the grid bias leakage circuit correctly adjusted the VTG will oscillate with a linear sinusoidal output optimised for maximum plate voltage and current swing, (maximum power transfer at the resonant frequency of the secondary), whilst keeping the grid bias currents within the maximum ratings for the vacuum tubes used.

When setting up this mode of operation it is most important to ensure correct phasing of the pickup coil in connection to the grid bias leakage circuit, and that the values of C1 and R4 are suitably adjusted to allow maximum swing of the plate circuit whilst keeping the grid bias within the maximum ratings. For this VTG with 2 x 811A vacuum tubes R4 is optimally between 1-1.5kΩ and C1 between 1.5-3nF. Considerable power dissipation occurs in R4 when running the VTG at higher powers > 400W, hence the need for a wire-wound power resistor (100W), and preferably as part of the forced cooled air circulation. Excessive or too little grid bias current will lead to a distorted and clipped oscillation, or in extreme cases, no oscillation at all and rapid vacuum tube degradation or destruction. Initial setting up is best done with low plate voltage ~ 500V and higher values of R4. R4 can then be progressively reduced to increase grid bias whilst ensuring a clean and stable output oscillation.

As seen in previous impedance measurement posts for the flat coil (1S-3P) the resonant frequency of the secondary can be adjusted by changing the capacitance of the variable vacuum capacitor in the primary Cp. When adjusted the oscillating frequency of the VTG automatically changes to track the changes in the secondary resonant frequency. At high values of Cp > ~650pF the impedance of the lower resonant frequency FL is dominant and the final oscillating frequency can be adjusted around this frequency FL in the range 1.5-2.2Mc/s. At low values of Cp < ~650pF the impedance of the upper resonant frequency FU dominates  and the final oscillating frequency is in the range 2.7-3.8Mc/s.

Adjusted in this fashion this configuration of the VTG is very well suited to continuous linear measurements for single wire currents, displacement and transference of electric power, and telluric transmission experiments. This configuration is best suited to exploring frequency regions centered around FU and FL, where oscillation is stable and conduction currents in the secondary of the experimental coil are > 10mARMS. This configuration is not suitable for exploring frequency regions far from resonance, at very low bias currents, or between the transition between FU and FL. In these cases it is necessary to use the VTG in modes 2 or 3 where accurate progressive frequency control is provided by an external source and the VTG acts as a power stage, whether that be as a tuned power oscillator, or as a linear power amplifier.

2. Variable frequency Hartley power oscillator

The Hartley power oscillator converts the VTG to a linear oscillator which can be adjusted for variable frequency in bands defined by the combination of band capacitors connected on the HPO module. The frequency of oscillation of the VTG is now determined by the resonant circuit formed on the HPO board, the pickup coil of configuration 1 is not connected in this arrangement. This configuration is suitable for measurements across the entire frequency band of interest, in the case of the flat coil 1S-3P between 1.5-3.5Mc/s.

The HPO must be manually tuned or retuned to a specific frequency of interest and particularly when changing the loading of the secondary coil circuit. The resonant frequency of the secondary coil is very sensitive to electrical loading, temperature, material losses, changing boundary conditions and proximity, which all cause deviations of the configured frequency. Any change in this tuned frequency  requires re-adjustment making the HPO configuration not well suited to experiments designed to explore different loads and operating conditions. It is however very suited to exploring circuit operation where the loading conditions are relatively fixed, and particularly in off-resonance, low current, and frequency transition regions.

In this configuration J1-J3 are left open and J5 is connected to the Hartley oscillator module along with the plate voltage and RF ground as shown on the circuit schematic of Fig 2. The bands of frequency can be adjusted by hardwired jumpers on the HPO module, where the static band capacitors are combined with the 1000pF variable capacitor, and form a parallel resonant circuit with L5 the HPO coil with an inductance of 3.0uH. The available bands are broadly as follows (static combinations of capacitors shown only):

a. 1 x 500pF = 2.4 – 4.1Mc/s

b. 1 x 1000pF = 2.1 – 2.9Mc/s

c. 1 x 500pF + 1 x 1000pF = 1.8 – 2.4Mc/s

d. 2 x 1000pF = 1.7 – 2.1Mc/s

e. 1 x 500pF + 2 x 1000pF = 1.5 – 1.8Mc/s

The setup of the HPO again requires a balance between largest plate swing (output power) without distorting the output wave, whilst restricting the grid bias within maximum parameters. The required frequency band is first configured using the hardwired jumpers, and the HPO grid bias variable resistor is set in the higher halve of its resistance range . The VTG plate supply is first set low at 500V and the HPO coil tapping point starts close to the RF ground end. The tapping point can be progressively moved upwards towards the plate voltage end of the coil until a point is reached where the output of the VTG is stable, clean, and with good power output as the plate supply is progressively increased up to the maximum ratings for the vacuum tubes. The HPO grid bias resistor can then be progressively reduced keeping the grid bias current within the maximum recommended. If no oscillation can be obtained the grid bias resistor can be progressively reduced until oscillation starts, whereupon the other adjustments discussed can be continued with.

The HPO can then be varied across its band frequency range according to the experimental requirements of the circuit, and the output power adjusted using the plate supply voltage. If the loading on the secondary coil changes significantly the HPO will need to be re-tuned and/or re-adjusted to provide stable power oscillation. Off resonance of the secondary coil in the experiment may require the HPO tapping point to be re-adjusted towards the RF ground to prevent excessive power dissipation in the coil. During continuous operation the temperature of the HPO inductor coil should be monitored in order to prevent over-heating. Re-adjustment will be required when moving often and rapidly between on-resonance and off-resonance operating points. As previously stated this configuration mode is best suited to exploration of operating points not accessible with configuration 1, and where the operating conditions do not change rapidly during the measurement cycle.

3. Linear amplifier output stage

The VTG can be operated as the power stage of a linear amplifier when correctly configured and connected to a suitable frequency generator with power amplifier output stage. No pre-amplifier stage is currently provided in the VTG so the driving signal source will need to generate an output between ~ 1-10W in order to the drive the VTG output to a usable output power. In this configuration the VTG can be driven in grounded grid, grid biased, or even as a cathode follower with some change of circuit connection and setup.

For grounded grid operation J1 is connected to the external signal source, and J6 is connected directly to RF ground. For grid driven mode J1-J3 are open and J5 is connected to the external signal source which in this case also needs to arrange for the driving signal to provide suitable dc grid biasing for the vacuum tubes used. For this reason grounded grid operation is preferred for simple linear amplifier operation.

When correctly setup and driven the VTG in this configuration can provide a very stable oscillation output which can be easily and finely adjusted by the external signal source, overcoming some of the adjustment problems of the HPO, but not exceeding the very good power output levels of the HPO. It has been found that the HPO is better when higher power experimentation is required, but the linear amplifier is better for overall signal stability and accurate adjustment.  This configuration also allows for non-sinusoidal waveforms to be applied to the experimental system.

4. Modulator stage

Modulation of the VTG is currently by cathode keying, allowing conduction through the vacuum tubes to be switched at low frequencies via mechanical switches and relays, or at higher frequencies < 100kc/s  by semiconductor MOSFET switches. Modulation of this kind was originally considered to be important in the exploration of the displacement of electric power where non-linear events play a very significant part in the unusual electrical phenomena observed within an electrical system. It has since been found through experimentation that the non-linear impulses generated by this modulation method are not of sufficiently low transition time and low pulse width to be particularly useful in the generation, observation, and measurement of displacement phenomena.

5. CW and burst oscillation modes

The VTG can be arranged in any of the configuration modes 1-4 and then operated in CW or burst modes simply by removing the plate supply tank capacitor at B+. When the tank capacitor is connected at the output of the HVBR a constant plate supply voltage B+ is applied to the primary load. In this case the output of the VTG in oscillation will be in CW mode providing a constant and continuous oscillation wave to the primary circuit LpCp.

When the tank capacitor is removed from the HVBR the output is an unsmoothed full wave rectified supply at 100c/s (based on UK line frequency). Applied to the VTG this produces bursts of oscillation inside the supply envelope, and has proven to be useful in establishing certain operating conditions in experiments orientated to displacement of electric power. These operating conditions and experiments are currently work in progress and will be reported in subsequent posts.

Overall the VTG has proven itself to be a versatile and reliable linear power source suitable to drive a wide range of experiments to explore the displacement and transference of electric power, and also some preliminary lower power telluric transmission experiments. Over several years of operation only 1 x 811A has been replaced after grid flash over destroyed one of the tubes when being used in configuration 2 with the HPO at an off-resonance operating point, with large output mis-match and high reflected power. The VTG has also sustained reliably through high power experiments where both plates of the tubes are glowing bright red for short periods of time. The 811A in RCA and Russian forms have proven themselves to be robust and reliable provided the heater element (spring tensioned) has not broken during extended storage. VTGs with other vacuum tube types and configurations will be presented in subsequent posts.

Click here to continue to part 2 of the 811A vacuum tube generator where the operating characteristics for each of the defined configurations are measured, optimised, and tuned.