A simple coast station Marconi transmitter is shown above.  It has higher power and a couple of modifications over the simple unit previously described.  The modifications include a reactance regulator, a safety gap and a quenching gap.

Power is supplied, not by a battery, but by a direct current motor driving an alternating current generator as shown in the top left of the schematic.  Typically the generator would supply supply 200-500 volts at 500 Hz.  The DC motor's rotor will draw an excessive amount of starting current (due to the lack of back EMF since the rotor isn't spinning) and so is reduced by the boxed 'starting switch'.       The starting switch box puts various amounts of current limiting resistance in series with the rotor.  As the motor gains speed, and the back EMF increasing, the operator moves the switch to the right in steps until it reaches its end of travel.  At that point the motor is up to speed and all the resistance is switched out.  Ten seconds was the usual start up time.  Later versions of the starting switch required no operator action.  The arm was pulled across the contacts by a solenoid, acting against the resistance of a dashpot.  The variable resistors allow some control over the RPM of the motor, and hence the generator's frequency, and output voltage.        The photo to the left shows the motor generator control panel from Estevan Point Wireless.  The motor starting switch is located at the middle bottom.  Mains and generator knife switches are in evidence.  Good practice positioned the knife switch handles so that gravity would tend to open the circuit, not close it--hence placing the handle in the top position energized the circuit.  Hand wheels are connected to the variable resistances.   Motor/generator input voltage and current meters are set along the top. Jack Bowerman photo.

     The operator of the station shown in the schematic had to take care where he touched his Morse key as there would be several hundred volts across its contacts.  Closing the key would place the generator's alternating voltage across the transformer primary windings and be stepped up to some 10-25,000 volts.  The arc across the spark gap would not only discharge the capacitors and cause the antenna system to ring at its resonant frequency, but would place a short circuit across the secondary transformer windings.  This short would be reflected back to the primary side and effectively shorting out the generator if it were not for the action of the reactance regulator in limiting current flow. 

    The discharge of the capacitors causes the aerial circuit to ring at its resonant (transmitting) frequency via the coupling coils.  Of course the antenna currents will couple back across into the capacitor circuit and mix.  The result is a transmitted signal that radiates some interfering sidebands.  The fix is to fit a 'quenching gap'.  The main capacitor discharge can blast through this gap, but only for a few cycles, in other words 'quenched'.  The antenna gets hit with these few large amplitude cycles, and rings for a few cycles at the transmitting frequency.   Also the reflected energy from the antenna does not have sufficient power to jump the quench spark gap.  It has been found by experimentation that 24 cycles of RF on each spark discharge was found to be the best in terms of tuning and reduction of adjacent frequency interference.

    If the key was held down the resulting radio frequency would appear as a group of individually decrementing sine waves, the group spacing proportional to twice the generator frequency (sparks take place at the positive and negative swings).  An operator at a distant receiver would hear a tone in his headset corresponding to the group spacing.  In most cases this would be 1000 Hz.

    This transmitter could be fitted with an AC motor and run off commercial power.  Doing this creates a small problem.  Commercial power is 60 Hz which will create a 120 Hz tone in a receiving operators headset--too low a frequency to be heard properly in the headphones at the receiving station.  The solution is to fit a rotary spark gap.

	The simple rotary gap at the left is simply composed of an electric motor spinning a heavily insulated disc.  Around the periphery of the disc are fitted metal plugs, in this case two.  As each plug is rotated into the centre of the spark gap, a spark ensues, and when the plug rotates out, the spark stops.  The high voltage isn't able to jump the gap without the assistance of the metal plug.  Thus the transmitter will have a tone corresponding to the speed of the motor and the number of metal plugs.
	It can be shown that this simple rotary gap can be improved upon by connecting it to the shaft of the motor and making the number of electrodes equal to the number of poles on alternator powering the transmitter.  The spark can now be adjusted to occur at any point on the alternating voltage cycle.  Apparently the best time to have the spark happen is when the alternating voltage from the transformer was at the near zero point.  At this point the charge on the capacitors is free to jump the gap without any 'interference' from the generator voltage.  The synchronous spark gap gave a very distinct tone, easy to copy through atmospheric and man made interference.      To the left is Jack Bowerman's photo of Victoria's synchronous spark gap.  The toothed wheel is another form of a rotary disk.  As can be seen it is mounted on the same shaft as the electric motor and generator and thus in synchronism with the generator's poles.  The teeth are fitted into brass ring mounted on the circumference of an insulating disc.  The high voltage appears at the top insulators and arcs as each pair of teeth rotates into position.  The knob at the right of the gap assembly is a screw adjustment permitting some slight advance or retard between the fixed and rotating gaps. This would be adjusted for the best sounding tone at a receiving station.
	The American Marconi Company's "Oscillation Transformer" as used in their 2 KW transmitters.  The coupling between the aerial and transmitter circuit was varied by rotating the top coil as shown.      In the schematic at the top of the page, a "short wave" switch changes the resonant frequency of the antenna system, and thus the radiated frequency.  Generally two frequencies would be used--one would be 600 Meters (500 kHz) and either double (1200 Meters (250 kHz)) or half (1000 kHz).  If the capacitor was switched into the circuit, the frequency would be lowered.  Since the antennas were never long enough, a loading coil would be required.  This was simply an inductance in series with the antenna lead (aerial tuning inductance).     The operator could make adjustments to the tuning by adjusting taps on the oscillation transformer and the aerial tuning coil and noting the affect on the RF ammeter.  The object was to tune for maximum current.     Photo from Fleming's Elementary Manual of Radio Telegraphy and Telephony 1915.
	Typical high voltage transformer of the time.  Often the case would be full of insulating oil.  Note the 'safety spark gap'.  If, for any reason, the main spark gap became disconnected for some reason, the high voltage would jump this gap instead of inside the transformer.  If that happened, the expensive transformer would require replacement.    Photo from Fleming's Elementary Manual of Radio Telegraphy and Telephony 1915.     Next page shows all these elements identified at a west coast Canadian station in the 1910-15 period.