Do This Modification At Your Own Risk

Above:  1440 rpm Four Pole Motors (Becomes 1500 rpm)

Above:  2880 rpm Two Pole Motors (Becomes 3000 rpm)

Above:  The Machined 'Flat' Is Ringed In Red. (motor is a two pole 2880 rpm. Flat = 40%)

The Rotor Modifications For a Synchronous Motor.

Do These Modifications At Your Own Risk

To make a Synchronous Rotary Spark Gap you will need either a synchronous motor, or an ordinary AC induction motor converted to 'salient pole' operation. The word 'salient' in this context meaning dominant.

The rotor of a synchronous or salient pole motor will always rotate at the same speed as the rotating magnetic field created by it's surrounding stator. The speed of this rotating field is dictated by the frequency of the AC mains supply and the number of poles the motor has.

Whilst an Asynchronous motor (an ordinary induction motor) will always rotate slightly slower (2% - 6%) than the rotating magnetic field. This is called slippage


The Theory

Let us suppose we have a synchronous or salient pole motor running while you are monitoring the AC voltage sine wave of its supply. Then when the supply voltage is at its peak, you were able to magically freeze both the motor's shaft and the AC sine wave.
If you then made a mark on the shaft in relation to the casing, you would find that every time the marks subsequently aligned, the AC wave will always be at its peak. This is the best time to arrange for the gap to fire, so you can be sure the capacitor will be as fully charged as it is ever going to be. (In reality though the firing point needs to be slightly retarded)
Dependant on the motor's speed and number of poles it has, you may have either 2 or 4 positions (or even more) around the path of the shaft where the AC voltage peaks.

A normal induction motor's rotor, as I mentioned, revolves slightly slower than the magnetic field that is influencing it, but by machining some flats on the rotor you cause it to lock too, and rotate at, the same speed as the rotating field.
As this speed is directly linked to the frequency of the AC sine wave mentioned above, you have achieved your goal of making sure the capacitor is as ready as it will ever be when it is discharged.

A normal induction motor before modification, is an Asynchronous motor because it is not linked to the AC cycle. This is because the rotor is revolving slightly slower than the rotating magnetic field created by the stator. So if you do the same magic trick above and freeze the action and mark the shaft and casing, you will find that they will not align on the next revolution of the shaft.
This means the mains cycle could be at any point on its sine wave curve when the electrodes align. The capacitor therefore may not be fully charged resulting in a missed firing. If you also have a resonant condition in the charging network this could result in very high voltages occurring that can damage the capacitor and transformer.
Asynchronous motors can still be used as long as the breaks per second (bps) of the spark gap is higher than around 400. This is because the effect of an undercharged coil is not so great then.


An excellent in-depth analysis of rotary spark gaps can be found here at the site of very knowledgable Richie Burnett.


If the motor is a 1440 rpm (1800 on 60Hz) it has four field windings and will need four flats milled onto the rotor, each at 90 degrees to one another. If the speed is 2880 rpm (3600 on 60Hz), it only has two field windings and consequently only needs two flats at 180 degrees to each other.
After modification the speed will (should) have increased a little to either 1500 or 3000 rpm (1800rpm or 3600rpm on 60 Hz), and it may run a bit hotter with less power.

The amount removed when forming the flat is critical. Too little and the motor will not be synchronous and its speed will constantly surge or 'hunt'. Too much and it will loose too much power and overheat.

With four pole 1440 rpm motors I measured the overall diameter of the armature and removed one quarter of this distance as a flat.
For a later two pole 2880 rpm motor that I also modified, the width of the flat was 40% of the rotor's diameter.
I did try 30% initially and this worked with just the bare 10 inch disk, but when I added all the Tungsten electrodes and their holders the added weight caused it to 'unlock' and it become asynchronous again. By removing another 10% to make it 40% overall, the motor then worked perfectly in synchronous mode.

Both of my modifications resulted in no discernible loss of torque and very little, if any, increase in the heat generated. The 1500rpm motor was a 0.5 Hp so any loss of power would have been noticeable, whereas the 3000 rpm is a 2 Hp, so it had plenty of leeway.




A Good Example of A Poorly Designed SRSG

The 1500 rpm motor had an 8 inch rotor with the electrodes on a 7 inch PCD. The 8 electrodes giving a break rate of 200bps at 1500 rpm. As the tungsten electrodes are 0.25 inch diameter the mechanical dwell time of the rotor was very poor. I attempted to overcome this by having staggered (in the vertical plane) electrodes.
This is achieved by having the right hand one adjustable up or down.

As can be seen despite the staggered electrodes I still suffered badly from power arching, caused I think by the closeness of the electrodes on the disc, and possibly the fact that I used a conducting ring on the rear of the rotor. Also the motor was only 1500 rpm, so for that reason I built the one shown below with a bigger rotor.

Above is my second attempt using a 10 inch diameter disc. This is still 200bps but without the conducting ring on the rear of the disc. This meant I needed two pairs of stationary electrodes, with the front set positioned at the rotor's mid height (say 9 o'clock) and the other rear set (hidden) at the 4:30 position on a clock face. When this is combined with four rotating electrodes on a 1500rpm motor it gives 200bps.



Unfortunately the 0.25 inch diameter electrodes combined with the 1500 rpm, still gave a poor dwell time, with power arcing at much over 3KW. For that reason I then put the 10 inch disc onto a 2Hp 3000 rpm motor, which meant I only needed four revolving electrodes and one stationary pair.

200bps 2 Hp 3000rpm

Above is shown the finished working SRSG (my third attempt) that used my 10 inch rotor from the second build but this time mounted on a 3000rpm 2 HP motor. It has two stationary electrodes and four revolving. All electrodes are 0.25 inch Tu.
The safety gap, which can partly be seen at the rear, consists of a 'horn' gap using two 0.25 inch brass rods. At the time my MMC handled 38Kv and I was using a pig with a BIL rating of probably in excess of 60Kv, so I set the safety gap to 9mm. At ~30kv/cm this gives a breakdown at ~27kv.
The horn shape means it acts like a mini Jacob's ladder and the arc extinguishes easily.

It would seem the old adage 'third time lucky' is true as can be judged from the performance shown below. (Stills were taken from a short video)


February 14th 2010 Update:

The coil was originally tuned to 11 turns untill I added another 10nF. To offset this I reduced the tap by half a turn which is what you see below. That was just an approximate guess so the tuning is probably a bit astray at the time of the test.

The only way to accurately measure spark length is a straight line between two fixed points. So when a ground strike occurs it is very difficult to accurately locate the point of the strike when you are some distance away. This ground strike's straight line distance was measured the next day by setting things up in the same position, and it appears to be at least 120 inches, also doing a reconstruction with the method used below(**) reinforces this. A question mark hangs over this though as John Freau's formula for spark length does not support 120 inch performance based on my power level. My power input according to PSpice simulations of my SRSG using my known component values, is 4.59kw, meaning John Freau's formula would only give me a length of 115 inches maximum. ** In addition to above I made a 3D model (Rhino) of the positions of the various points in the picture based on actual measurements taken on the ground. The longer streamer length of the two pictures then measured as 144 inches. This is obviously wrong, but it adds some credence to the 120 inches physically measured at the estimated strike location. Because of local opposition to the coil (see here) I will have to wait for a better opportunity to make some accurate and controlled measurements - rather than (most likely) wishful thinking!

Using both the methods mentioned above this was estimated at 100 or more inches, so a conservative 8 foot. The majority of the streamers were hitting the shrubbery behind unfortunately, so any real objective testing is not possible in the garden. This is a classic example of building a coil bigger than your surroundings can really accommodate - I doubt I will be the last to do that though!