A spark gap will have a very repeatable breakdown voltages for a given atmospheric conditions. For mostly mechanical reasons, uniform field gaps (using, for example Rogowski or Bruce profile electrodes) are not used as much as sphere gaps where the spheres are quite a bit larger than the gap. There isn't a convenient analytical expression for the breakdown voltage as a function of sphere diameter and gap, as there is for a uniform field gap, however, there is a lot of empirical test data, and sphere gaps are by far and away the most common way of measuring high voltages with a spark gap.

Typical accuracies are 3% for gaps less than half the diameter of the sphere and 5% for the gap larger than the diameter of the sphere. As the gap gets larger, the field between the spheres gets more and more nonuniform, and as a result the scatter in the data gets larger. A rod gap represents sort of the ultimate in non-uniform gap, and is often quoted at +/- 8% accuracy.

Sphere Gap Breakdown Voltage Table

A series resistor is usually put between the source and the gap to limit the breakdown current and to provide some damping of the high frequency oscillations. It is typically 100K to 1 Meg for AC or DC voltages, and no more than 500 Ohms for impulse voltages.

For AC peak and DC measurements, the voltage is gradually
increased until breakdown occurs. The mean of 5 measurements that
fall within ± 3% is used as the value. For impulses, a 50%
flashover voltage is calculated from the mean of two
measurements, described as follows, which must be within 2% of
each other. The first measurement voltage is set so that out of
10(?) impulses, either 2 or 4 flashovers occur. The second
measurement is set so that out of the 10 impulses, either 6 or 8
flashovers occur. *(N.B.Presumably, there is some ANSI
specification for this process, which will be used to update this
section.)*

Sphere gaps can be arranged either vertically, typically with
the lower sphere grounded (earthed), or horizontally. The
surroundings do have an effect on the breakdown voltage, as they
alter the field configuration. Standard clearances are specified
for spheres of various sizes in both configurations. These
clearances reduce the effect of the surroundings to less than the
specified accuracy (e.g. 3%). In the following: **D**
is the diameter of the spheres, **S** is the spacing
of the gap, S/D <= 0.5. **A** is the height of
the lowest point of the HV sphere above the ground. **B**
is the radius of clearance from surrounding structions.

D (cm) |
A (max) |
A (min) |
B (min) |

<= 6.25 | 7*D | 9*D | 14*S |

10-15 | 6*D | 8*D | 12*S |

25 | 5*D | 7*D | 10*S |

50 | 4*D | 6*D | 8*S |

100 | 3.5*D | 5*D | 7*S |

150 | 3*D | 4*D | 6*S |

200 | 3*D | 4*D | 6*S |

Actual values (meters)

Sphere Diameter (cm) |
A (max) meters |
A (min) meters |
B (min) for max gap (D/2) meters |

<= 6.25 | 7*D | 9*D | 14*S |

10-15 | 60-75 | 0.80-1.20 | 0.60 |

25 | 1.25 | 1.75 | 1.25 |

50 | 2.0 | 3.0 | 2.0 |

100 | 3.5 | 500 | 3.5 |

150 | 4.5 | 600 | 4.5 |

200 | 6.0 | 800 | 6.0 |

The insulator supporting the upper sphere should be less than 0.5 D in diameter. The sphere itself should be supported by a conductive metal shank no more than 0.2 D in diameter and at least D in length (that is, the sparking point should be at least 2D from the lower end of the upper insulator).

The high voltage lead should not pass near the upper electrode. Ideally it should be led away from shank avoiding crossing a plane perpendicular to the shank at least 1 D away from the sphere (i.e. 2 D away from the sparking point, until it is outside of a sphere of radius B from the sparking point.

The top of the lower electrode should be at least 1.5D above the (presumably) grounded floor.

Horizontal gaps are much the same as vertical gaps, except that both electrodes are insulated. The insulators should be longer, at least 2D long (putting the sparking point at least 4D from the supports: 2D for the insulator, 1D for the shank, 1D for the sphere). And, both spheres should be the appropriate clearance from the floor or external objects.

Copyright 1998, Jim Lux / sphgap.htm / 5 May1998 / Back to HV Home / Back to home page / Mail to Jim