Should you ground the driver circuit?

[Zipdox]

Should the driver circuit GND be connected to earth?

[davekni]

My preference is to ground the driver circuit.  That way if any stray fields attempt to charge the driver circuit, it won't cause any breakdown of GDT insulation.

[Zipdox]

My preference is to ground the driver circuit.  That way if any stray fields attempt to charge the driver circuit, it won't cause any breakdown of GDT insulation.

To follow up on this (sorry for necropost), wouldn't this cause interference if the bottom of the secondary is grounded too? Is it a bad idea to ground the secondary to mains ground? There's a lot of dispute around the internet about this.

[davekni]

Grounding is a surprisingly complex topic, and not just for Tesla coils.  It's also a debated issue for circuitry inside computer cases and almost any other metal-housed electronics device.  For electronics, the issue with noisy grounds is usually EMI (electromagnetic interference), exceeding regulatory limits.  Issues are concentrated around cables entering/leaving and where to connect cable ground and shield wires.  That's why ferrite beads are so commonly integrated into power supply cords and monitor cables and such.  The internal design couldn't keep ground clean enough, so EMI needed to be attenuated externally.

Probably the best-known grounding strategy is called "single point ground".  That works really well for analog audio equipment.  Tesla-coil frequencies aren't all that much higher for primary resonance, but the H-Bridge switching edges have a lot of energy at much higher frequency.  H-Bridge and controller circuitry are more reliable with local ground planes given local high-frequency current.

The single-point strategy has all the grounding/grounded elements wire separately to a common point somewhere.  That could include your counterpoise,  control circuitry, H-Bridge VBus (through small Y-caps if electrically hot), and bottom of secondary coil.

In some ways, the version where bottom of secondary and counterpoise are wired together but not to line ground is a variation of single-point ground, where the single point is where your breaker box is tied to a ground rod and/or water piping.  But that's only partial, as the control electronics and VBus share a line-wiring path back to this "common" ground.  Still may be OK.

I prefer the opposite approach of one large ground-plane.  I lay aluminum screen and/or foil and/or sheet metal around and under my coil and to my scope and VBus power supply.  (I always monitor performance with a scope when running.)  Everything is grounded to this metal plane, including line ground.  It also has a lot of capacitance to physical ground.  My ground-plane is an over-kill, but makes accurate scoping easier.  (I have large ferrites around my scope cables as well.  Some day I need to test how far I can back off and still keep reliable operation.)

[Zipdox]

Grounding is a surprisingly complex topic, and not just for Tesla coils.  It's also a debated issue for circuitry inside computer cases and almost any other metal-housed electronics device.  For electronics, the issue with noisy grounds is usually EMI (electromagnetic interference), exceeding regulatory limits.  Issues are concentrated around cables entering/leaving and where to connect cable ground and shield wires.  That's why ferrite beads are so commonly integrated into power supply cords and monitor cables and such.  The internal design couldn't keep ground clean enough, so EMI needed to be attenuated externally.

Probably the best-known grounding strategy is called "single point ground".  That works really well for analog audio equipment.  Tesla-coil frequencies aren't all that much higher for primary resonance, but the H-Bridge switching edges have a lot of energy at much higher frequency.  H-Bridge and controller circuitry are more reliable with local ground planes given local high-frequency current.

The single-point strategy has all the grounding/grounded elements wire separately to a common point somewhere.  That could include your counterpoise,  control circuitry, H-Bridge VBus (through small Y-caps if electrically hot), and bottom of secondary coil.

In some ways, the version where bottom of secondary and counterpoise are wired together but not to line ground is a variation of single-point ground, where the single point is where your breaker box is tied to a ground rod and/or water piping.  But that's only partial, as the control electronics and VBus share a line-wiring path back to this "common" ground.  Still may be OK.

I prefer the opposite approach of one large ground-plane.  I lay aluminum screen and/or foil and/or sheet metal around and under my coil and to my scope and VBus power supply.  (I always monitor performance with a scope when running.)  Everything is grounded to this metal plane, including line ground.  It also has a lot of capacitance to physical ground.  My ground-plane is an over-kill, but makes accurate scoping easier.  (I have large ferrites around my scope cables as well.  Some day I need to test how far I can back off and still keep reliable operation.)

I'm building the electronics of my coil into a computer PSU casing. The ground of the power connector is wired to the case with a spade connector. Would it be acceptable to wire the bottom of the secondary to this connector too?

[davekni]

Yes, that's fine for secondary grounding.  But, don't bundle the secondary ground wire with any control or H-Bridge circuitry.  Keep it separate from the coil to that ground point.  It's best to also have a counterpoise - some conductive sheeting on the ground around the coil.  If you include a counterpoise, wire the secondary directly to the counterpoise, then ground the counterpoise to that IEC connector ground lug.  That keeps the highest-frequency components of any ground strike more local.