Grounding issues...

[Manz]

Hello everyone.
Here you can find a brief overview of my coil before talking about the problem that popped up the other day...

Inverter: skm100 fullbridge with a 4700uF 450V electrolytic capacitor
MMC: 97.5nF 8kV (4S10P 39nF 2kVdc mkp capacitors)
Secondary coil: 10*38cm 0.2mm wire. 100*38cm topload around 125khz resonant frequency
Driver: ud2.9 in pulseskip mode
Primary: 10turns, 4mm copper tube



The coil has a 24V psu that powers the driver and a relay for the bus capacitor precharge and a 12V one to power a fan.
I connected the 12V-24V psu grounds and the negative output to the coil's RF ground together with strike rail,inverter heatsink and driver's enclosure.


I run the coil for several minutes when all of a sudden I saw a small spark on the 24V psu screw terminals shorting ground to line tripping the breaker.
It seems like everything is working fine and just the 24V power supply died so it will be an easy fix.
I guess for some reason (probably bad grounding) the RF ground potential was high enough to jump to the nearby line, how can I avoid that? Should I just leave everything floating so it doesn't happen again?
I will for sure find a way to improve the RF ground like using a metal mesh but now I'm scared of it happening again....

[davekni]

Many builds include "Y" cap(s) from H-bridge power to ground.  That provides a path for any arcs that do manage to hit primary coil.  Not certain if that is your issue or not.

I will for sure find a way to improve the RF ground like using a metal mesh

Yes, some form of counterpoise along ground/floor is good practice.  Aluminum foil works fine, though not as durable as other solutions.

[Manz]

I'm thinking about buying some chicken wire to make a decently sized counterpoise (probably around 5 square meters) and I will also add the 100nF "Y" caps.

I still don't know if it's better to leave the power supplies floating or to connect the to RF ground like I did

[davekni]

I still don't know if it's better to leave the power supplies floating or to connect the to RF ground like I did

My preference is to ground everything possible.  Any floating metal has the possibility of picking up electric field and then sparking to something.
BTW, last month after reworking a bit of my QCW coil, I forgot to connect ground to buck converter (which also grounds following H-bridge for coil primary).  After a bit of running, fried the 19.5V laptop supply that feeds buck converter and H-bridge gate drive circuitry.  Replaced that supply and connected ground and it runs fine.

[Manz]

good to hear the power supply was the only damaged part.
Do you also connect the negative output of the power supply to RF-ground or just the shielding?

[davekni]

Do you also connect the negative output of the power supply to RF-ground or just the shielding?

All grounds connect together for my coils.  Includes counterpoise, boxes for electronics, control supply (19.5V for most of mine) negative side.  Vbus- (H-bridge negative) is tied either through "Y" cap or directly to ground too.  I built an isolated PFC supply for Vbus for my normal DRSSTC and my QCW, so Vbus- can be grounded.

[Manz]

Thank you for your insight I really appreciate it.
I'd like to hear if someone else had the same grounding problem I had with my coil and how they solved it

[AstRii]

All grounds connect together for my coils.  Includes counterpoise, boxes for electronics, control supply (19.5V for most of mine) negative side.  Vbus- (H-bridge negative) is tied either through "Y" cap or directly to ground too.  I built an isolated PFC supply for Vbus for my normal DRSSTC and my QCW, so Vbus- can be grounded.

Isn't the earth eventually connected with neutral in the fuse box? If you connect Vbus- directly to earth, aren't you shorting out a diode in the bridge rectifier?

Also, are you sure this is a good practice? Wouldn't it be better if driver is insulated from earth so that any ground strikes won't interfere with the circuit as much?

[davekni]

Isn't the earth eventually connected with neutral in the fuse box? If you connect Vbus- directly to earth, aren't you shorting out a diode in the bridge rectifier?

My coils are not typical.  I've built isolated "PFC" units to generate Vbus for my coils.  That way I can adjust Vbus without a Variac, and ground Vbus- for easy scoping.  (My largest Variac is 15A at 120V, not enough for larger coils.)  My isolated "PFC" units limit power to what line can handle (10kW for my 240Vac unit and 1.8kW for my 120Vac unit) including during initial bulk cap charging.
For normal coils, a Y cap is recommended between Vbus- and ground to handle any stray secondary current (arc to primary or just capacitive coupling).

Also, are you sure this is a good practice? Wouldn't it be better if driver is insulated from earth so that any ground strikes won't interfere with the circuit as much?

Insulated from earth ground is fine as long as insignificant current manages to get from secondary to primary.  I prefer to protect from that, with a Y cap if using a conventional line bridge or voltage doubler.

Good questions.  Please feel free to challenge my reasoning here.

[AstRii]

My coils are not typical.  I've built isolated "PFC" units to generate Vbus for my coils.

Alright, that makes sense now.

Insulated from earth ground is fine as long as insignificant current manages to get from secondary to primary.  I prefer to protect from that, with a Y cap if using a conventional line bridge or voltage doubler.

You mean as during a primary strike? I agree at this point that the primary strike current must return back not to interfere or destroy transistors but wouldn't it be better to use Y cap to earth from one of the inverter outputs? Wouldn't that prevent stray currents to flow through IGBTs and limit the overvoltage caused by a primary strike?

Please feel free to challenge my reasoning here.

I'm not sure if I'm competent yet for such task but in my experience having the driver not insulated from earth, causes EMI issues. For example, in one of my coils the UD+ driver's on board FPGA would "restart" everytime when the ground strike rail was hit.

[davekni]

I agree at this point that the primary strike current must return back not to interfere or destroy transistors but wouldn't it be better to use Y cap to earth from one of the inverter outputs? Wouldn't that prevent stray currents to flow through IGBTs and limit the overvoltage caused by a primary strike?

Y caps from inverter outputs might be better for primary strikes, but would inject high frequency noise into line ground and add capacitive load to H-bridge.  I think secondary current is generally low enough for IGBTs to handle.  The issue without Y caps to Vbus is that H-bridge voltage rises enough to break down GDT insulation.

I'm not sure if I'm competent yet for such task but in my experience having the driver not insulated from earth, causes EMI issues. For example, in one of my coils the UD+ driver's on board FPGA would "restart" everytime when the ground strike rail was hit.

First, yes, isolated electronics can work fine when shielded.
When grounded, as there any path through driver ECB ground?  In other words, was ground current likely to pass through driver?  For example, if driver power supply has a path back to line ground, perhaps a second ground connection is causing current through board.  Ideally board ground plane would handle that without causing any disruption, but ground planes are not perfect.
FPGAs are quite sensitive.  I use FPGAs regularly for interrupters and other optically-isolated purposes.  For these, FPGA board is grounded, but has no copper connection to power circuitry.  My only project with FPGA driver is my latest QCW coil.  FPGA board is grounded and adjacent H-bridge, isolated only by GDT insulation.

Thank you for sharing your experience.

[Manz]

Just a little update.
In the end I ended up grounding the secondary to RF ground and I left everything else inside floating and it seems to work allright.The RF ground is a groundrod that goes about 90cm deep into the soil in my garden.

I'm still not sure why it should be better to ground everything together.

If we assume the RFground voltage does not rise during operation then everything should be fine but if the RFground is bad  (for example if the soil is very dry) I expect the voltage to start rising and if it gets just to a couple of Kv it could arcover to mains wires or inside grounded power supplies (as I think happened to me) . Am I getting this wrong?

[davekni]

If we assume the RFground voltage does not rise during operation then everything should be fine but if the RFground is bad  (for example if the soil is very dry) I expect the voltage to start rising and if it gets just to a couple of Kv it could arcover to mains wires or inside grounded power supplies (as I think happened to me) . Am I getting this wrong?

Sounds reasonable.  Two other considerations:  First, this presumes that all strikes from top load hit something tied to your RF ground, not to any other objects, especially floating ones.  Second, any ground strikes, even to strike ring or other object tied to RF ground, have high frequency components.  Top load is suddenly discharged.  That current pulse may make a momentary large voltage drop through inductance of long wire to ground rod.  I don't have any data to indicate if this is a real concern or not.  I try to avoid any such long paths just in case inductance is an issue for ground strikes.