GDT electrostatic shielding / Tesla Coils / Forums

Forums

4hv.org :: Forums :: Tesla Coils

« Previous topic | Next topic »

GDT electrostatic shielding

Move Thread

LAN_403

Steve Ward

Mon Aug 11 2008, 08:21AM

Registered Member #146 Joined: Sun Feb 12 2006, 04:21AM
Location: Austin Tx
Posts: 1055

It was mentioned before that multiconductor shielded cable was good for GDTs because you can use the shield as the primary and each conductor as a secondary, producing a very low leakage inductance GDT. Then it was said that even better is triaxial cable, where the middle shield is connected to the low side IGBT so that the capacitance between the high side winding and the driver winding could be blocked by the middle shield connected to -Vbus.

So ive been using plain old multiconductor cable now, but im wondering if i shouldnt add simply an electrostatic shielding between the conductors and my primary. This seems to achieve the same thing as triaxial cable (except i get more conductors in there) if i connect the middle shield to -Vbus. Is this really equivalently good?

I just so happen to have a large amount of 2 conductor shielded audio cable, and an absurd amount of old RG-58 coax. Ive tested that i can quite easily and cleanly remove enough shield from the RG-58 and slip it over my audio cable, and presto, i have my 2 isolated screens. Hopefully tomorrow i can make a new GDT for my CM300 driver and maybe somehow notice a difference between having the extra screen in there, vs not.

Any ideas how i can quantify the difference? Maybe i will see less EMI on the LV side of things when i scope it out? Currently there is a lot of noise on everything.

If all goes well, i will document how to make the ultimate GDT

Bennem

Mon Aug 11 2008, 08:53AM

Registered Member #154 Joined: Sun Feb 12 2006, 04:28PM
Location: Westmidlands, UK
Posts: 260

Hi Steve,
This might be of interest to you,
its a cut taken from 'GDT arcing problem' thread.
Its written by Richie Burnett in answer to using screen stereo cable for GDT design.

"It does indeed achieve the lowest possible leakage inductance. The only downside is that it results in a relatively high inter-winding capacitance between secondary connected to the bridge leg mid-point (source terminal of the top MOSFET) and the low-voltage primary side. This tends to capacitively couple the high dv/dt from the bridge leg mid-point back into the drive electronics. The symptom is narrow but intense spiking of the 0v rail on the low-voltage drive side when the power devices switch.

If you have tri-axial cable you can get close to the ideal gate-drive transformer for a bridge leg by winding it like this... The drive side primary consists of the outer screen and therefore its flux encompases the two inner conductors. This guarantees high coupling factor to both secondary windings and low leakage inductance. The secondary driving the low-side MOSFET consists of the inner screen, and the secondary driving the high-side MOSFET consists of the centre conductor of the cable. When constructed in this way the low-side secondary winding (which is connected to DC bus negative) acts as an inter-winding faraday screen between the noisey bridge output voltage on the centre conductor and the sensitive low voltage drive electronics on the outside shield. All of the high frequency hash from the bridge leg mid-point is capacitively coupled back to the DC bus negative. It is kept on the power side where it belongs, and out of the control and driver electronics."

All the best,

Mel.

Steve Conner

Mon Aug 11 2008, 09:30AM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

I used Faraday shielding on all my GDTs and CTs for the last few years. My coils have been reliable, but I've no proof that this was due to the Faraday shielding.

On my isolated gate drivers, I found that I could actually spike them on with dV/dt, and introducing a Faraday shield between primary and secondary of the signal isolating transformer cured that.

Steve Ward

Mon Aug 11 2008, 04:19PM

Registered Member #146 Joined: Sun Feb 12 2006, 04:21AM
Location: Austin Tx
Posts: 1055

Hey Mel, that was actually the response by Richie that i was referring to

.

I suppose i feel now that the extra shield is simply good engineering, so im gonna try it out and see if its not too troublesome to implement. Unfortunately, this means retrofitting 7 DRSSTC drivers with new GDTs... that will be a good days work for sure (especially when i have to custom make the cable!).

Conner, do you have a picture of how you did the ES shield on your CTs? Was it just a metal tube around the primary conductor?

I used Faraday shielding on all my GDTs and CTs for the last few years. My coils have been reliable, but I've no proof that this was due to the Faraday shielding.

Well, i have not used any faraday shield, and my coils have been 100% reliable so far too, though i have had noise problems, which only showed up when adding the secondary coil.

Mathias

Mon Aug 11 2008, 07:36PM

Registered Member #1381 Joined: Fri Mar 07 2008, 05:24PM
Location: Hungary
Posts: 74

by "custom made" triax do you mean that "any"* coax cabel will do if you just pull yet another shielding on it ? (maybe also isolate with some liquid pvc or just tape to finish)

kinda having a hard time finding allready made cable in my country.

by the way , if it "work" the way it was described , then a coax with some wires twisted on it would also work ,least it would be alot easier to do.

*(had some of trouble with those cheap bubble/foam PE type coax cabels)

also for example to drive a fullbridge what would be "better" ?
- to use a bifilar triax winding on a single core (kinda large core will be needed for that 16-20+ turns)
- or two separate GDTs for each half bridge ? (2 smaller cores)

Don't really see any real difference.

Steve Ward

Mon Aug 11 2008, 08:23PM

Registered Member #146 Joined: Sun Feb 12 2006, 04:21AM
Location: Austin Tx
Posts: 1055

by "custom made" triax do you mean that "any"* coax cabel will do if you just pull yet another shielding on it ? (maybe also isolate with some liquid pvc or just tape to finish)

Yes, but in my case i have initially tried making the triaxial wire that has 2 inner-most conductors in parallel, and then 2 shields outside of that.

So after some initial testing, im not seeing a real benefit yet of this faraday screen. My testing method is simply my scope probe looking at the noise on the clock input to the flip flop (just an arbitrary node in the circuit). Realizing that my scope is a ground path as well, ive tried many configurations of where to tie the shield and whether or not to connect my PCB ground to the heatsink ground.

Connecting a .1uF coupling cap between -Vbus and the heatsink seems to make a difference in all cases, usually making things slightly worse always. I also noticed that the long the wiring was for this capacitor connection (like if i used a clip lead), there was enough EMI generated to start messing with my fluorescent lamp! The noise on my board was also tremendous. Shortening the lead length seems to make a very big difference. I suppose there must be some current flowing through that cap :P.

More testing for me, this time i'll take some notes of the details of each setup so i can maybe get a more conclusive answer as to whether the faraday shielding is worth the hassle or not.

Tom540

Mon Aug 11 2008, 09:47PM

Banned on 3/17/2009.
Registered Member #487 Joined: Sun Jul 09 2006, 01:22AM
Location:
Posts: 617

I'm guessing no because if the coil is already 100% reliable then why bother? Maybe try a .001uF cap to heat sink instead.

Steve Ward

Mon Aug 11 2008, 11:37PM

Registered Member #146 Joined: Sun Feb 12 2006, 04:21AM
Location: Austin Tx
Posts: 1055

Hi Tom,

One of the reasons i was going down this path was because of some other noise related issues, and i thought perhaps the GDT capacitive coupling was contributing to the problem. The noise would cause extended pulse widths as either my fiber optic Rx would extend the pulse (note that its a DC response receiver!), or my D-FF would keep re-triggering. I have verified both of these issues with the scope, but ive also managed to work around them in various ways like adding small filter caps certain inputs on the ICs, and extending the electrostatic shielding on the fiber Rx.

Anyway, from what i can see, the faraday shield is offering me no extra reduction in noise on my PCB vs a GDT without the faraday shield. I did find that connecting the PCB ground to the heatsink DID increase the noise because it was providing a good return path for the currents coupled through the GDT, just as expected. The faraday shield honestly didnt seem to help this a whole lot.

What i did finally try was a simple common mode choke on the primary side of the GDT (no faraday shield). I use coax to run to the GDT itself to keep all loop inductances low, but this allows me to use a more generous length of wire, so i pass it through a small ferrite about 4 times. This cleans up the noise on the PCB nicely. Adding a common mode choke on my scope probe helped even more :P. The important thing now is that im able to connect my PCB ground to the heatsink ground, and keep my .1uF decoupling cap between -Vbus and heatsink. All of this is in the hopes of making the electronics of the system look like a big chunk of metal when it comes to a streamer firing its way back at the tesla primary.

Any thoughts on this common-mode choke solution? It seems like a decent fix for rejecting switching noise, and is a heck of a lot easier to add to all of my drivers than rewinding with triaxial wire.

GeordieBoy

Tue Aug 12 2008, 01:37PM

Registered Member #1232 Joined: Wed Jan 16 2008, 10:53PM
Location: Doon tha Toon!
Posts: 881

> Any thoughts on this common-mode choke solution? It seems like a decent fix for rejecting switching noise, and is a heck of a lot easier to add to all of my drivers than rewinding with triaxial wire.

I've written at some length about this issue in the other thread Steve W.

To combat common-mode noise capacitively coupling back through the GDT, take the following 3 actions in this order of preference:

1. Keep the dv/dt rate on the power side down to something sensible. As low as you can get away with. The faster the voltage slews, the more displacement current it will try to drive across the GDT's isolation barrier.

2. Reduce inter-winding capacitance in the GDT, with careful layout, or preferably by winding such that the low-side secondary encloses the high-side secondary forming a faraday screen. This returns all high-frequency shit back to the power side DC bus negative where it belongs.

3. As a last resort try putting a common-mode choke on the wires between the driver IC's and the primary winding of the GDTs. This will increase the loop impedance in the common-mode current path and decrease the peak current that a given dv/dt will be able to drive around this loop.

Although the purpose of the common-mode choke is to add lots of common-mode inductance, it will also add a small amount of differential mode inductance which will appear like leakage inductance of the GDT and degrade gate-drive waveforms slightly so check them after it's addition.

-Richie,

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob