What core to pick for a GDT

[Zipdox]

I assume the color has some sort of meaning. Videos on youtube suggest to avoid the light green ones. One person also said to avoid ones from computer power supplies, but that's the only source I can find for them really. So which one from the pic should I pick?

[klugesmith]

In my limited experience, computer PS have two kinds of ferrite cores.
E-I shape in high frequency transformer or flyback.
Toroid, with two identical windings,  in common-mode choke at AC input.  I think made of lossy ferrite material.

As for color codes, I never heard of a standard that would apply to more than one maker of toroids.

[Zipdox]

I googled CH330066 and found this:

Powder cores are distributed air gap cores made from ferrous alloy powders for low losses at high frequencies. Small air gaps distributed evenly throughout the cores increase the amount of Direct Current (DC) that can be passed through the winding before core saturation occurs. Molybdenum Permalloy Powder (MPP) cores are ideal for low loss inductors such as switching regulators and noise filters.

High Flux, Sendust and Mega Flux cores are the preferred choices for Power Factor Correction (PFC), switching regulator inductors, in-line noise filters, pulse and flyback transformers and many other applications requiring low losses at high frequencies.

[ritaismyconscience]

Don't use powdered cores. I've had pretty good luck using the white cores on the EMI filters.

They look like this:


Just unwind the wires and remove the white plastic. There should be a ferrite core inside, which works very well for GDTs.

[Zipdox]

Don't use powdered cores. I've had pretty good luck using the white cores on the EMI filters.

They look like this:


Just unwind the wires and remove the white plastic. There should be a ferrite core inside, which works very well for GDTs.

I just tested the powdered core and it was absolute trash. I also tested the dark green one and it worked pretty nicely. The dark green one was part of an EMI filter in the PSU and from videos I've seen the dark green ones should work well.

[prabhatkumar]

I can send aliexpress link of the core which i used for my own driver ( Though it blew up ) but still the gdt works well you . You can select the size of the core as per your requirements in that page . The one here is a bit small maybe but should work ( I guess).
 If you want to salvage them then try checking out some 3 phase inverter or something like that sort . The three phase wires are wrapped around the toroid.
https://www.aliexpress.com/item/32861649468.html?spm=a2g0s.9042311.0.0.27424c4dIDbkW4

[johnf]

You need a core with high permeability ie ui of around 1000 or more
powderediron cores have permeabilties of up to around 100 that is why they do not work well for GDT's
as pointed out emi cores are usually of high permeabilty ferrite, but they can also be lossy by design and not quite so good for GDT's. Another tye of core is to be found on VGA leads ie a very large bead these are good also but you will not be able to use cat5 or 6 cable due to the center hole not being big enough. The internal twisted cores from the CAt 5,6 cable wil however fit and the ferrite is usually a slightly better grade loss wise than the mains common mode choke type.

[John123]

Fair-rite 77 for up to 100khz and fair-rite 78 for 200-500khz, random cores can work but its potluck.

[Mads Barnkob]

Do not try to reuse unsuitable material ferrite/powder cores to be used in a high frequency driver/feedback application. Shit in, shit out.

I recommend this guide: https://thedatastream.4hv.org/gdt_index.htm

[rikkitikkitavi]

In my limited experience, computer PS have two kinds of ferrite cores.
E-I shape in high frequency transformer or flyback.
Toroid, with two identical windings,  in common-mode choke at AC input.  I think made of lossy ferrite material.

As for color codes, I never heard of a standard that would apply to more than one maker of toroids.

They come in four varieties:
* Input filter chockes at high voltage- typically small common mode chockes made out of lossy ferrites
* PFC correction inductor, large (a few cm ) - powdered iron cores- totally unsuitable for GDT due to losses and leakage inductance (they have low Al value and hence ration Lm/Lleak will be unfavourable)

*Large EI type cores- can be used, but will likely be to large and induce an unfavourable Lm / Lleakage ration

*Output filter inductors- powdered iron cores

Of these I would be for input filter chocke inductors of suitable size (design as a transformer) - yes, they are lossy, but only at high frequency.
at typical drive frequency the losses are managable, and anyhow efficiency of a few hundred milliW transfered GDT power is of little issue.

But dedicated cores are better of course.

Important here is how the secondary voltage looks, a good drive rise/fall and no ringing etc.

Mads posted a very good link. A GDT is only as good as its windings!


( I guess now that all the posts here in this thread points in more or less the same direction  but it is not unwise to repeat important things...

[ElectroXa]

Hi 

there is also a fifth type of core, although it's less common in appliances.

Indeed, I found some Metglas core in some appliances, like this core :
https://zupimages.net/up/20/12/xhah.jpeg

There is an overview of Metglas properties here : https://metglas.com/magnetic-materials/
Does anyone have tried this material for GDT, and is Metglas suitable for GDT construction?

Thanks 

[ritaismyconscience]

Metglas doesn't work

Tried it and it caused the capacitor on the output stage of the driver to start sparking.

[John123]

Hi 

there is also a fifth type of core, although it's less common in appliances.

Indeed, I found some Metglas core in some appliances, like this core :
https://zupimages.net/up/20/12/xhah.jpeg

There is an overview of Metglas properties here : https://metglas.com/magnetic-materials/
Does anyone have tried this material for GDT, and is Metglas suitable for GDT construction?

Thanks 

Oh that looks interesting, like some kind of ferrous tape rolled up in a donut shape.

[davekni]

Metglas is a brand name for amorphous magnetic materials.  They're formed by extremely-rapid cooling (molten metal poured onto the circumference of large water-cooled spinning disk).  The rapid cooling reduces crystal formation, making more amorphous glass-like material.  Amorphous metal is just one material used for tape-wound cores such as the link posted previously.  They are typically used around 20kHz.  Tape-wound cores are available in other materials as well, including standard silicon-iron (electrical steel), cobalt alloys, etc.  Most tape-wound cores are larger than what is useful for Tesla coils.  I used a silicon-iron tape-wound core with 50um thick material for a 1kHz application at work.  The amorphous metal drastically reduces hysteresis losses in the core, but is only a little better for eddy-current.  Amorphous metal tape-wound cores use very thin tape to minimize eddy-currents.

[Zipdox]

Hi 

there is also a fifth type of core, although it's less common in appliances.

Indeed, I found some Metglas core in some appliances, like this core :
https://zupimages.net/up/20/12/xhah.jpeg

There is an overview of Metglas properties here : https://metglas.com/magnetic-materials/
Does anyone have tried this material for GDT, and is Metglas suitable for GDT construction?

Thanks 

Yeah I have one of the cores in a plastic cover. I tried it and it's absolutely worthless for a GDT.

[klugesmith]

I wonder how come it didn't work. 
Maybe eddy current losses get you, at very high frequencies, in spite of the thin laminations.
As Dave warned, Metglas conducts like a metal (because it is one).

I think amorphous metal toroid cores are used in some GFCI current transformers. 
Can be laminated as a stack of ring-shaped punchings instead of tape-wound.
High permeability and low hysteresis are handy when you need to detect 5 mA in a single turn winding.
Low remanent flux is valuable when you need to remain sensitive after a strong magnetizing pulse, like from a short-circuit ground fault.

[davekni]

Can be laminated as a stack of ring-shaped punchings instead of tape-wound.

I didn't realized it was ever stamped and stacked.  Wonder if that is just for low-frequency applications like GFI with thick laminations.  I'd read that amorphous metals require very hard tooling for stamping.  They also naturally come as long ribbons due to how they're made.  Perhaps the ribbons are too wide to use for small GFI cores, and perhaps stamping is easier than slitting wide ribbons.

[klugesmith]

Dave, I failed to confirm the stacked-rings construction during 1/4 hour of searching on Internet.  That doesn't mean it does not exist.  If I ever saw it with my own eyes, it would have been after removing the 1000 turns of fine wire from a GFCI transformer.

Not counting stacked rings of regular transformer steel sheet, from more than 100 years ago, before it was superseded by tape-wound strips.  Aside from ease of fabrication and lack of waste, winding performs better because of grain orientation.  Stacking could save a step if you were making an air-gapped toroid.