Low-capacitance transformer?

[jan.didden]

Hi,

I am designing a high-voltage amplifier to directly drive ESLs. I'm not ready to disclose the design yet, but I DO need a power transformer with a large isolation between primary and secondary, at least 3kV peak, as well as very low capacitance between the primary and secondary.

I've got a quote by one manufacturer which specs 250pF between prim and sec which is too much. I want to get (far) below 100pF.
I need about 20W power, 115/230V prim. Sec voltage should be 6V nominal.

I *think* I might need to wind my own, like on a C-core with physically separated primary and secondary. Anyone knows where I could buy kits of cores and bobbins or any other solution?

thanks,

jan didden

[Andrew Eckhardt]

I suppose your circuit swings the power supply secondary with audio voltage and therefore needs low capacitcance to primary? The best way to drop capacitance is to increase the transformer frequency, which makes all parts of the transformer smaller.

Maybe include a switching power supply with your amp? Or maybe HF linear.

[jan.didden]

Originally Posted by Andrew Eckhardt

I suppose your circuit swings the power supply secondary with audio voltage and therefore needs low capacitcance to primary? The best way to drop capacitance is to increase the transformer frequency, which makes all parts of the transformer smaller.

Maybe include a switching power supply with your amp? Or maybe HF linear.

Yes. I did that for the screen supply of the upper tube (those CCFL transformers are great & cheap!). But I also need a solution for the heater (6V @ 2.7A). I'm hesitating to go the DC-DC converter route as I'm not familiar in that field and I want to build an audio amp not embarking on a new field of study.

What do you mean by 'hf linear'?

jan didden

[zigzagflux]

I would say you would choose either a split bobbin or electrostatic shield. You can reach less than 10pF with a properly arranged shield between windings.

[Andrew Eckhardt]

For "HF linear" I meant sine drive to the transformer primary as from something like a class AB power amplfier. Probably too complicated to be worthwhile. I mean it would be easy enough to wire up but a lot of circuitry for the job.

Maybe for only 20 watts you could make a 60Hz cut core or any dual bobbin (as in one bobbin on either side of the core) arrangement work for less than 100p parasitic. I'd figure you'd want to select a bobbin for the secondary that will get you a square winding cross section (to minimize overall winding surface area, and therefore parasitic capacitance) while allowing for quite a bit of insulation/distance to the core. A full shield can drop the overall pri-sec coupling capacitance but of course it doesn't drop the capacitance from secondary to shield.

[jcx]

lowest pri-sec C in standard xfmrs comes from physical separation of the windings - "worldtrans" international accepted double reinforced insulation mains transformers with split bobbin construction have 5-10x less parasitic coupling C than layer wound types (toroids are the worst)
dual bobbin "semi-toriodial" C, R core types are intermediate in coupling C since they still put pri, sec side by side in both bobbins to cut leakage

the double insulated mains xfmrs must pass 4KV hi-pot but may not be safe to operate at continuous 3KV - possibly chaining one (or more) similar construction, physical size 1:1 mains isolation xfmr in series would give you a acceptable hobbyist solution (add V equalizing R divider??)

I have measured few 10s of pF in the dual bobbin EI types - series connection can cut that by a factor of N

I'm not saying that would be "safe" - you decide

electrostatic shields are not helpful in this application - multiple shields can give impressively low in-out C numbers at the cost of high winding to shield C - so you have to have a lo Z "AC gnd" (or signal node in Jan's app) termination to "bootstrap away" the shield to sec C

[jan.didden]

Originally Posted by zigzagflux

I would say you would choose either a split bobbin or electrostatic shield. You can reach less than 10pF with a properly arranged shield between windings.

Well to be specific, I would need low C from the secondary to the rest of the world...

jan didden

[jan.didden]

Originally Posted by Andrew Eckhardt

For "HF linear" I meant sine drive to the transformer primary as from something like a class AB power amplfier. Probably too complicated to be worthwhile. I mean it would be easy enough to wire up but a lot of circuitry for the job.

Maybe for only 20 watts you could make a 60Hz cut core or any dual bobbin (as in one bobbin on either side of the core) arrangement work for less than 100p parasitic. I'd figure you'd want to select a bobbin for the secondary that will get you a square winding cross section (to minimize overall winding surface area, and therefore parasitic capacitance) while allowing for quite a bit of insulation/distance to the core. A full shield can drop the overall pri-sec coupling capacitance but of course it doesn't drop the capacitance from secondary to shield.

Yes that would work, at least if I can insulate the sec to at least 6kV peak. Are there any companies that sell this type of xformer cores and bobbins?


jan didden

[jan.didden]

Originally Posted by jcx

lowest pri-sec C in standard xfmrs comes from physical separation of the windings - "worldtrans" international accepted double reinforced insulation mains transformers with split bobbin construction have 5-10x less parasitic coupling C than layer wound types (toroids are the worst)
dual bobbin "semi-toriodial" C, R core types are intermediate in coupling C since they still put pri, sec side by side in both bobbins to cut leakage

the double insulated mains xfmrs must pass 4KV hi-pot but may not be safe to operate at continuous 3KV - possibly chaining one (or more) similar construction, physical size 1:1 mains isolation xfmr in series would give you a acceptable hobbyist solution (add V equalizing R divider??)

I'm not saying that would be "safe" - you decide

electrostatic shields are not helpful in this application - multiple shields can give impressively low in-out C numbers at the cost of high winding to shield C - so you have to have a lo Z "AC gnd" (or signal node in Jan's app) termination to "bootstrap away" the shield to sec C

I just looked at isolation transformers from Coilcraft and Halo; some are spec'ed for 4kV RMS isolation which would be perfect. They are developed for DC-DC converter chips from TI or Maxim.
Now if I could find a ref design...
I guess this type of transformer would have low secondary capacitance? Not spec'ed as far as I could see.
Anybody has experience with diy dc-dc converters?

jcx you have a link to some of this 'worldtrans' stuff?

jan didden

[pieter t]

Hi Jan,

I use c-cores exclusively for power transformers and all kinds of audio transformers, mainly for tube amplifiers.
Besides expertise in winding technology for electrostatic (high voltage) applications is nearby.
And that in your own country....maybe you drop me a note to see what can be done.