Transformer shielding - Page 2 - diyAudio
Go Back   Home > Forums > Amplifiers > Power Supplies

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 30th September 2012, 04:20 PM   #11
diyAudio Member
 
Join Date: Mar 2008
Antonio,

the electrostatic shield is inside the transformer, in between windings. Its purpose is to remove (in theory. in practice it can only reduce) electrostatic (capacitive) coupling. to do that it must be conductive, and connected (via a low impedance connection - thats where theory smacks head-on into practice) so a suitable "AC return" (in a smps we're talking about a node that is DC - +Vdc_in, +12V_out etc) - a node whose voltage ISNT bouncing up and down a lot (i = C*dV/dt is what its trying to stop). you've got that.

problem is, its sitting inside the main field of the transformer. in order for flux to get from one winding to the other, its got to pass through (or "cut") the shield itself.

this will generate eddy current losses in the shield - little loop currents will flow in the shield, regardless of where or even if the electrostatic shield is connected.

so the internal electrostatic shields need to be thin - if they are > five skin depths thick, they will completely block (read as: turn into heat) the field from passing through the shield. this would be bad.

this casually ignores the actual field distribution and the effect of the core, but the guts of the argument is valid.

so you use the thinnest material you possibly can get. And even better is to use a more resistive material == less conductive => skin depth is smaller for that material.

A second reason to make the electrostatic shield thin is that it makes magnetic coupling worse - not just because of eddy current losses (which reduce transferred power) but because there is now a gap between the windings. the bigger that gap is, the worse the coupling.

I like to think of internal electrostatic shields, and external flux shields, in terms of tightly and loosely coupled coils. its not the correct way to analyse them, but it is easily understandable.

Belly bands are loosely coupled, which can be modelled as a very high leakage inductance => constant current source (kinda. it is as leakage -> infinity, but its a useful approximation to make). so the losses are I^2*R_bellyband where I is constant. ergo minimum loss needs a low resistance belly band. (this is actually a reasonable approximation)

electrostatic shields are tightly coupled - so low inductance. Its completely wrong, but I've always thought of them as the dual of belly bands - tight coupling => voltage source => V^2/R_electrostatic_shield losses. therefore we need to maximise electrostatic shield resistance. Unfortunately this is an analogy not an approximation - which is obvious when you start to look for the return path from the shield "resistance" - there isnt one, the losses are caused by eddy currents.

so its a crappy analogy for the electrostatic shield. but it does make it easy to remember.

And finally re. belly bands:
be careful with externally gapped core legs - eg in a flyback transformer. almost all of the stored energy is in the gap, and a reasonable chunk of that bulges out the sides of the gap - fringing flux. just like with a winding, it can be good to space the belly band out a bit from the external gaps (eg with tape) otherwise the belly band might intercept a fair chunk of the fringing flux. one or two gap lengths is enough, and as long as the belly band is much wider than the gap it'll work very well.

personally, I never use external gaps - only ever center-leg gaps. I have to be careful about extra winding losses, but my xfmr doesnt spew flux everywhere.

Last edited by Terry Given; 30th September 2012 at 04:26 PM.
  Reply With Quote
Old 30th September 2012, 07:25 PM   #12
diyAudio Member
 
Join Date: Nov 2010
Location: The City, SanFrancisco
Terry

Thanks for your detailed reply and insight.
Must warn you I am fairly slow (usually I eventually get it), so far I understand the practical aspects (spacing and increased leakeage inductance) but still confused about the shield cutting the flux.
I can see how it cuts the leakage flux but still not seeing how it cuts the core flux.
So for the sake of arguement if one assumes there is no leakage flux would the shield still have eddy currents induced?

Thanks
-Antonio
  Reply With Quote
Old 30th September 2012, 10:00 PM   #13
diyAudio Member
 
Join Date: Mar 2008
Antonio,

my pleasure.
Oh, I see. consider a single wire, in free space. the H field is a series of circles surrounding the wire, in equal amounts at any radial angle. thats the key. the H field decreases with increasing distance from the wire, out to infinity (or at least far enough that we cant measure it any more)

now lay the wire down on a flat piece of ferrite, about 5 magnetic skin depths thick (so no flux makes it out the other side of the ferrite). the H field distribution changes, but the wire is still surrounded by H field. its just on the ferrite side all of the H field has been squashed into the ferrite, rather than extending out to infinity.

but on the opposite side to the ferrite, the H field still extends out forever - in kinda mangled half-circle-ish loops.

and thats more-or-less whats happening inside the xfmr - the core doesnt pull in ALL of the winding flux, just most of it.

Anywhere within the winding volume that isnt either core or winding will be filled with air-cored flux. any of those air-cored flux volumes that are not shared with both (all) windings will be leakage flux. remember the flux falls off with distance, and so does the "sharing" of air-cored flux volumes.

Thats why bifilar winding gives the lowest leakage - the wires all share the same physical space, and all air-cored flux volumes too.

if you have two windings, one atop the other - the outside of the outer winding has flux from that winding which is not shared by the other winding. if you surround the outer winding with ferrite (eg pot core) that flux gets coupled into the core, which reduces leakage. otherwise (eg ETD core) it doesnt - and thats why non-interleaved windings have high leakage inductance.

and for a really lousy winding technique, look at many mains transformers - Primary and Secondary are often side-by-side on the core. this makes electrical isolation a breeze, but maximises leakage inductance as the two windings share far less volume - they are as far apart as it is possible to be.

you can also control the leakage inductance by deliberately spacing windings apart.

and lastly if you look in a decent (old) transformer/magnetics text (eg magnetic circuits and transformers, MIT) they show in detail how to calculate leakage inductance, and its all about calculating the not-core-or-winding volume.

HTH

so to answer your question: in order for there to be no leakage flux (perfect coupling) all you need to do is have infinitely thin conductors all occupying the exact same volume. luckily near enough is good enough - 44AWG is a nightmare, imagine how annoying infinitely thin wire would be.

and this would also require the shield to be infinitely thin - at which point its eddy current losses would indeed be zero.

Alas wires, insulation, shields, bobbins etc all conspire to fill up our transformers with space that isnt conductor, and prevent windings from occupying the same space. so there is always leakage flux.

Last edited by Terry Given; 30th September 2012 at 10:04 PM. Reason: forgot to answer question
  Reply With Quote
Old 1st October 2012, 02:28 AM   #14
diyAudio Member
 
Join Date: Nov 2010
Location: The City, SanFrancisco
Terry

You are one good explainer (love your reason for editing).

Thanks your explanation really tied it all together for me.

For whats it worth, many years ago I designed HV (and LV) low power/noise supplies. One required 10KV p-s isolation. Started with a C-I core (I think thats what there called nowadays) with p and s on opposite sides, what a nightmare. Ended up that way too, but with a pair of coupling windings against the core added.

Similarly used aluminized mylar strips (or just copper tape) for the LV torroid core shields. Lots of fun. Never went smaller than #38.

Many Thanks
-Antonio
  Reply With Quote
Old 1st October 2012, 04:40 AM   #15
diyAudio Member
 
Join Date: Aug 2010
Quote:
Originally Posted by magnoman View Post
Terry

Started with a C-I core (I think thats what there called nowadays) with p and s on opposite sides, what a nightmare. Ended up that way too, but with a pair of coupling windings against the core

Many Thanks
-Antonio
What are coupling winding’s.
  Reply With Quote
Old 1st October 2012, 12:39 PM   #16
diyAudio Member
 
Join Date: Nov 2010
Location: The City, SanFrancisco
PowerBob,

I put a single layer winding against the core underneath primary as well as one under the secondary. These two additional windings were connected to each other. The effect was to reduce the leakage inductance of both p and s significantly, as you can think of it as two transformers (note the p and s windings were physically spaced away from the cores to account for the HV isolation). Came up with it in desperation so I dont know if it has a real electronic term but I think I've since seen something similar.

Thanks
-Antonio

Last edited by magnoman; 1st October 2012 at 12:48 PM.
  Reply With Quote
Old 1st October 2012, 01:57 PM   #17
diyAudio Member
 
Join Date: Mar 2008
Antonio,
I thought thats what you meant. nicely done coming up with that solution!

I've seen them referred to as "flux-cancelling windings" in a few papers on High-Voltage pulse transformers. when the windings are concentric (they occupy a single core leg, eg center leg of E core) one is placed on the inside (closest to core) and one on the outside (farthest from core).

the two windings are connected in parallel. If the flux cutting each winding is identical then the induced voltages of each winding are identical (V = N*dFlux/dt), and no current flows. An electrical model is a pair of voltage sources with their returns connected together, and their outputs connected together through a pair of series inductors (one for each winding).

when one winding sees more flux than the other, the resulting change in induced voltage causes current to flow between the two windings. This current compensates for the flux imbalance....damn, I cant remember exactly how! I'll have to look it up.

but work it does - I've used it to good effect, but I didn't think it up - I pinched it from an old book.
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
Shielding ..... east electronics Solid State 9 20th July 2009 07:34 AM
shielding a E-I transformer rs1026 Chip Amps 6 10th September 2004 12:39 PM
Shielding peterr Everything Else 2 17th September 2002 11:57 AM
Shielding a transformer SteveG Parts 5 26th July 2002 04:21 PM
Transformer shielding shadow Solid State 11 16th May 2002 10:26 PM


New To Site? Need Help?

All times are GMT. The time now is 09:23 PM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright ©1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2