I was looking to add some shielding around a power transformer. So a bit of research and sure enough, what I want is a can made of mu-metal. It is about 5000 times as effective as plain steel. Well, you can't just get a sheet and form a can as it has to be annealed after forming in a pure hydrogen environment. So, I guess that reduces me to electrical steel, also known as the laminating steel used for transformer cores as it is a good 400 times better than plain steel.
Has anyone had any success in adding shielding around transformers? Anyone know where to get electrical steel sheets or boxes made of it?
Has anyone had any success in adding shielding around transformers? Anyone know where to get electrical steel sheets or boxes made of it?
low carbon steel is so cheap, available that you just use it, may take more, a couple of air -spaced layers but its not worth getting special material unless you know what you're doing, have insane space, weight constraints - and it usually means knocking down the level with cheap steel, air space 1st to avoid saturating the mu metal anyway
My experience is that thin mu-metal made into a box around a transformer is not terribly effective and not as good as heavier plain steel. Data sheets can be misleading. You'll get more field reduction for your money by reorienting the transformer; don't feel it has to be mounted parallel to the chassis walls. Distance is also a better friend than fancy metals. Depending on your construction, and this may be hard to believe, a heavy aluminum plate, about 5/8" minimum, will kill fields very effectively by dissipating them as heat from eddy currents. Strange but true.
Yes, eddy currents cause an opposing field provided there is a place for the currents to flow, i.e shorted winding from copper around the windings. Aluminum is considered in the literature as useless as a magnetic shield for LF as it has the same permeability as air. I do have some half inch plate so I may play a little. For power transformer shielding, we are talking about less than 1Khz.
I guess mild steel it is. Most of my issues are to retrofit existing problems where I have very little leeway on position, or it is impracticable to remote locate the transformer.
Turns out, mu-metal is not suitable for a DIY. Although it has a permeability of 50,000 vs 100, if you bend it at all, it is then just as poor as soft steel. It has to be annealed as it's magic is dependent on the grain structure. I was hoping to find a source of electric laminating steel in less than hundred ton lots.
I guess mild steel it is. Most of my issues are to retrofit existing problems where I have very little leeway on position, or it is impracticable to remote locate the transformer.
Turns out, mu-metal is not suitable for a DIY. Although it has a permeability of 50,000 vs 100, if you bend it at all, it is then just as poor as soft steel. It has to be annealed as it's magic is dependent on the grain structure. I was hoping to find a source of electric laminating steel in less than hundred ton lots.
I've mentioned the aluminum trick before and had it dismissed, but don't be too quick. Long ago I was working with amplifiers for atomic force microscopes where the slightest hum will destroy the image and feedback stability. It took reorienting the power transformers at an angle, putting the low level circuitry in the lowest field direction, and then putting in the 5/8" aluminum plate between the transformers and amplfier circuitry. No box, just a flat plate. The effect requires very thick plate; I seem to remember that 1/2" isn't effective. Some alloys are better than others, but I think plain old 6061 is fine. The plate brought the residual 60 Hz field down to a negligible level. All this assumes you've already done everything else correctly regarding star ground and such.
you can search "skin depth", a calculator:
Ness Engineering Tech Data - Skin Depth and Skin Effect
gives 1/2" for 6061 @60 Hz - and one skin depth is only ~40% attenuation
Ness Engineering Tech Data - Skin Depth and Skin Effect
gives 1/2" for 6061 @60 Hz - and one skin depth is only ~40% attenuation
I thought with mu-Shield you can bend the material, but you can't kink it. Supposedly, you can also drill mu-Shield with little ill effect.
Will try the "Aluminum trick" -- where had you posted it previously?
One of the Linear Tech application notes had a nice graphic of the emf profiles of different transformer configurations, but I can't find where I put the dang'd thing.
Will try the "Aluminum trick" -- where had you posted it previously?
One of the Linear Tech application notes had a nice graphic of the emf profiles of different transformer configurations, but I can't find where I put the dang'd thing.
Last edited:
Try a search on mumetal and mushield. I've bought small quantities in the past with no problem. It's not cheap. If you find a junk oscilloscope, the CRT shields are usually some form of mumetal. I know I posted the aluminum trick quite a few years ago, but everybody knows aluminum and copper can't shield magnetic fields! Fun- drop a strong neo magnet down a piece of copper water pipe.
Mu metal is not particularly suitable for shielding power transformers as apart from the difficulty of working it without subsequent heat treatment it tends to saturate and then is ineffective. It is very good for shielding low level fields which is why it is used for signal transformers and tape head enclosures. Stalloy or similar alloys are deigned for the higher flux levels found around power transformers and are also easier to work.
The thickness of standard 1-ounce/foot_square copper foil,
at 35 microns or 1.4 mils, provides 1/E 1/2.178 attenuation
at about 5MHz, with progressively better attenuation for
higher frequencies per theory. Each 1/E is a neper, or 8.9dB,
thus 500MHz, 100x faster, would have 10 skin depths in that
same 1.4mil copper. Using a copper PLANE, to allow circulating
currents, provides 10 * 8.9 or 89dB for the high-speed-edge
energy of microprocessor clocks.
A 5MHz MPU clock would only be down 8.9dB, but assuming
1nanosec edges, the transient energy experiences wonderful
attenuation. This should permit integrating sensitive analog
circuits near modern MPUs, IF A PLANE (Vdd or GND or just SHIELD)
is between analog and MPU.
at 35 microns or 1.4 mils, provides 1/E 1/2.178 attenuation
at about 5MHz, with progressively better attenuation for
higher frequencies per theory. Each 1/E is a neper, or 8.9dB,
thus 500MHz, 100x faster, would have 10 skin depths in that
same 1.4mil copper. Using a copper PLANE, to allow circulating
currents, provides 10 * 8.9 or 89dB for the high-speed-edge
energy of microprocessor clocks.
A 5MHz MPU clock would only be down 8.9dB, but assuming
1nanosec edges, the transient energy experiences wonderful
attenuation. This should permit integrating sensitive analog
circuits near modern MPUs, IF A PLANE (Vdd or GND or just SHIELD)
is between analog and MPU.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.