Smps Toroids on a Ground Plane

I recall somewhere an author recommended that the inductors be placed on a ground plane..

That's probably nice for little inductors..Like buttons. But my toroid is 1/2" tall and about 1.5" wide..

I don't believe putting this toroid flat down on a ground plane is going to help reduce EMI emissions. Correct? All the copper windings are too far from the ground plane..

Would it help to Faraday shield the toroid to reduce EMI... ? (I'm trying to avoid killing my neighbors AM radio reception.) Could I blanket my toroid with some foil and ground the foil? Or don't bother...?

Toroid Conditions f=100khz I=2A average, 200mA ripple V=170V peak, 0.4 duty square wave (hard switching)

This toroid is in a earth grounded metal box. The smps power ground is not connect to earth ground. The toroid is part of an offline unisolated smps..If the two grounds connect.. .Poof!!! So ..I'm guessing a faraday shield will be connected to power ground. That would make the shield "hot" relative to earth ground. Shocking to touch when live but safe in a closed earth grounded box. D from BC

Reply to
D from BC
Loading thread data ...

In my experience toroids don't emit much anyway; their flux is contained in the core. Have you any reason to think otherwise? I would pay more attention to the PCB layout, e.g. minimising the area of high dI/dT current loops. Also make sure input and output wires are filtered.

--

John Devereux
Reply to
John Devereux

Yes..I know toroids are good at keeping in a magnetic fields but there's an electric field too .. When high switching voltages exist across a toroid..can the electric field be a problem...? I don't know much about electric field interference. D from BC

Reply to
D from BC

Yes, indeed. Most inductors in switching circuits get connected to a high dv/dt node at one end and a lower dv/dt node at the other. Placement of the high dv/dt pin(s) should be away from potential EMI and cross-talk victims. Many inductors can also be placed in either of

2 orientations on the PCB (due to pin symmetry), and sometimes it makes a big difference which placement you happen to use. This has to do with the way that the 1st turn on one end comes from inside the core and vice versa, which means that the highest dv/dt locus moves a bit. There is also a extra 'turn' on most toroids that generates field that is not in the core at all. This is the turn that the current takes as the coil turns progress around the core. We'll probably have Mr. Sloman chime in on this fine point. A few standard toroid inductors are wound with a crossing turn halfway around, so that this effect is mostly cancelled. You can also have inductors wound with 2 windings and accomplish the crossover turn on your PCB. In any case, the ground place can help by concentrating the E field in a region around the high dv/dt turns, thereby reducing parasitic currents to other structures. Of course, this can also 'inject' noise into your ground system...push down here and it pops up there. Paul Mathews Paul Mathews
Reply to
Paul Mathews

Good colloquial "push down here and it pops up there" :)

Crossing turn halfway around?? I don't understand yet.. Maybe after some more coffee.... :)

Still thinking... D from BC

Reply to
D from BC

Probably Whidbey Island speak for "twisted windings". Bifilar, trifilar and so on.

--
Regards, Joerg

http://www.analogconsultants.com
Reply to
Joerg

Probably not. "Non-progressive" windings, like the Ayrton-Perry "bootlace" winding technique, can be applied to single wire windings or to bifilar, trifilar and rope windings.

Check out "Coaxial AC Bridges" by B P Kibble and G H Rayner, ISBN

0-85274-389-0, which includes a lot of stuff on winding fancy inductors and transformers for high precision work. The stuff about non-progressive windings is in section 4.2.

formatting link

-- Bill Sloman, Nijmegen

Reply to
bill.sloman

Ah, the secrets of RF engineering. Thanks for explaining. That seems to be one of those books from the days when engineers were self-taught and knew their stuff by heart.

When I was a kid an old engineer showed me how to build inductive wideband distributors and combiners. Whenever I asked him why this or that had to be exactly as shown his answer was: "It just don't work no other way".

--
Regards, Joerg

http://www.analogconsultants.com
Reply to
Joerg

Imagine a toroid with 10 turns. Beginning on the inside of the core, wind 5 turns, covering about 160 degrees around the core and ending outside the core. Then, cross over the the core, that is, route the wire on a diameter right over the core central axis to a point near the first turn (about 10 degrees to the bare side of the 1st turn) and continue the remaining 5 turns. You'll end up with a crossing turn that goes over the top of the entire core, with 5 turns progressing CW around the core and 5 turns progressing CCW (or ACW for some folks). There are many variations on this approach that can be used to accomodate multiple windings, minimize parasitic capacitance, etc. A few suppliers of power toroids do this as standard practice. Paul Mathews

Reply to
Paul Mathews

Photos here:

formatting link
and here:
formatting link
look particularly at Figure 2

Paul Mathews yes, on Whidbey Island

Reply to
Paul Mathews

You have the choice whether to mount a toroid horizontally or vertically and how to orient the high dv/dt vs low dv/dt ends. The geometry scales: At an equal number of core diameters, the ground plane has equally beneficial effects, regardless of core size.

High dv/dt end will generate parasitic currents according to the relation: I = C dv/dt. These currents WILL flow, regardless, so minimize C and dv/dt and provide the shortest possible intentional paths for such currents. Shields are meant to perform this function, but you must have a low impedance connection to the common node of the source of the dv/dt, otherwise the shield itself will have significant dv/dt. The highest dv/dt is often associated with switching elements and associated ringing, so rectifier technology, gate drive, and snubbers are all very important. Shields with large dimensions relative to the inductor (if you have the room for them) will have a low capacitance and corresponding low parasitic currents. However, the larger dimensions then present a challenge to make a sufficiently low Z connection to the common point. Faraday shields are supposed to completely enclose a field generator and theoretically neutralize magnetic and electrostatic fields. A properly wound toroid has very little external field, so Faraday shields are seldom seen except where very high isolation from magnetic fields is required.

Positioning a shield very close to a switchmode toroid is an excellent way to generate very high parasitic currents to the shield. Unless the shield is very well connected to a common return point for the generated currents, the shield itself will have high dv/dt and radiate to other structures. See earlier comments.

Safety capacitors can be used to direct parasitic currents through particular paths, returning them to their sources without much affecting leakage between mains and other conductively accessible parts. Of course, this is why there are limits on how much capacitance you can use for the purpose. Advanced safety topic: study carefully.

See some interleaved comments above: Paul Mathews

Reply to
Paul Mathews

Thanks, Paul. In noise critical apps toroid winding can become an art.

The closest I came was while fishing with a former boss of mine. Didn't catch anything, we should have motored on to Langley for a nice breakfast and a stroll.

Some of my co-workers at ATL in Bothell were from Whidbey Island. Quite a commute every day with the ferry ride and all. Many had an old car on the other side (Mukilteo?) and a bicycle on Whidbey Island. Whenever they talked about the island it sounded like they lived in paradise.

--
Regards, Joerg

http://www.analogconsultants.com
Reply to
Joerg

ok..I got the structure now.. To help me understand what's going on I'm going to try this trick:

I'm going to pretend the wound toroid is a power reostat. Let's say I put that in my smps..(but it still acted like an inductor) In my app..one end of the inductor is 170VDC and the other has 300Vpk

100Khz square wave. Moving the wiper along the core (normally wound) and the wiper has an increasing square wave amplitude. I'm imagining the electric field is like this too.

Now I switch to that special winding technique.. Turning the reostat from the beginning and the waveform amplitude starts off small...increases....jumps to max...then decreases to half.

But I'm still a little fuzzy on how this minimizes electric field interference with neighboring components... ....I'm gonna have to have some more coffee :)

D from BC

Reply to
D from BC

Magnetic field effect: The parasitic turn from normal toroid winding around the circumference in one direction produces field outside of the core. The effect of the field depends on what loops might couple to it, and ordinary shielding is ineffective at preventing such coupling. You cancel this field with the crossover turn approach.

Electric field effects: The usual winding approach brings opposite ends of the coil near to one another. High dv/dt of one end relative to the other, along with their close proximity, means current flow through a relatively high parasitic capacitance. As you suggest, dv/dt varies continuously around the turns. Using the crossover-turn winding technique, you end up with the low dv/dt end close to the middle of the coil rather than the end, where it is adjacent to half the dv/dt. It takes a complex model to approximate the overall effect, since you'd need to consider all of the turns and not just the ends. However, by simply measuring the self-resonant frequency, it's easy to demonstrate that the crossover-turn technique can raise resonant frequency by a factor of 4 or so. At the higher resulting ring frequency, you can use much lower snubbing capacitance, as an example of one benefit. These used to be considered 'RF' techniques, but switchmode power supplies have entered that realm.

Side note: Best read for offline power switchmode is anything by Sanjayit Maniktala. Paul Mathews

Reply to
Paul Mathews

In case someone uses a search engine the spelling of his name would be Sanjaya Maniktala.

Also very good reading are the older Unitrode app notes, now TI and hopefully still on their server. If you have the "Unitrode IC Data Handbook" from around 1990 don't ever think about tossing it. Tons of valuable SMPS info in there.

--
Regards, Joerg

http://www.analogconsultants.com
Reply to
Joerg

Totally agree Paul, his "Switching Power Supply Design + Optimization" (ISBN 0-07-143483-6) has a ton of info. Sorry to see him voted off Am. Idol. Harry

Reply to
Harry Dellamano
[snip]

Resonance increased by a factor of 4 or more!!! Wow :)

Thanks for the book suggestion. I'll check out some online bookshops.. D from BC

Reply to
D from BC

The general technique of designing windings so that high dv/dt is maximally far away from low dv/dt can be applied in many different ways on various kinds of cores. 'Progressive' winding is a fairly well known example of this, but there are many other less well known methods, and you can easily dream up your own. Figuring out a way to make a technique mass producible is the real trick. In my experience, the majority of magnetics suppliers can't offer much creativity, and many don't really know much beyond the basics of how to use their winding machines. Paul Mathews

Reply to
Paul Mathews
[....]

It is sometimes easier to design in two inductors in series than to get one made with special winding methods. On things like EFD formers, you can get the ones with multiple winding bays. You can wire the resulting sections in series on the PCB.

You can also specify the winding to be done with "wire rope". Twisting 7 strands together isn't quite Litz wire but the high frequency Q is a lot better. Once a winder understands the concept of how the rope is done, they don't seem to have any problem with doing it for you. (At least, the winder I use)

Also, if you are going after every % you can get, the snubber can be made so that some of the energy ends up back in the input supply.

Reply to
MooseFET

Sorry to have come across this thread late, but what is the "crossover turn" technique?

I've been trying two methods to minimize susceptibility to axial fields in Rogowski sensors and tranformers (toroids with non-magnetic cores): 1) start with a circumferential turn in a groove machined into the former, then wind on a single layer winding in the opposite direction, or 2) wind two layers, reversing the feed for the second layer.

Both seem to be OK at reducing axial flux sensitivity, but still having trouble getting sufficient winding uniformity (another subject).

But in general, I'd like to know about anything that could raise self-resonant frequency 2 octaves.

Regards, Tony

Reply to
Tony

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.