# EMI CHOKE QUESTION

Discussion in 'Electronic Design' started by Yzordderrex, May 19, 2005.

1. ### YzordderrexGuest

I have recently been given responsibility for EMI choke design. The
company I work for makes motor controls.

I am looking at a 12amp servo drive (single phase 230v input) that will
pull about 150 amps peak on the input side when balls to the walls 300%
output.

The EMI filter choke is wound like any other single phase unit you
would see on a toroid. Winding A on one side, and winding B on the
other side of the ring. Each winding taking up about 160 degrees of
ring so that there is some voltage creepage distance.

I have wound a 10 turn tertiary winding on the core that I use to
measure inductance. I jam my 30v bench supply into the tertiary
winding an watch the rate of rise of current to measure the inductance.

When I excite the core with 150 amps not all of the flux is cancelled
so that the inductance goes to nearly zero. I'm assuming that the
filter is probably useless at the peaks of the line currents.

Would I be better off using an insulated wire and wind each coil say
330 degrees and interleave the coils - bifilar?

We have gear to measure and I will probably proceed on this path for

regards,
Bob

2. ### Paul MathewsGuest

Yes. And, for extreme cases, you can consider using coaxial cable
Paul Mathews

3. ### YzordderrexGuest

Ok, thanks Paul.

I just finished winding a choke bifilar and the inductance stayed
constant up to 50amps.

Unfortunately my supply only goes to 50amps. I would like to get
flavor at 150amps.

I may look at a Rubadue style wire to wind units.

regards,
Bob

4. ### colinGuest

dont forget that although the flux on each side of the toroid cancel
eachother out so there is no circular flow around the core, they do however
add up to cuase a net flux across the core, wich if high enough wil cuase
enough flow through the air to saturate the whole core.

Colin =^.^=

5. ### Paul MathewsGuest

Good point. Cross-winding midway around takes care of that problem.
Paul Mathews

6. ### YzordderrexGuest

Colin, Paul,

Seems you two have lost me. Not sure what a net flux across the core
is, and cross-winding is a term I'm not so familiar with.

One thing I did realize is that winding bifilar will tend to decrease
the differential mode inductance - So differential noise components
might then actually be worse.

regards,
bob

7. ### YzordderrexGuest

Colin, Paul,

Seems you two have lost me. Not sure what a net flux across the core
is, and cross-winding is a term I'm not so familiar with.

One thing I did realize is that winding bifilar will tend to decrease
the differential mode inductance - So differential noise components
might then actually be worse.

regards,
bob

8. ### YzordderrexGuest

Colin, Paul,

Seems you two have lost me. Not sure what a net flux across the core
is, and cross-winding is a term I'm not so familiar with.

One thing I did realize is that winding bifilar will tend to decrease
the differential mode inductance - So differential noise components
might then actually be worse.

regards,
bob

9. ### YzordderrexGuest

Colin, Paul,

Seems you two have lost me. Not sure what a net flux across the core
is, and cross-winding is a term I'm not so familiar with.

One thing I did realize is that winding bifilar will tend to decrease
the differential mode inductance - So differential noise components
might then actually be worse.

regards,
bob

10. ### Paul MathewsGuest

The differential mode component you mention is just leakage inductance
between the common mode windings. Most of this flux is 'in the air',
rather than 'in the core'. If you absolutely must minimize the number
of components in your design, then it may be worthwhile to depend on
leakage inductance for DM benefits. However, on the down side, winding
methods that increase leakage inductance also increase the likelihood
of imbalances that lead to core saturation. At the point when the core
saturates, the CM properties of the component pretty much disappear.
When you place windings around a core, proceeding uniformly in the same
direction, there is a general flow of current in the direction of those
windings. For a toroid, this has the unintended effect of one or more
'turns' effectively following the core and producing a field through
the core axis rather than within the core material. You can cancel
this effect by alternating the direction that the turns advance around
the core. 'Cross-winding' of single-layer toroids involves winding ~
half way around the core, 'jumping across the core center to the
opposite side, and winding an equal number of turns advancing in the
opposite direction around the core. All turns have the same direction
with respect to the core itself. Some mfgrs of toroidal inductors
always follow this practice. For a horizontally mounted toroid, it's
also quite easy to specify 2 windings wound in such a way that they can
be series connected to achieve the cross-winding effect. If done
properly, cross-winding also reduces the distributed capacitance of the
inductor, making it effective at higher frequencies.
Paul Mathews

11. ### Terry GivenGuest

dont use a supply, use a monster cap. Charge the cap up, and "splat" it
across the choke. Easy to get 1,000A that way

just choose Ecap >> Echoke at Idesired

Cheers
Terry

12. ### JeffMGuest

[Yzordderrex reposted his response]

The new beta of Google Groups has a much lower latency than the old
iteration.

If you get a Server Error message,
wait a minute or 2 and refresh the thread page.
Usually your post has gone thru,
it's just that the ACK page hasn't come back.

BTDTGTTS

13. ### MarkGuest

does your system need common mode or differential filtering (or both)
to solve the EMI problem?

Mark