[Amps] Choke anomoly

[Manfred Mornhinweg]

Jim,

> I measured the DC resistance, using my  fluke 87..and it measured
> exactly 10.0  ohms.    Then re-measured using my  B+K  875-A... and
> it spits out 8.7 ohms.

Such a big choke can easily fool a multimeter, due to the high
inductance and low resistance. If the multimeter uses anything other
than a constant, regulated DC for measuring resistance, the inductance
will affect the reading and make it useless.

Some multimeters, and often the better ones, use pulsed DC when
measuring resistance, to conserve battery power. These won't work right
with such a choke.

> Then I noticed the   4 x 5 x 12  small lab supply leaned over... and
> STUCK to the  left side of the  choke! The dahl choke uses the same
> pair of  hypersil C cores as their  10 kva  CCS plate xfmrs,
> identical, except the choke, of course has just a single winding.   I
> had a 8 inch long, chrome plated letter opener close by, and it stuck
> to the left side of the choke, like a magnet.     Heres the kicker,
> the same metal letter opener, would NOT stick at all when placed on
> the RIGHT  side of choke.   Choke is  13 inches  wide  x  8 inches
> deep,  x  11 inches tall.

Where magnetic flux leaks out of the core, ferromagnetic objects will be 
attracted. Instead where the flux stays well confined in the core, 
attraction of external objects will be very low, maybe unnoticeable. So 
either your choke has its air gap made on that side where it attracts 
stuff, or due to some asymmetry in the assembly it has an unintentional 
air gap or magnetic bottleneck on that side.

> I can see where they have done the cut on the cores, then
> polished, the with winding in place, the tops of each core are put
> back on, aligned, then the  pair of cinching straps used to mash each
> top of each core to the mating lower cores.

I understand this is a double-U core made from wound tape. So it has 
just two magnetic joints, rather than the three of a doble E or an E-I 
core. Either one or both of those magnetic joints must have a gap in 
between. ALthough called airgap, it's normally filled with some 
non-magnetic spacer, rather than air. Try to see if your core is gapped 
just on one side. If so, then of course that side will attract objects, 
and the other side will not, or at least much weaker.

If instead the gap is distributed on both sides, then maybe it's 
asymmeric - intentionally or not.

Consider the possibility of the tow U pieces being symmetric, assembled 
with spacers between, but one of the spacers being steel and the other 
being non-magnetic ("air" gap). In that case of course the side with the 
non-magnetic spacer will attract objects.

All normal, constant-value chokes using iron lamination cores will have 
an airgap (or two), while transformers will not. Instead swinging chokes 
may use no airgap at all, or only a tiny one. If yours is a swinging 
choke, the asymmetry might be due purely to manufacturing tolerance, the 
two pieces meeting better on one side than on the other.

> Then I turned down the current limit on the  lab supply, and all hell
> breaks loose on the  digital display voltage and current on the the
> lab supply.    I quickly disconnect the pair of test leads from the
> lab supply, and get a small spark.   Then with nothing connected to
> supply,  3 secs later, smoke billows out of the lab supply, so yanked
> the plug out of the 120 vac outlet.   My guess is the back EMF
> cooked the supply,   should have had a rvs connected diode, and or a
> MOV  across the choke, and  or output of lab supply.

I have designed some variable voltage, variable current power supplies, 
and I can tell you that it's mighty hard to get them to be stable when 
working into highly reactive loads! Using current-limited operation into 
a choke is the nastiest condition possible: The choke tries to hold the 
current constant, while the power supplies tries to force  certain 
specific current. The two clash head-to-head! The control loop gain 
soars to almost infinity, together with a full 90 degree phase shift in 
the load, that adds to the 90 degrees phase shift of the supply's power 
stage, making a nice 180 deree phase shift and thus a formidable power 
oscillator. The power supply tries to force a square wave current 
through the choke, which reacts with voltage spikes that can easily 
reach kilovolts, at the full current that's flowing, thus putting out 
pulses that can exceed 10 kVA, and your poor little power supply had no 
chance to think what to do...

RIP, poor little power supply... :-)

> So
> how come the magnetism on left side and nothing on right side ? Is
> this normal, or is choke defective ?

It's either due to intentional asymmetry (air gap on one side), or 
unintentional asymmetry (residual airgap on one side). I see no reason 
for worrying, the choke should still work fine.

> Using the PSUD software, the  C-L-C  config has merit, for reducing
> ripple down to extreme low values.   Interestingly enough, a C-R-C,
> using a simulated  8.7 ohm  resistor, also works very good, not as
> good as a choke,  but extremely effective.   The  C-R-C config would
> reduce ripple by a factor of  10.  And thats with the same dcV drop
> across the 8.7 ohm resistor as the  8.7 ohms across the choke.   The
> C-L-C config would provide for a ripple reduction with a factor of
> 550.

Nice - but a modern switching power supply would reduce the ripple much 
more, give you a highly regulated voltage even while the line voltage 
varies by 30% or more, and as a bonus give you almost perfect power 
factor - all while being about 50 times lighter.

And of course, about 200 times more complex, too... ;-)  Nothing comes 
free in this life...

Manfred

========================
Visit my hobby homepage!
http://ludens.cl
========================