That first (furst?) Wurth part is interesting. They wind the wire on some sort of rod, then stick it sideways into the square ferrite box. It is sort of shielded.
At rated current, 17 amps, the L is down about 40%, which means the shielding won't be so good there. And the temperature is up 50C!
John
Speaking of measurement...
I switched my 4192A to show the series resistance part of the impedance and did some more measurements.
SRU1038 22uH, using no-leads (leaded component test fixture).
no plane Groundplane f L R L R
10k 20.41 0.082 20.30 0.091 50k 20.30 0.151 19.91 0.231 100k 20.19 0.319 19.67 0.439 200k 19.99 0.79 19.41 0.930 500k 19.52 2.59 18.93 2.73 1M 19.10 5.36 18.50 5.56 2M 18.90 9.80 18.29 10.10Not sure what to make of these TBH
( snipped-for-privacy@yahoo.com)
( snipped-for-privacy@yahoo.com)
from the tw=
parallel to=
The cros=
material whic=
well, du=
field
check the=
having t=
crew mi=
isolati=
Funnily enough I was going to try that, I have a spectrum analyser with tracking generator. Switching frequency is 370 kHz - Anyone care to lay bets what the insertion loss will be? :)
cast=
geometry
Or not.
Watch this space...
two
to
ross
hich
due
these
g to
might
7ation
st
8try
ot.
ISTM the furst is the wurst. Most of the flux goes thru the rod and up the box thru the top, but potentially a great deal goes down the box walls to the bottom rim of the box.
^ >>>>>>>> v ^ v ^ oooooo v
+
two
to
ross
hich
due
these
g to
might
57lation
st
88etry
The apparent resistance is increased by the following factors:
50K 1.52 100K 1.38 200K 1.17 500K 1.05I expect those are at least in part real increases in dissipation caused by stray flux inducing circulating currents in the ground plane.
Both inductor and ground plane e.s.r.s will show skin effect at higher frequencies. I suspect we're seeing that too, and that the inductor's added e.s.r. dominates over ground plane losses at higher frequencies (beca use the inductor winding path length is sooo much longer than the ground plane' s circulating current's path length).
Cheers, James
OK, this is how I set up the test, two of the parts facing each other across a piece of (mostly) bare FR4:
(the holes on 0.1" pitch are not plated through)
SMB connector to each one, I mirror-imaged everything to keep the E- field coupling down as best I could. One part connected to the spectrum analyser input, the other to the TG output, set to 0dBm.
Results:
100 kHz - 1 MHz :Not much to see at low frequencies, but a nasty big peak at 40ish MHz, which corresponds to the SRF of the part. We have a big, broad emissions peak at 65 MHz, I suppose the presence of the ground plane in the real product could be shifting the resonance.
Thanks everyone for the suggestions for alternative parts, I will try some of them and post back here in a week or two.
On a sunny day (Fri, 13 Nov 2015 11:09:04 -0000) it happened RBlack wrote in :
Nice pictures.
The basic solution to prevent magnetic coupling is: 'put the coils at right angles to each other'. that would require a small extra rising PCB in this case. But the second rule: 'as far away from each other as possible'. So why put these on opposite sides of the board? As you mention adding a ground-plane it seems you are making a new layout. In that case move the coils further away from each other.
Yes, the parts were deliberately placed in the 'poorest' configuration. This was a test to see how magnetically 'leaky' this particular inductor design is. The answer seems to be 'very' - I will repeat the test on the replacement inductors proposed upthread, once I have samples of them.
On a sunny day (Fri, 13 Nov 2015 14:36:59 -0000) it happened RBlack wrote in :
Ha, my misunderstanding Yes coupling coils this way makes a good rotary transformer, say 2 potcore halves...
Have something to add? Share your thoughts — no account required.
Ask the community — no account required