Connectorized feed-though filters?

[Odysseus]

I'm familiar with the use of feed-through caps/filters on RF equipment to eliminate leakage over the incoming DC power, but everything I've seen uses soldered wire terminations. What options are there for feed-though filters which are connectorized?  The best I've found is really intended for mains filtering and quite bulky.

[radar_macgyver]

How about a filtered D-sub? There are several kinds available. The simplest just use a puck of ferrite with holes drilled for each pin. More sophisticated types implement a C or pi filtering network on each pin. The pi network style tends to be very pricey.

[T3sl4co1l]

Hmm, what kind of connectors were you hoping for?

Form factor?  Attenuation?  Impedance, frequency range?  Price?

The filtered D-sub for example, it's filtered as such, yes -- but the roll-off is very gentle, a few dB to low 10s dB depending on circuit impedance.  Maybe that's enough, maybe not.

Tim

[radar_macgyver]

but the roll-off is very gentle

Similar to what you'd expect from one of these, no? (I was assuming these were what the OP meant)



The feedthroughs I've used range from 100pF to 10nF, I've seen similar values available for filtered D-subs which I've also used for one project.

[T3sl4co1l]

No, worse than that, assuming typical conditions (50 ohm source and load).  Those ferrite plates likely offer ~100 ohms at 100MHz, or about 6dB, and not much more (mostly less) at any other frequencies.  A 1nF cap rolls off at 6.3MHz (-3dB) and a feedthru worth its trouble should perform asymptotically above there (-20dB/dec).

But it depends.  If there are caps either side of the ferrite bead, the local impedance can be quite low, giving acceptable filtering.

Tim

[cdev]

Fair-rite makes cores that have six small holes through them.

How about a filtered D-sub? There are several kinds available. The simplest just use a puck of ferrite with holes drilled for each pin. More sophisticated types implement a C or pi filtering network on each pin. The pi network style tends to be very pricey.

[coppercone2]

well you can make an adapter that goes over it.

[radar_macgyver]

A 1nF cap rolls off at 6.3MHz (-3dB) and a feedthru worth its trouble should perform asymptotically above there (-20dB/dec).

But it depends.  If there are caps either side of the ferrite bead, the local impedance can be quite low, giving acceptable filtering.

Tim

A little confused. This connector for example has a 1nF cap, so how would it perform worse than an equivalent feedthrough? Unless you were referring to one of these which has just the ferrite puck.

[T3sl4co1l]

Yes, the ferrite plate type.  The D-sub with caps wouldn't be much worse than a proper (coaxial?) feedthrough (or so I would hope).

Sadly, neither one gives an impedance or s12 plot.

Tim

[Odysseus]

Thanks all! Filtered D-subs actually sound like a very good solution. I will have to keep this in mind on my next project. Yes, I'd consider ferrite only filters to be insufficient.
I realized that you could use just about any coaxial filter, but that seems like a painful mix-up waiting to happen.
If there are any other filter/cap form factors you are aware of, I'd be grateful to know about them.

[German_EE]

Why not use the wire-ended feedthroughs and add sockets on the wires, something like the female connectors on an RS-232 socket?

[coppercone2]

for dsub there is a x2y capacitor bank you can use IIRC.