Block of Aluminum RF Black magic

[johnh]

picked up this  block of Rf black magic

N connector either end

Weighs 4kg  135mm * 75mm * 263mm

Can anybody tell me what it was used for?

[n45048]

Some kind of funky waveguide/air gap/lighting arrester?

[GEuser]

It's tuneable (see the screws that go in the posts) , a band width of 10MHZ or so , my guess a Cavity filter ?

[Gribo]

Yes, looks like a cavity filter. But what are those copper things inside?

[VK5RC]

Aren't they sampling probes to transfer the signal from one cavity to the next?
Cavity filters can be a lot sharper than 10MHz, some Ham VHF repeaters use them on the antenna which is used as both transmitting and receiving antenna at the same time, but the frequencies are only 0.6Mhz apart. You would need somewhere in the order of 160dB isolation at 0.6MHz.
eg http://forums.qrz.com/showthread.php?241866-Phelps-Dodge-VHF-Repeater-Duplexer-on-eBay

[KJDS]

That is a coupled coaxial cavity filter.

It's is a seven section quasi-elliptic with three finite frequency transmission zeroes, two on the low side and one on the high side and I'll guess that it's an 850 or 900MHz cellular basestation transmit band filter. It's almost certainly the full cellular bandwidth and I'll guess less than 1dB insertion loss through the passband and more than 60dB of attenuation through the receive band, which will be rather close to the passband.

The extra bits of copper in the middle are what creates the finite frequency transmission zeroes. The one on the right will cause a single zero below the centre frequency and the one on the left will create a pair either side of the passband.

One important thing to note is that sometimes it is necessary to place the entire structure on a hotplate in order to solder something internally. When your apprentice has just done this, if asked to inspect it then don't pick it up with bare hands to get a closer view. These are heavier than expected.

[ajb]

That is a coupled coaxial cavity filter.

It's is a seven section quasi-elliptic with three finite frequency transmission zeroes, two on the low side and one on the high side and I'll guess that it's an 850 or 900MHz cellular basestation transmit band filter. It's almost certainly the full cellular bandwidth and I'll guess less than 1dB insertion loss through the passband and more than 60dB of attenuation through the receive band, which will be rather close to the passband.

The extra bits of copper in the middle are what creates the finite frequency transmission zeroes. The one on the right will cause a single zero below the centre frequency and the one on the left will create a pair either side of the passband.

So, Magic.  Got it.

[TSL]

So, Magic.  Got it.

LOL after that description I'd bet a few went WTF?

This is somewhat similar to a filter used in ham circles called an Interdigital filter but, in this case, also using cavities. - don't be fooled by the name, it's got nothing to do with digital transmissions.

Each cylinder forms a cavity at the frequency of operation, each centre pole is about 1/4 of a wavelength long and is capacitively tuned by the proximity of a screw opposite the end of the pole.

The wave front propagates from cylinder to cylinder and each tuned rod will selectively attenuate either a single frequency or form a low pass or high pass filter with its neighbor to only permit the desired signal to exit the entire arrangement.

More info can be found here...
http://www.amateur-radio-wiki.net/index.php?title=What_is_a_cavity_filter%3F

and here...

http://www.w1ghz.org/QEX/Waveguide_Interdigital_Filters.pdf

and google is your friend

cheers

Tim

[KJDS]

So, Magic.  Got it.

There's very little in the way of magic, just a lot of very well understood maths. We're almost at the point of being able to scan the spec and output the manufacturing drawings with no human intervention in between.

When I say it's well understood maths, it's probably fairer to say that there are a very few who truly understand the maths, but most have written software engines to allow mere mortals to create high performance filters that perform very close to expected on a first iteration.

There are probably no more than a couple of dozen people in the world who you could put in a room with a computer and some basic reference texts that would be able to create the design software. It just happens that virtually all of those capable have done so. The software that works very well all remains propriety and is far in advance of the filter design wizards in even high end RF EDA packages.

[Richard Crowley]

We use gigantic versions of these for tower/antenna sharing between several high-power TV transmitters. The filters allow the primary channel signal to go up the tower to the antenna, but filter the power from feeding back into any of the other transmitters connected to the same feedline/antenna.

[Richard Crowley]

That red "pylon" antenna to the left on the top of this tower is broadband enough to handle the entire UHF broadcast band, and can handle a total of 1MW.
And probably has a practical limit of 6 different TV transmitters on different channels feeding it.

https://maps.google.com/maps?q=45.525096,-122.746609&num=1&t=h&z=20

[nfmax]

Definitely magic! Look, the sorcerer's apprentice has even left his broom!

[grumpydoc]

Definitely magic!

Yup

Personally, I know I'm way out of my depth once frequencies get high enough that the connections start to look like plumbing.

[GK]

We use gigantic versions of these for tower/antenna sharing between several high-power TV transmitters. The filters allow the primary channel signal to go up the tower to the antenna, but filter the power from feeding back into any of the other transmitters connected to the same feedline/antenna.

I used to build that stuff. Recall climbing into combiners and filters, from the top, head fist, to clean them out prior to assembly and getting dizzy upside down on metho fumes.

[Richard Head]

It looks like there's a dual directional coupler on each output and a thin flexible coax feeds the reflected power to a Bird 1kW attenuator for spectral measurement.
I suppose the copper tubes are rigid air dielectric coaxial cables, not waveguides.
Great picture. Thanks for sharing it.

[jeremy]

The software that works very well all remains propriety and is far in advance of the filter design wizards in even high end RF EDA packages.

Very interesting! Can you give some examples of this?

[KJDS]

It looks like there's a dual directional coupler on each output and a thin flexible coax feeds the reflected power to a Bird 1kW attenuator for spectral measurement.
I suppose the copper tubes are rigid air dielectric coaxial cables, not waveguides.
Great picture. Thanks for sharing it.

The way those filters work is that you have a pair of identical filters mounted between a pair of 3dB hybrid splitters. The signal is put into one port of the hybrid on the input end of the bank and then the two identical filters are on the two 3dB ports of the hybrid. Anything outside the bandwidth of the filters is reflected back to the isolated port of the hybrid and anything on frequency is passed through and appears on one port of the output hybrid. That output signal can then be connected to the other port on the next hybrid pair which is tuned to a different frequency.

I've drawn a very poor sketch to show this