Seeking Your Interest In RoadTesting Ethernet or SMA Eval Kits

Thanks for the comments.

or maybe CWDM? CWDM technology have gotten way cheaper recently...

The slip ring application is very interesting.

Although you could have multiple fiber links in fairly close proximity, you could only have one fiber in a slip ring application because a slip ring must rotate around a single free-space optical link. You could, of course, use OFDM to send multiple channels though a single link. I'm a little bit tempted to try this technology in a slip ring application.

You ask some detailed, specific questions that I'm not sure the datasheet alone would answer them. I can ask Broadcom.

Randall

This is a very interesting device and I never knew this existed (up to 5Gbps!). The reason I find it interesting is because I see it as the easiest way to implement an optical slip-ring. At work, we use some very complicated  (and expensive) optical slip rings to pass fiber links across rotating things. Something like this could simplify that task by at least two orders of magnitude. The SMA version of the kit is more interesting because The TX/RX can be directly connected to any FPGA Eval kit with transceivers to test. Those typically come with SMA TX/RX ports too so it should be fairly straight forward to test. BERT tests can also be done this way.

The thing I don't like too much is that there isn't much information about the device itself (detailed datasheet?) on the Broadcom website, and also pricing is not available anywhere. For example, I wanted to understand if multiple of these can be used in the vicinity, and how accurate the optical alignment needs to be, etc. This information would dictate what specific use cases it could be used in, for example building slip rings with multiple fibers, etc.

I have interest, but not the equipment to do an adequate test.

I've had that same experience.. trying a high-value attenuation and getting nowhere near it! 

I was wondering how to seal each stage, in the easiest-to-assemble way I could think of. I sketched the following, it still needs work, and might not be the best way.

If (say) it was a 4-stage (where each stage shouldn't be more than about 25 dB) then that's 100dB of attenuation if desired.

The idea being that each section could be boxed with a screening can. The resistors (or at least some of them) would be large (e.g. 1210 sized) SMD, so that they could dissipate some power.

Once the cans are soldered (with a high-power iron), the adjacent cans joining each stage would need to be bridged with solder too, to prevent RF finding a way to the outside world between the stages.

The cans are cheap in quantities of 10. Slightly taller versions would be better, but this is all I could find and might be OK:

The SMA connectors could be soldered on the other side, and then solder used to fillet all the way around the edges of each connector. Also every via hole could be soldered over if required!

Don't know if it's worth building though, because it would be hard work to solder the cans without a high-power iron. Also it would only work up to a few hundred MHz.

Regarding a coupling loop near an attenuator (if it isn't fully shielded), or close to coax braid, it would work, but would be hard to know what the attenuation would be (and it might vary over frequency). But might be good enough! 

Interesting device. I'm thinking of applications in network switches where it can avoid some links between stacked PCBs but there are challenges such as the presence of dust and dirt particles on the optical sensor causing noisy data transfer. Still, I would like to test this as we have some good test equipment for this at the University of Colorado Boulder. High-Speed Digital Engineering department. 

Hi shabaz .  The box is just a lazy man's way of terminating a SO-239 connector.  The bolt on the side is because I gave up trying to land the shield on the aluminum.  It's just a ring ternimal, bolt and nut.  I bought a number of the bodies thinking to build attenuators but they turned out rather poor.  Right now, I only use attenuators for radio direction finding - so I purchased a few and made one out of strip board that attenuates 46dB.  The calculator says 116dB but the coupling says... "I don't think so." 

I've been experimenting on and off with trying to attenuate 500mW to 750mW signals.  I have intentionally bought and made inefficient antennas.  You would be surprised how effective a 3-stack of 30dB attenuators works as an antenna.  Wouldn't a coupling loop near the dummy load act like a potential transformer?  It would be hard to get accurate level readings without plotting a bunch of known values.  I may have to try that out.

Hi Mike, 

Interesting! Glad you're finding what works. Also, just curious, is the box an attenuator?

Depends how it's built of course, if it's super-high attenuation in there (e.g. 30dB or more) the RF might be coupling unexpectedly despite the attenuation.

I found that out when trying to attenuate by 100dB or something (I wanted to do a similar thing, feed a transmitter into a radio receiver), and ended up using separate 20dB and 10dB attenuators (each with their individual metal cylinders) all chained up, and it was totally impractical admittedly! Partitioning inside the enclosure (e.g. walls of copper-clad PCB) could be another idea.

Recently I found this nice thing on eBay (but it's not high power, so would still need a higher-power attenuator to front it):

I've not tried it out yet! Hoping it works. 

EDIT: This would be another way to do it; using a coupler to get less of the signal (and pass the rest into a large dummy load), but this one only offers 15dB reduction, and only supports 4W. Maybe there are higher-power DIY versions suitable for HF use, but I've never checked to see.