[BC] WLS dead cxr prob. this a.m. at ~6:50

[Mike McCarthy]

The Moseley unit fails to 0 V if the unit should fail or loose power.  At 
-60dBu signal, the meter output is something like 3V dropping to a little 
under 2 V at -80dBu.  It's marginal BER/fade point is -90dBu and total loss 
at -95dBu.  If the RX level drops to -80dBu for more than 30 seconds, the 
macro is triggered to "phone home" with the codec and then place it on the 
air.  Takes about 2 minutes which is usually sufficient for fades around here.

It's not the BER which is the salvation.  It's the ability to have a 
"smart" comparator with anything that can give you a QoS indication and 
then have a sequence of commands to make the switch when the appropriate 
threshold is reached.

MM

At 10:25 AM 7/1/2006 -0400, Paul Christensen wrote
>>That's the beauty of the GSC3000.  When the signal strength on the STL 
>>goes below a certain threshold, a macro is triggered to dial the back-up 
>>codec and then put it on the air automatically before the STL fades.
>
>That's an indispensable technology to have in these instances as well as 
>microwave paths affected by rain fades (e.g., 23 GHz switching to 950 MHz 
>or telco backups).  I suppose the only thing the GSC3000 could not protect 
>against is a total failure of the unit, and hence the loss of the 
>BER-based switching control.  Mike, maybe you can clarify on that 
>point.  This type of sampling and path switching may be best located 
>external and independently to the very equipment it's sampling for that reason.
>
>Prior to the availability of a software solution or even BER detection, we 
>had to design the path-fade hysteresis algorithm by hardware.  In once 
>instance, the only triggering source available was a DC voltage that 
>appeared on a 23 GHz gunnplexer.  In a nutshell more or less, here's how 
>the scheme worked...
>
>- Sample the gunnplexer at the antenna through series R and lightning 
>abatement;
>- Route the sample down a spare conductor of the gunnplexer control cable;
>- At the mod/demod equipment below, install a simple A/D converter based 
>on a 10-segment LED bar.  The gunnplexer dynamic range was approx 0-10VDC 
>and produced 0 to 10 LED bars, depending on signal;
>- Control switching sample at LED segments 3 and 9 (corresponds to 
>gunnplexer V of 3V and 9 V, respectively);
>- When the path fades to 3 segments, circuit automatically switches to 950 
>MHz or telco;
>- When the event is nearly complete, circuit automatically switches back 
>to main STL when 9 LED segments is achieved BUT must remain at 9 segments 
>for a continuous time of 2 to 5 minutes (this protects against rapid, 
>ragged-edge switching as the signal goes above and below a predetermined 
>level).
>
>So, to take this hardware mess and do it all with software based on what 
>really counts (BER), is a real salvation today.
>
>Paul
>
>
>====================================
>Paul Christensen, CPBE, CBNT
>LAW OFFICE OF PAUL B. CHRISTENSEN, P.A.
>3749 Southern Hills,  Jacksonville, Florida 32225
>Office: (904) 379-7802   Facsimile: (904) 212-0050
>pchristensen at ieee.org
>
>
>_______________________________________________
>This is the BROADCAST mailing list
>To send to the list, email: broadcast at radiolists.net
>For sub changes, archives and info on this other lists: 
>http://www.radiolists.net/