[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Serial D-1 Jitter

     Several things you might look at:
     1.  Is it possible that the foam core of some of these cables has 
     been smashed or deformed.  This will case reflections.  In rare 
     cases, a tight bend radius in the cable dressing can cause 
     reflections, depending upon the cable type and outer jacket 
     Where we all have rolled over video cable in the NTSC past with 
     heavy equipment, it's strictly a NO NO in a serial digital plant.
     2.  Are you using "self Normalizing" patch panels.  Debate has 
     been raging for years over the pro's and con's of using self - 
     normalizing patch panels vs. non - normalizing with "hair pins" 
     for 270mb/s video.  I have used both.  
     Critics of self - normalizing say that over time and non-use 
     (inserting a patch card in a patch jack) can cause the self - 
     normalizing contact to corrode or become dirty, thus inducing 
     distortions in the transmission chain.
     3.  Are all of your cable, connectors, patch panels, truly 75 
     ohm?  If you started with a NTSC analog house and slowly migrated 
     to serial digital was there a chance older transmission hardware 
     was re - used.  For years we all used, at on time or another, 50 
     ohm connectors and in some cased patch panels.  This was of no 
     concern when signals only went out to 10mhz.  Now with signal 
     going out to 270 mhz, it's critical.
     4.  The most obvious is, how long are your cable runs?  Maximum 
     length without reclocking will vary from cable types.  I've run 
     Belden 1505A out to 300 meter with a successful packet recovery, 
     however my recommendation is "don't try this at home!" ... that's 
     really pushing your luck.
     5.  You stated that you were using a GVG (I assume Series 7000) 
     serial digital router.  This router (at least the one I put in 
     some years ago) reclocks on the input.  You will note in the 7000 
     specifications, that it will pass signals up to 360mb/s.  This 
     360mb/s spec. was for provisions for increasing luminance 
     resolution to 18mhz sampling for 16:9.  However, the trade off is 
     "More Jitter" vs. adhering to 270mb/s, at least with today's 
     current chip-sets.  This condition is true even if the signal is 
     The pro's and con's of 18mhz luminance sampling is beyond the 
     scope of this discussion.
     6.  One last item:  would any of this trouble be associated when 
     using a DVR-2100/OS.  If you are reclocking while using the 2100 
     in off speed operation, that signal comes out serially at a 
     "nonstandard" data rate.  Some devices tolerate it better than 
     With a cable of known reliable 270mb/s performance, try 
     bypassing your patch panel, or try a whole different cable path 
     and see if the jitter performance changes or is consistent.
     More more info I recommend an article in the August 1992 SMPTE 
     Journal written by Takeo Eguchi, from our Sony Atsugi 
     Technologly Center, titled: 
     "Pathological Check Codes for Serial Digital Interface Systems".
     We may have an all serial digital pathway now, but it is still 
     susceptible to analog degradation.  The good news is that when 
     the signal performance is marginal at best, visually it's still 
     "great", when it steps over that fine, sometimes invisible line, 
     it "catastrophically fails".
     Hope this has been helpful.
     Nathan Simmons
     Sony Digital Production Systems
     Hollywood, USA  
______________________________ Reply Separator _________________________________
Subject: Serial D-1 Jitter
Author:  telecine at xyzoom.alegria.com at sonycom 
Date:    1/24/96 1:36 PM
                      Subject:                              Time:  12:56 PM
  OFFICE MEMO         Serial D-1 Jitter                     Date:  1/24/96
Dear fellow Telecine Group members,
We have a pretty big GVG serial router for our D-1 signals here ( 96x96 ). 
Some of those signals are derived from Parallel only devices such as 
Digiscan-4 Ranks or older Accom DNR's and are serialized using Miranda 
parallel to serial converters.  Others are from devices with direct serial 
outputs such as DaVinci 8:8:8 Processors.  Some of the signals exhibit very 
little jitter when measured on the AAVS Digital Signal Analyser but others 
have up to about 1.60 ns of jitter which sometimes causes failure of the 
signals intermittently.  The jitter doesn't seem to be related to any class of 
device, some Miranda derived signals are low jitter, others are high.  What 
has been the experience of some of the rest of you with dealing with this sort 
of thing?
Dave Corbitt