I don't recall exactly, but I believe the KEYE Tech Dir told me that the ROVI box interface is very simple, "just hook it up, and plug it in" with no control or clock inputs or parameters to be set by the station at all, a pure "black box" as far as they're concerned. So since we're all speculating here's my best guess at what's going on:
ROVI supplies the encoder to the station and tells them that they have two choices regarding how to install it:
1) DISCRETE (dedicated bandwidth) mode: This is the mode P Smith has been describing. In this mode, the ROVI box has only 3 connections to the outside world, AC power, Ethernet, ATSC-Out. It gets ALL external data including and required clock references from the ROVI mothership via the ethernet connection. The ATSC-Out feeds an INPUT to the station MUX which must be configured to allocate a certain amount of bandwidth to the ROVI box (as well as every other data source that is feeding it).
2) INLINE (scavenger bandwidth) mode: In this mode there are the three exernal connections listed above plus a fourth; ATSC-In. In this configuration, the OUTPUT of the station MUX feeds the ATSC-In connection of the ROVI box (ie ROVI box is "downstream" of the station MUX). ATSC-Out from the ROVI box presumably feeds straight into the station transmitter. In this mode, the station MUX does not allocate any specific bandwidth for the ROVI box which then has to act as a "scavenger" trying to shove data into any available empty space in the ATSC stream (like the boxes being packed into the train cars of the previous analogy).
To a station Tech Dir trying to maximize HD-PQ or subchanels (or both), INLINE mode would have a certain appeal. It would leave him with a little extra bandwidth that he could use for other purposes. So here is what I think is happening nation-wide:
- Some stations have installed the ROVI box in DISCREET mode. These stations have never (and will never) see the clock skew issue.
- Some stations have installed the ROVI box in INLINE mode, but (to date) have not fully allocated their ATSC bandwidth. These stations have not seen the clock skew issue SO FAR, but they are vulnerable to it in the future if they add more subchannels or try to increase PQ.
- Some stations have installed the ROVI box in INLINE mode and have fully allocated their ATSC bandwidth. These are the stations that have seen the clock skew issue either intermittantly or chronically based on bandwidth usage.
I realize this is all highly speculative for someone who's never set foot in an affiliate or seen a ROVI encoder, but we now have two separate cases (KEYE and WUSA) where the documented fix proposal from ROVI says, "Relocate the ROVI encoder further "upstream" within the station infrastructure and allocate a specific amount of ATSC bandwidth to it". If my theory is correct that is the same as saying, "stop using INLINE mode and start using DISCREET mode". The response from KEYE was "No way!". The response from WUSA was apparently, "We're working on it."