On Fri, 24 Sep 2004 14:42:54 -0400, "JLB"
wrote:

As for digital signal processing is concerned, just because you have
digitized the signal doesn't mean that you have eliminated all of the analog
problems.

Hi Guys,

This reminds me of the problems I had teaching the digital types
Shannon's law for BER.

When I designed the black box for the 757/767, all of my digitized
readings (taken from 600 leads) was passed over to a specialized tape
recorder (25 hour capacity).

Problem was that this digital signal was fed into the recording head
without any bias. Many may be unaware of the advances in audio tape
recording BW that came by the addition of bias, and more, that it
reduced the head's tendency to erase its own stored signal. In
essence, with no bias, you were recording data with two strikes
against you. This was not the hallmark design for a data sensitive
product. Worse yet, was this digital mentality had recorder
specifications that allowed for a S+N/N of 2.

What were the comments I heard in response? "This is not HiFi, it's
digital data, on/off."

Within weeks I was drawn into their simulator lab to view a simulated
cockpit of an KLM aircraft that was being used to display flight
recorder data that was rather -um- noisy (much too much for KLM's
engineers to make sense of it). I watched that plane hit the ground
several times as they struggled to recover digital bits lost in analog
noise.

Several (many) years later I was called to consult on the TWA flight
800 data - more noise that lead to hints of missile strikes. The
panel's best spin on the topic was that it was old data printed
through the new data (even though the difference in time would not
correlate to the two data sets overlapping). We reported it as
exploding gas tanks, Tehran reported it was revenge for our shooting
down one of their civil aircraft during the first Gulf War. Like the
first attack on the Twin Towers, government and the media shrugged off
correlations.

73's
Richard Clark, KB7QHC