Decca 3

The above image shows the playback digital to analog converter (top 1U rack case), the record
analog to digital converter (lower black box), and the video recorder 'codec' with time code
electronics (large silver box). Unfortunately I have little technical information on this Decca
system, I know the sampling rate is 48Khz, and that they used a very powerfull error correction
system that would not conceal errors if the data stream failed. It just muted the output, so that it was
quite obvious to the engineers if there were any problems.

(Here we see the system running in playback mode.)

Apparently the data was written into a memory in rows and read out in colums so as to spread a
given data word over the tape. A dropout would then have less chance in losing a whole word. The
error correction could then cope with the broken words and reconstruct them, and the parity data
was also scrambled along the tape track. The coding was initially Miller coding and later changed
to Bi-Phase coding . The Miller light shows errors in the data stream as received from tape if the
data contravenes the raw data coding rules due to dropouts. The processor may also be set to count
tape dropouts which indicates the overall state of the tape. I believe that the tape transports could
be set to shut down if too many dropouts occoured.

The last versions of this IVC based system used 20bit A to D converters,

From the start the tape transports could handle 18 bit data, but in 1976/7 16 bit A/D conversion was
all that was possible.

Time code data with the facility for User bits was incorporated in the begining part of each data
frame. I believe though that Decca's time code system preceded the S.M.P.T.E. standard.

They also used a 'word clock' signal, which I believe reduces the digital 'jitter' in the recording
process. One may remember that this was an advanced system in its day, still earning its living over
20 years later...

The digital converters in this particular example are dissimilar, and are obviously of different dates
and manufacture. It was apparent during our visit that many more D to A's than A to D's were
produced. This is logical because there would be a need for more monitoring (play back) than
encoding (acquisition) machines.

This Decca made signal processor seems to do a number of various things:

1) Takes the data stream from the input A to D converter and changes it into a format suitable for
recording by the modified IVC one inch video recorder.

2) Generates 50Hz synchronizing pulses to interleave between the digital audio data so that the
video recorder will have a signal to lock to.

3) Provides a data error correction system.

4) 'Scrambles' the digital data to spread data words along the tape track to improve the
performance of the error correction system. Because this was a true data recording system, there
was no compatibility with commercial equipment that just put audio words into a standard video
waveform.

5) Concurrently generates and reads in house Decca time code (hours, minutes, seconds, frames
and user bits).

6) Simultaneously takes the off-tape signal (the transport has simultaneous playback) and
converts it back to a format suitable for the audio output D to A converter.

7) Provides a LED PPM type display of the off tape and 'E to E' digital signal level. This
actually was the World's first digital peak reading meter, It also has a linear decay
characteristic.

8) Indicates 'PLL' - This checks on the Phase lock loop oscillator in the processor, as it locks to
the incoming data stream.

9) Indicates corrected and uncorrected errors for each channel. The 'Miller error' light in this
late system just indicates dropouts.

10) The time code display can be switched to count the actual numbers of errors on the tape.
This is a performance check of tape and system. Although having perfect error correction, by
counting corrected errors in say 1/2 minute gives a health indication of the equipment.

Dither was not required in this machine as there was generally enough random noise in audio
input for the A/D . None was required in recording or D/A conversion - only in post production .

Because the video recorder has true 'confidence' recording (separate video/data record and
replay heads) with proper 'E to E' switching, and that the processor encodes and decodes both
audio and time code simultaneously, the output and the various processor indicators are always
showing what is actually happening on the tape. There also seems to be a simple form of self-
diagnostics built in.

Mr Griffiths remarks that 'off-Tape monitoring during a session was seen as ESSENTIAL. Many
Sony users recorded sessions only to find on play back it was no good or not even there ! A most
important reason for not using commercial equipment. Manufacturers just did not understand the
cost implications of losing a recording. A Pavarottii concert would be Millions !'