Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/18/84; site hydra.UUCP
Path: utzoo!linus!philabs!cmcl2!harvard!think!mit-eddie!cybvax0!frog!hydra!die
From: die@hydra.UUCP (Dave Emery)
Newsgroups: net.video
Subject: Re: problems with HiFi VCR's
Message-ID: <134@hydra.UUCP>
Date: Fri, 4-Oct-85 23:25:53 EDT
Article-I.D.: hydra.134
Posted: Fri Oct  4 23:25:53 1985
Date-Received: Wed, 9-Oct-85 06:39:58 EDT
References: <38@utecfc.UUCP> <20600020@hpcvry.UUCP>
Organization: Charles River Data Systems, Framingham MA
Lines: 76

> >> 5) Worst of all, a distortion problem which seems level and frequency 
> >> dependent.  This is really apparent with solo recordings (for example, a 
> >> female opera singer or a trumpet) where there is no other music to mask
> >> the problem. I would describe the distortion as a harsh, jagged kind of 
> >> sound which accompanies the music. This occurs at levels below the red 
> >> LED (0dB). I recorded some sine wave test tones at different levels and 
> >> frequencies. The performance seems OK at low frequencies, but at high 
> >> frequencies the same harsh, jagged sound emerges, somehow superimposed 
> >> over the test tone.
> 
> Because of the way the signal is recorded on the tape (in 1/30 second
> slices) the playback signal contains a 30 Hz switching transient.  The
> system has circuitry to remove this noise and normally does very good 
> job of it.  However, (for reasons I don't understand) it fails miserably
> on high tones without much background.  The worst case of this being test
> tones.  This happens on both Beta and VHS HI-FI and has been reported
> in Video Review.  I have confirmed it on all of my machines.  There
> does not seem to be anything that can be done about it.  

> ---  Raan Young (hp-pcd!raan)


	I believe that there is quite simple explanation of this 
phenomonon.   It has to do with the effect of a slight time delay
error in the signal coming from one spinning head versus the signal coming
from the other.   

	Assuming a clever and distortionless circuit for handling the
switch between heads, there still exists a form of distortion that
should be worst on pure high frequency notes.  That distortion
is phase modulation of the high frequency signal at the head switch
rate due to slight errors in the angular position of one head versus
it's opposite number 180 degrees across the head drum.  This can
cause one head to be reading FM audio delayed by a few hundred
nanoseconds to a couple of microseconds or so  (1 us corrosponds to 39 seconds
of arc in head alignment for a head drum spinning at 1800 rpm).
The effect of this difference in delay is to cause a phase shift
in the audio demodulated from the fm carrier every time the electronics
switches from one head to the other.  Or in other words to phase
modulate the high frequency notes with a 30 hz square wave.

	At 10 khz a signal has a period of 100 us and a jitter
of as little as 3 microseconds would cause a phase modulation of
10 degrees.  A phase modulation of 10 degrees results in first order
sidebands about 21 (approx) db below the tone, this corrosponds
to a 'IM' distortion of about 10 percent (Pardon my simple
estimates, I haven't bothered with more terms of the series than
the first couple).   I am not an expert on phychoaccoustics
but I suspect that a tone this impure (with sidebands stretching
out from + and - 30 hz) won't sound as sweet as a nice pure
10 khz tone with little or no sideband energy near it. 

	Obviously the distortion is much less at lower frequencies.

	It may very well be that this distortion is much less audible
when AM modulation (by instrument and musician) generates the dominent
sideband energy, thus only passages with pure high frequency notes 
will sound bad.

	It is also possible that the head switching electronics
also introduce a transient due to balence errors of various sorts between
the channels.  I do not know whether this is worse in magnitude than the
phase modulation.

	The time delay could be corrected by use of ccd or other types
of analog delay elements to compensate for the mechanical errors in the
heads (the tv video sync pulses make a nice reference for allowing such
a circuit to correct dynamically).  This kind of time base correction
is actually applied to video coming out of broadcast vtrs, and it
would substantially eliminate the error due to this cause but would
probably be expensive.

	  David I. Emery    Charles River Data Systems   617-626-1102
          983 Concord St., Framingham, MA 01701.
	  uucp: decvax!frog!die