[sdiy] Kurzweil K1000 strange problem sounds solveable..

[Bob Weigel]

Ok...I'm getting, even after hard reset, on any sound, a very obnoxious 
parsing of the sound being played back.  ALWAYS at the same points after 
attack.  Regardless of which sample is being played.  So I'm 
thinking...there's a missing line somewhere..or... anyone experienced on 
these machines?  Other than key dirt and a data slider my K1200 has been 
solid.  Someone brought this too me the other month and I hate to put it 
to rest. -Bob
harrybissell wrote:

>Ola
>
>Matrix 1000/6/6R use a programmable timer chip called 82c54. You
>basically
>choose how much the incoming clock pulse frequency is divided by writing
>to
>the different data registers (from the CPU) of the timer. The clock
>pulse
>is generated from a high frequency crystal oscillator (MHz) (very
>stabile).
>The resulting clockplulse (0-1-0-1) is then fed into the CEM3396 which
>shapes the pulse into sloped waveforms (amongst other things as PWM,
>filtering, VCA etc..)
>
>The Matrix 6 / 6R / 1000 actually have high frequency VCOs. These drive
>the
>82C54 chips. They are used for very small variations in tuning, they are
>not
>wide range.   Most of the frequency control comes as described above,
>from
>changing the programmed counts in the 82C54 chips.
>
>Matrix 6 / 6R has TWO HF-VCO (one per oscillator).   The Matrix 1000
>shares
>a single HF-VCO between the two oscillators.   (to be perfectly correct,
>the 6 and
>6R have a total of three HF-VCO and the 1000 has a total of two HF-VCO)
>This is said
>to account for the difference in identical patches sounding different in
>in the
>Matrix 6 / 6R and the 1000.
>
>These chips output a pulse that is used to SYNC the CEM3396 waveform
>converters.
>The CPU is responsible for TWO actions here... outputing a voltage that
>will set the
>rate that the DCO capacitor is charged at... and setting the counts for
>the 82C54
>reset pulses.
>
>The advantage of this method is that the accuracy of the current source
>can be a little
>'off'... but since the reset is locked to the 82C54 chip the error will
>be a slight variation
>in the AMPLITUDE of the wave (which you cannot hear easily) and not in
>frequency
>(which is VERY obvious when its off...)
>
>There is an internal calibration mode where the computer establishes a
>relationship
>between the current charging rate, and the reset timing.  This will
>eliminate or reduce
>any distortion in the wave due to component changes or temperature.
>(they calibrate the filters during another part of this routine).
>
>Depending on the frequency range of the DCO, they actually change the
>current source
>scaling (allowing a much higher current for higher notes) to preserve
>accuracy. This change does NOT occur at a single point, but seems to be
>related to whether you are
>ascending the keybaord, or descending.  There is a hysteresis in it. Its
>VERY weird and
>not at all what you'd (I'd) expect.
>
>An AMAZING amount of the Matrix 6 / 6R / 1000 system is in software. The
>hardware
>is really quite simple.
>
>Here ends more than you ever wanted to know about the Matrix 6 / 6R /
>1000.  I've been studying the design so I can hack guitar synth
>waveforms into the Matrix 6R / 1000.   :^P
>
>H^) harry
>
>
>
>
>
>  
>