For those of you who are into
sampled sound synthesis or just
hardware freaks, I've just built
an 8 bit sampler that can
accurately sample at a maximum
rate of 324545Hz or about 3uS per
sample.
    I've used an ADC020 that uses
32 comparators to do 2 four bit
flash conversions then derives an
8 bit result. The actual conversion
takes about 800ns (due mostly to
the settling time of its analog
switches). 3uS is as fast as the
68000 in Amy can transfer the data
in ML with all interrupts disabled
(nonDMA).
    Instead of connecting it to the
printer port as is usually done,
I've cut the chip selects going to
the CIAs and decoded address bits
12,13 and 14 with a 74F138 and use
the decoded outputs along with the
old BFxxxxH select to enable the
CIAs at their usual address
(BFDxx0H and BFExx1H). This freed
up the rest of the address space
from BF0000H to BFF001H.
    Then I added a 6821 PIA
chip on the lower order data lines
at address BFC001H to BFC301H and
the ADC020 is connected to port A.
Altho the ADC020 would interface
directly to the 68000 bus I wanted
more control over the timing and
a level of protection in case
something should go wrong with the
ADC. I used a LH0033 buffer amp on
the A/D input because it has a
1500V/uS slew rate and 100MHz band-
width (mostly because I already had
them on hand) but any op-amp with a
reasonable SR (not 714) would do.
    Since the reference ,Vref(+)-
Vref(-), need not be ground
referenced I set Vref(-) to 1.4V
with 2 1N914 diodes and used a 2.5V.
Zener from there to Vref(+) because
the output of the LH0033 and most
op-amps can't go all the way to
ground (the LM324 does but its a
dog). This results in a 0 reading
from the A/D with an input of about
0.5V and full scale (255) at
about 3.12V. I also added a 1N4732
zener to the input of the LH0033 to
protect it from going higher than
+4.7V or lower than about -0.7V
below ground. All of this fit
nicely on top of the mother board
and under the RF shield inside my
A1000 with the nesessary leads run
through the holes that the bottom
case posts fit through and soldered
to the bottom side. Finaly, the
dreaded trace cuts were made near
pin 23 of the CIAs on the solder
side.
    It all works great and after
using it for a month now with a
variety of software I've had no
problems.
    To make it work with
Perfect Sound and AudioMaster all
I had to do was write a small
program to set up the PIA in my
startup-sequence and I changed
the reference to BFE101H in each
program to BFC001H. Unfortunatly,
AudioMaster seems to have a
maximum sample rate of 24000Hz and
Perfect sound has a maximum of
about 28000. I have written some
stand-alone programs that do
sample at a full 324KHz and save
to disk or ram.
    I'm sure your now saying:
"That's all well and good but the
amiga won't play a sample at more
than 28000Hz". The advantage of
the higher speed comes in doing
DSP (Digital Signal Processing).
For example you could run the
sample through a digital low pass
filter then upsample, reducing
the size of the sample and
increasing the pitch (rate) so
it could be played at 28000Hz.
This would result in a much
cleaner sample and reduce any
noise. The higher sample rate is
also very important doing waveform
analisys (using an FFT program
I've writen).
    I'm also planning to write
a program that would allow me to
use it as an oscliscope.
    If you have any comments or
suggestions please leave a message
or write to:
                       
    Michael R. Walters
       129 Columbo
    Tustin, CA 92680