ComputerVoltageSources

At 11:25 PM 10/31/2007, you wrote:

>It wouldn't be that hard to do. It's been 2.5 years since I wrote my
>trig-gate-cv delay but the basic concept is I have three 128 event
>buffers. The first is used to timestamp an event (trigger high,
>trigger low, etc.). The second buffer is to record a trigger or gate
>transition (high or low). The third buffer is to record a CV value.
>When I detect a trigger, I timestamp it with a 'delay' factor and put
>the timestamp, trigger, gate, and CV in the buffer. The buffer is
>circular and at 128 events long it never fills up.

Check out the Modcan CV Recorder 57
It samples low frequency analogue input and stores the values to SRAM.
When the 512K x 8 ram is full it plays it back in a loop.
Loop start and end points can be set with panel controls and record
/playback speed is also continuously variable.
I can store about 4.3 seconds of 4 channels at a sample rate of 20kHz.
A 12bit quad DAC is used for the output. Sampling is done using the
dsPIC analogue inputs.
I pack 2 12bit values into three 8bit memory locations to extend ram.

Multiple samples strung together can be done by setting the loop
start/end and it will only record between these limits.
All four channels can be recorded independently.
It was pretty easy to do as it is just a data logger but it does take
a fair bit of expensive hardware.
64 pin dsPIC, Quad 12bitDAC, SRAM numerous rail to rail op-amps etc.
and uses about 180mA of current.
It is the most awesome joystick movement recorder though.

Bruce