[sdiy] MIDI-CV DACs Part #2

[Peter Grenader]

Bud Lite, thank you.


on 9/18/02 2:56 PM, Paul Schreiber at synth1 at airmail.net wrote:

> Now let's look at issues with the DAC itself.
> 
> When you think of a DAC, the first thing that pops into your head is "how many
> bits", but
> *really* the first thing needs to be "What is the Integral Non-Linearity (INL)
> error?" Unless you
> are an 'IT Professional', and the first thing you think of is "Who is buying
> the beer?"
> 
> INL is the overall ACCURACY of how well the DAC 'hits' the output you tell it
> to. You can have a
> 24-bit DAC, tell it to output 1.000V and if it outputs 1.0175V that's a
> problem. DACs are
> designed to be used one of 2 ways: as an open-loop, absolute voltage
> generator, or in a "set
> until happy" system (like say a volume control, where the user doesn't care
> what the digital code
> is to generate the output, just the output matters). In a MIDI-CV converter,
> the WHOLE IDEA is to
> translate a SPECIFIC code (the MIDI note) to a SPECIFIC voltage. That is why
> you really have to
> slog through all of this stuff :)
> 
> Much DAC confusion is associated with MIDI being "only 7 bits of information".
> Heck, an 8-bit DAC
> is one better, right?!?? Errr....nope. All that means is that there are 127
> possible outputs, but
> it says NOTHING AT ALL about how ACCURATE those 127 outputs have to be. I have
> designed systems
> with 4 output voltages, but they had to be within 4uv at 2PPM ACCURATE. So, do
> not confuse "code
> length" to "DAC bits".
> 
> So, just *how* do you select the DAC??!?
> 
> Part #3 shortly.
> 
> Paul S.
> 
>