[sdiy] Bar Graph Driver as Quantanizer

[Kyle Stephens]

Some musings...

While looking at the LM3914, a bar/dot graph driver, I realized that it's basically a quantinizer. With a few tweaks and a stable voltage reference, it'd serve as such. I think... Help me here if otherwise.

Datasheet's here for reference: http://www.national.com/ds/LM/LM3914.pdf

One big question, is how could I use just 2 chips(*) for several octaves (vs. 1 chip per 10 notes...?). Some kind of dynamic reference voltage offset...? Not unless that's not a completely unreasonable solution. The datasheet says you can chain together up to 100 outputs. With 12 discrete keys per octave, that's 100/12 = effectively 8 octaves.

(*) At least 2 chips would have to be employed anyway, as they only have 10 outputs each. So, call it 2 chained ICs for 20 total outputs, 12 of which would be used for 12 "bits" of resolution/levels of quantification. That's assuming the "just 2 chips" configuration is possible, otherwise it's just a step-after-another progression.

There's also the LP3943, which has 16 outputs to start with. It's a low power IC though - has that any share of potential issues?

Datasheet: http://www.national.com/ds/LP/LP3943.pdf


The 3194 has a 'bar' and 'dot' mode; bar is 'from ground up,' with each output giving a fixed voltage, which when summed would give you the quantinized CV.

Switch to dot mode, which triggers the output designated closest to the input voltage, and you got a voltage controlled event generator; between your preset step range (dunno the exact hysteresis thereof...), gate 'a' fires, next step gate 'b', etc.

On the note of hysteresis, I looked it up in the datasheet just now and it said there isn't any built in, and there's a fadeover of 1mV or more between outputs (apparently less if the next "step" isn't on the same IC, for chained circuits). I'm not sure how to address that... A comparator, or a specific output-voltage referenced Schmitt trigger, per output??
(x12 is kinda steep). How many cents off would ~1mV bend a VCO anyway?

Any other thoughts on this? And again, is it even an effective, let alone feasible idea? Just wana scope things out before I hit the breadboard.


Also: a grateful if belated thank you to everyone who responded about my MS-20 filter clone questions!


_Kyle