FWD: RE: DAC resolution

[joseph s ramstrom]

----- Forwarded Message Starts Here -----

From: 104065.2340 at compuserve.com
Date: Tue, 9 Mar 1999 13:34:01 -0500
To: Fraser_ Colin J
Subject: RE: DAC resolution

Message text written by "Fraser, Colin J"
>> -----Original Message-----
> From: Terry Michaels [mailto:104065.2340 at compuserve.com]
> Sent: 09 March 1999 16:37
> To: Fraser, Colin J; synth
> Subject: RE: DAC resolution
> 
> +/- 0.1% of full scale, which is 128 notes, is +/- one-eighth 
> of a note. 

Full scale is 256 notes - that would be +/- quarter of a semitone - that
probably would be noticable - I must have been lucky :-)
I have measured the output voltages of for the first 12 notes on an 8 bit
DAC, and the accuracy of each 0.083mv step was better than I could measure
with a 3.5 digit mm.

> The
> advantage of using 8 bits of a 12 bit DAC for this application is the
> maximum differential linearity, or note to note error, is 16 
> times smaller,
> because the 12 bit DAC has 16 times more resolution, and the maximum
> differential nonlinearity will be only 1/16th of a semitone. 

But surely you can only take advantage of this increased resolution if you
use the extra bits ?
Wouldn't you have to measure the input to output error of your DAC and
store
the precise 12 bit value required to give each of the 256 output voltages
you were after for 8 bit resolution ?
 
> The maximum
> deviation of a straight line over 128 notes will also be 1/16th of the
> maximum deviation of an 8 bit DAC.

Ignoring the theory and specs for a moment - inside the DAC is a ladder of
resistors setting the voltage level for each bit.
In most DACs these will be laser trimmed to the correct values.
Why, assuming the manufacturer wants to make the most accurate part he can,
would the dividers in the 8 bit DAC be so far less accurate than the
dividers for the first 8 bits in a DAC with more bits ?


Colin f

<
Hi Colin:

You don't access the extra 4 bits, you tie them low. You are still "using"
the extra bits by doing this.

This whole issue is based on the real world capabilities of DACs and the
way they are made.  The point I'm trying to make is you will always get
better results with a 12 bit DAC, for not a lot of extra expense.

My perspective is as an electrical engineer, and although I have not used a
DAC for pitch control, I have designed and built many products that used
various types of DACs.  Manufacturers don't necessarily want to make the
most accurate part they can.  They want to make parts that perform within
the listed specs, subject to all the limitations of making such a device: 
manufacturing cost, yield, reliability, etc.  It costs more to make a 8 bit
part with 12 bit accuracy, and though the manufacturers might
"accidentally" achieve that level of accuracy, it is secondary to getting
the parts out the door.  Making an 8 bit DAC to 12 bit accuracy will
probably require a longer time spent in the laser trim process, in a more
accurate test jig.  They will get more rejects of parts that won't make it,
so the yield will be less.  All this costs money, and you can trust me when
I say the manufacturers won't spend the extra money if they don't have to.

I didn't mean to say your doing this wrong, it looks like 8 bit DACs seem
to work for you just fine for you.  I guess it is more correct to say I
would use a 12 bit DAC to generate control voltages if I were doing this,
mainly because I would worry much less about pitch accuracy, for a minimal
increase in parts cost.

Terry Michaels

----- Forwarded Message Ends Here -----