[sdiy] Consider this DAC

Sun Mar 14 11:50:11 CET 2010

Piano tuning standards are inadequate for the discussion here for
several reasons:
1. As Antti mentioned already, the notes are two- or three-fold unison
already and will beat no matter what;
2. Piano timbre is acyclic which hides the phasing artefacts which are
easy to find in synth timbra
3. The stretched tuning of a piano, which becomes extremely apparent
in the top octaves, hides exactly the effect of tuning deficiencies
which become clear on a synthesizer
4. The way that piano is played. It is a different instrument from a
synthesizer and I hope everyone finally agrees that for all intents
and purposes the only thing it (sometimes) shares is the (unfortunate)
keyboard layout.
Sustained high-register notes are not used nor are they expected on
the piano; they are short, sudden stabs; sometimes they have some
sub-harmonic sustain after a glassy hit. Quite often the highest
octaves have such weird timbre (due to stretched tuning *and* tuning
difficulties) that they are only used as inharmonic stabs /
counterpoints / percussive noises. You'll notice that a lot in jazz.

All this means that high tunning accuracy in the highest registers,
such as postulated in this thread, isn't necessary in the piano.

What's more, I think there's no reason to look at industry standards,
better use your brain. Right now the industry standard in our craft is
the Juno Stage - want to follow industry standards? MIDI has been an
industry standard for 20 years and it kills music. Industry standards
in VCF modules are bad ripoffs of the EMS or Moog filter. Behringer
gear met and still meets all industry standards, such as
ISO-1234567-whatever and noise floor and THD.

Magnus, if you just want to detune VCOs relatively, you can use a
second DAC (especially feeding a V/Hz input). But then you can ask
what happens if we use more than one instrument in a song (i.e.
always) and I will say that you're right with the absolute accuracy
having to meet relative accuracy standards.
I think that the answer could be that numerically controlled VCOs
should be controlled via a DAC in V/Hz mode. This coupled with an ADC
and all that put together into one big feedback mechanism for an
adaptive tuning algorithm could work quite well.

The adaptive algorithm would only have to work when the pitch is
expected to be constant, i.e. when the CV from the DAC is not
changing, because that's where I think it really matters - if pitch is
changing rapidly, it's more difficult to notice lack of accuracy. So
for example, find the moment when the DAC output is constant, and feed
statistical data (i.e. frequency noticed vs frequency desired) back to
the brain.
Or, you could generalize the notion of 'constant', and take a
relativistic approach: if the desired pitch right now is x Hz, and
speed of change is xHz/s, then the statistical data should indicate n
cycles (+- several) during the testing time. That tests the first
derivative. You could go and check the second derivative fitting too.
This first-derivative approach can be expressed in this way: if you
look at the graph of your function of pitch (vertical axis) vs time
(horizontal axis), then the segment which you are looking at should
fall inside a box of specific height defined by the specific time you
are testing for.
The first-derivative approach (note that I am not sure this is the
right term, I'm sure someone has named this already and it's probably
got a smarter name) also answers the question: 'how to make sure that
my notes are in tune between eachother when I am performing a
glissando?'

Joachim, I heard the Q+ has something like 'first order tuning error
correction', is that what it is? I never figured out what that meant.

With regards to the recurring noise floor question: the biggest
problems right now are bad layout; in physically bigger (than 0.5 m^2)
systems, the use of single-wire signaling; the parts themselves give
us loads of accuracy.

With regards to midi tuning: the fact that it specifies an
evenly-distributed tuning table only means that no note will be on the
pitch that it should be at; this sort of bare-data approach was nice
in the 80s when processing one byte could take minutes, but nowadays
there's no reason to not specify the exact tuning in a literal way,
i.e. send a string with the content "325.14555".

On Sun, Mar 14, 2010 at 09:59, ASSI <Stromeko at nexgo.de> wrote:
> On Saturday 13 March 2010, Antti Huovilainen wrote:
>> Please correct me if I'm wrong, but a TL072 for example has 18
>> nV/sqrt(Hz) voltage noise. For 20 kHz bandwidth (RC lowpass at
>> 10kHz meaning settling time of less than 0.2 msec) the resulting
>> RMS noise is 2.5 uV. That's quite a lot less than 1mV.
>
> Look up the graph of noise vs. frequency to see that you've missed to
> integrate a substantial amount of noise.  The flicker noise corner is
> at around 400Hz for the TL072 and the noise figure on the first page
> is for 1kHz and a source impedance of 50Ohm (guess what?).  There's
> also this pesky little thing called offset and offset drift (with both
> temperature and common mode voltage) that will need attention.  But
> maybe I'm still missing your point.
>
> My definition of 1mV _precision_ is that there is at most +-0.5mV
> deviation from the ideal CV across all operating points.  For a 10Vpp
> signal range that results in about 14bit precision across the whole
> system (in voltmeter speak: 4½ digits) - which is doable, but not as
> easily or cheaply as you seem to believe.
>
> Getting back to the CV->VCO example: "merely 1mV precision" across the
> whole system (just having this for the CV is not going to cut it)
> would allow you to set up two VCO five octaves apart (say 220Hz /
> 7040Hz, after calibrating the beats out with both VCO at the same
> frequency) and have them beating at less than 2.5Hz.  Not only that,
> if you reverse which VCO plays the lower and the higher frequency, you
> still get less than 2.5Hz beats.  Modulating both VCO with the same
> voltage never produces any faster beats than said 2.5Hz and
> transposing the patch four octaves lower (13.75Hz / 440Hz) produces
> less than 0.2Hz beat frequency of the 32nd harmonic of VCO1 vs. the
> fundamental of VCO2.  Lastly, if you switch that hypothetical synth
> off, let it sit for a year and switch it back on, you would get the
> same results (not exactly the same beat frequencies, mind you, but in
> the same range, perhaps after warm up).  Ian Fritz might have such
> wonderfully precisely calibrated VCO, but there surely are a lot of
> systems that would fail this test.  In other words their precision is
> worse than 14bit/1mV and yet they are still perfectly useable.
>
>
> Achim.
> --
> +<[Q+ Matrix-12 WAVE#46+305 Neuron microQkb Andromeda XTk Blofeld]>+
>
> Waldorf MIDI Implementation & additional documentation:
> http://Synth.Stromeko.net/Downloads.html#WaldorfDocs
>
> _______________________________________________
> Synth-diy mailing list
> Synth-diy at dropmix.xs4all.nl
> http://dropmix.xs4all.nl/mailman/listinfo/synth-diy
>

Cheers,
D.