Temperature-compensated resonance ( was Re: [sdiy] RE: [AH] Parametric EQ)

[cheater cheater]

On Mon, Feb 22, 2010 at 17:51, Tom Wiltshire <tom at electricdruid.net> wrote:
>
> On 22 Feb 2010, at 14:54, cheater cheater wrote:
>
>> Tom,
>>>
>>> Aww, c'mon! Who ever bothered having a temperature-compensated resonance
>>> control anyway? Certainly no synth I own has one. You're lucky if you get
>>> a
>>> tempco in the filter *cutoff*, let alone the resonance.
>>
>> This only shows the sad state synth designs are in. Most high quality
>> equalizers have tempco Q.
>
> Can you back that up with examples? I'm curious to see some.
> A quick search turned up this extremely expensive EQ
> (http://www.yoursoundsource.com/mw-avalon-ad2077.html) but it doesn't
> mention tempco Q.

Most mastering EQs have discrete settings on Q, frequency, and gain,
made up with tempco or better yet temperature-unaffected circuits.

>>> One thing I'm really learning about synth building is that you can often
>>> get
>>> away with a lot less than you think you need. It's amazing, but the human
>>> ear really isn't "hi-fi" at all. I thought I needed 16-bit parameter
>>> values
>>> for everything, but 12 will do, and often 8 is enough.
>>
>> The way I think of this is that I can't figure out all ways that
>> people will be using my synth/eq/computer program/other invention,
>> therefore I shouldn't compromise where *I* think that *I* don't need
>> the accuracy. I usually go and ask experts, I'm definitely not the
>> best expert in the use of anything I created.
>
> I'm not talking so much about 'possible use' as the limits of human
> perception. Digital 'stepping' is the big evil that everyone is trying to
> avoid, and the question is "how many steps do you need before they become
> imperceptible?". This does depend on the situation (a slow sweep of a highly
> resonant filter is about the worst case scenario) but is a practical
> question that can be answered by experiment. Whilst some people's ears might
> be more sensitive than others, it's not a subjective point.
> A similar example is the question of how small a pitch difference is audible
> ("about 6 cents" is the accepted answer, with caveats). Obviously some
> people have a much better ear for pitch than others, but if your oscillator
> is producing discrete pitches every 0.01 cents, you don't need to waste time
> and effort trying to  "improve" it to 0.001 cents accuracy. For synth use
> "better than human hearing" is a useful day-to-day definition of "good
> enough".

You are judging the limits of human perception by your perception?
That's brave to say the least :-)

The limits of perception are, by definition, a subjective point. Our
subjective opinion is dictated by perception, and our perception
determines our subjective opinion; they are one and the same.

Pitch perception can be backed up with extensive research. Can you
back up your hardware design decisions with a similar amount of
research with large populations of geographically and otherwise
statistically various individuals?

Just to give you a practical example: at 10kHz, your accuracy of 0.01
cents can create a beat interval of 20 seconds, 9 seconds, and so on.
A 0.001 cent accuracy will be much finer and will allow creative usage
of the parameter at hand.

Every time you come up with a reason why you can't notice a
difference, someone else will be able to tell you why they absolutely
need it.

D.