[sdiy] VCO reset time

[Magnus Danielson]

From: harrybissell <harrybissell at prodigy.net>
Subject: [sdiy] VCO reset time
Date: Tue, 01 Jun 2004 23:27:32 -0400
Message-ID: <40BD4924.8614A1E1 at prodigy.net>

> Hiya kids

Hi toy-boy!

> I'm playing with a new experimental VCO core.

Can we play with you?

> I will NOT tell you what it is...

C'mon!

> however if it has merit I'm going to publish it in EDN Design Ideas (it
> is cool enough that it will probably get accepted - after all they publish
> some really marginal stuff sometimes...)
> Of course if someone offers me the same $100 I'll probably talk :^P
> (you will need a lot more to get me NOT to talk)

So, if we meet I offer you a beer... so start singing boy! ;O)

> Several questions...
> 
> 1)  What is a good ramp reset time ?   How fast have you seen ramp reset
> be ?

I've seen ~736 ns:
http://rubidium.dyndns.org/~magnus/synths/friends/stopp/tek00016.png

This could have been improved with a little will and not a major rebuilt.
Oh, the above is the ASM-1 as I built it and modified the reset timing to get
it up and working. The transistor Rds(ON) is a limiting factor on the reset-
time obviously. Too high Rds(ON) for a reset-time and your oscilator fails to
oscillate. It boils down to choice of transistors and nothing else.

> 2) How many octaves do you expect (in tune) without resorting to HF
> compensation

These are heavilly connected IMHO. The less ramp reset time the less HF
compensation. If you do a good job and one of them then you can reduce or
totally remove the other.

However, the HF compensation of series-resistor type is needed not because you
reset the capacitor as such, but in the way it is performed, by shorting it
with a transistor. This has the unfortunate effect of temporarilly stop the
capacitor charging which at first seems like a good idea, but it is not, since
now the effective time the capacitor is not charged creates the cycle time
error which naturally requires compensation and the resistor is able to
seemingly "precharge" (which it doesn't, it just looks like it) the capacitor
just enought to compensate for the missing charge.

Now, what you could do as an alternative is to time a charge-dump into the
integrator. The charge shall be such that U = C * Q makes the capacitor jump
the /|U is needs to "reset" and C is naturally the capacitor capacitance.
The comparator would trigger the charge-dump and a current is pulled for some
little time. The best way to acheive this would be to have another comparator
voltage at the bottom such that the current is turned off when it has hit
"rock bottoms". This sounds like a double comparator for a RS-flip-flop or a
comparator with hysteresis. Such an oscillator would not require HF
compensation. Comparator-speed, integrator speed, current-source speed and
current-source current forms the limits for the sharp edge.
I can also come up with a handfull of different variants but I'm sure the
above proposal is a fair one.

> (I have not tried hf comp yet...)
> 
> Because I don't want to prejudice anyone... I won't say how fast my ramp
> reset is... but I will after I get some opinions.  I don't know how many
> octaves I'm getting ... I need to instrument differently. I'm thinking I
> mught do a lissajous figure and a binard divider to tell how well I'm
> doubling octaves

Ah well:
http://rubidium.dyndns.org/~magnus/synths/friends/stopp/asm1vco.html

By that time we didn't have a propper frequency counter. Now I have _much_
better capabilities both at work and home both in resolution and accuracy.

> Gentlemen (and ladies ;^) start your opinions !!!

I have an opinion, OK? ;O)

Yes, we *can* do much better if we want and need to. Question is how far we
need to take it. For musical purposes I'm not sure we need to take it much
further, but there is always a fight for perfection. I consider triangular
cores a very interesting concept and they might compete very well with
sawtooth cores IMHO. When you do a triangular core, you create the sawtooth
only to be consumed as a sawtooth and possibly a sawtooth-based PWM (as a
distinct thing from the triangular based PWM which is a different animal).

Cheers,
Magnus