[sdiy] Odp: Re: SAW core VCO flyback time

[Ben Bradley]

I've designed but not yet built a triangle oscillator using a single
integrator and mux on the input. This uses two current outputs from
the exponential converter, one going to a current mirror, so there's
the same amount of current in both directions. The mux, driven by the
usual comparator with large hysteresis, switches the integrator input
between these two current sources. I'm not sure how a disconnected
current source will act when connected, but an extra switch can
connect the non-active current source to ground. This way the current
sources are always (except for switching times) connected to a virtual
or actual ground. Thinking another moment, a signal diode/1N4148
between each current source and ground, connected to forward conduct
when the mux switches off, may be a better solution.

Good symmetry of course depends on the current sources being as close
as possible.

What parts do people use for multiplexers? I recently got a Vishay
thru-hole DG613 from Digikey (DG613DJ-E3-ND), and then got an email
that the thru-hole part is on Last Time Buy. It's expensive at $3, but
it switches remarkably fast. It has odd power voltage requirements and
will only do about +/-10V, bu that's still better than the standard
CMOS series parts. I've since done more research on analog switches
running on +/-15V, and it appears the minimum cost is around $1.50 to
$2.


On Sat, Aug 27, 2016 at 7:20 PM, Roman <modular at go2.pl> wrote:
> Benefits:
> - there's no fixed time period of reset pulse involved, that affects HF
> tuning
> - ramp always starts at 0V, which is not the case at higher frequencies in
> mostly used saw core design
> - integrators can use slower opamps, easier to find with super low bias
> current, while buffer after mux can be as fast as you wish, like SR=2kV/us
> because high input bias is no issue here
> - waveform is cleaner because slower reset time, that does not cause
> integrator opamps to ring/overshoot (let's assume that final buffer part is
> done right and does not ring either)
> - fastest saw reset slope in history
> - and most of all it's such a cool idea to try!
>
> Roman
>
> Dnia 27 sierpnia 2016 23:31 Tom Wiltshire <tom at electricdruid.net>
> napisał(a):
>
> Ok, I get it. Thanks Roman.
>
> I guess you could do triangles the same way, but with one integrator doing a
> down slope instead. Mux would switch back and forth between the two slopes.
> The result would be one octave down, clearly. Not sure what the benefit over
> a one integrator triangle core would be exactly.
>
> Tom
>
> On 27 Aug 2016, at 21:00, Roman <modular at go2.pl> wrote:
>
> One ramp is going up, and when reaches maximum, the MUX switches to 2nd
> ramp, which then starts exactly at this moment.
> Meanwhile 1st ramp is slowly decaying by not-so-fast switch, let's say it
> takes a few microseconds, and is held at 0V untill the 2nd ramp reaches its
> maximum, and MUX switches back again to 1st ramp.
> So reset FETs are not driven by short pulses but out of phase square signal.
> MUX switches between ramps making the reset slope fast as lightning.
> I have this core somewhere in one of my notebooks, or maybe loose paper.
> Never built it hough as it was a bit overkill.
>
> Roman
>
> Dnia 27 sierpnia 2016 19:37 Tom Wiltshire <tom at electricdruid.net>
> napisał(a):
>
> How would this work, Rick? I'm not seeing it.
>
> Does it need the two ramps to be running at half the output rate so that you
> always switch in the middle of the ramp? And don't you just swap flyback
> time for switching time as the potential glitch?
>
> Thanks,
> Tom
>
> On 27 Aug 2016, at 17:49, Rick Jansen <rick.jansen at xs4all.nl> wrote:
>
> You could do a dual integrator and alternate their outputs: no flyback time
> at all.
>
> rick
>
>
> On 27/08/2016 18:27, Mattias Rickardsson wrote:
>
> On 26 August 2016 at 06:42, David G Dixon <dixon at mail.ubc.ca> wrote:
>> I stopped building sawcore VCOs because I didn't like flyback.  That's why
>> I
>> only build tricores now, and shape them into saws.  They sound better
>> anyway, and sync in a more interesting way.
>
> I'm not a big fan of them either, as far as tracking and waveshaping
> is concerned.
>
> But regarding the sound - I made a very bad (meaning slow flyback) saw
> core, and now people shout that it sounds so great and they want it in
> more designs.
>
> Alas, the ever so unrewarding nature of good engineering! ;-)
>
> /mr
> _______________________________________________
> Synth-diy mailing list
> Synth-diy at dropmix.xs4all.nl
> http://dropmix.xs4all.nl/mailman/listinfo/synth-diy
>
>
> _______________________________________________
> Synth-diy mailing list
> Synth-diy at dropmix.xs4all.nl
> http://dropmix.xs4all.nl/mailman/listinfo/synth-diy
>
>
> _______________________________________________
> Synth-diy mailing list
> Synth-diy at dropmix.xs4all.nl
> http://dropmix.xs4all.nl/mailman/listinfo/synth-diy
>
>
>
>
> _______________________________________________
> Synth-diy mailing list
> Synth-diy at dropmix.xs4all.nl
> http://dropmix.xs4all.nl/mailman/listinfo/synth-diy
>