[RE: VCOs]

[Harry Bissell]

Some addl thoughts... the ESD diodes shouldn't be a player in the inaccuracy
of the circuit because the operating point will be far away from the point
where they would be biased even a little bit. The circuit will run at 1/2 Vcc
plus or minus the hysteresis band. The very limited amplitude should keep
these errors almost constant. That doesn't mean it works, it means this isn't
the biggest problem. Can anyone think of a use for a sawtooth wave where we
have no control over amplitude, bot can control frequency. It's no sweep
generator... ;-) Harry Bissell

The problem is even worse than diode leakage. C-MOS schmitt-trigger
threshholds vary with manufacturer and aren't guranteed. Sawtooth amplitude
is proportional to the hysteresis. And the C-MOS gate has a very limited
ability to recharge the cap. especially with the series resistor and diode.
It's a cute circuit though.
	Thierry Rochebois has a VCO design on his website (don't know the
address but someone in this group will...) that generates a Triangle wave,
Charge Time = Discharge Time. I haven't tried it but it looks like the way
to avoid the problem. Any Comments on his design?   Harry

> -----Original Message-----
> From:	Martin Czech [SMTP:martin.czech at intermetall.de]
> Sent:	Friday, February 05, 1999 7:51 AM
> To:	synth-diy at mailhost.bpa.nl; MHELIN at tne01.ntc.nokia.com
> Subject:	Re: VCOs
> 
> > Hello,
> > 
> > Are there any VCOs that produce saw waveform by recharging the
> > cap instead of discharging it as usual? I mean something like
> > this: http://www.uta.fi/~helin/sch/s-vco.gif. Usually in
> > relaxation type ocillator the cap discharge time has to
> > be constant, discharge can't be detected by comparator,
> > but in this case the charging can be detected (here by the schmitt-
> > trigger, but comparators could be used as well). Are there any problems
> > with this kind of method (like temperature or supply dependency,
> > possible current spikes etc.).
> > 
> > -Mikko
> 
> Charging, discharging, where's the difference ?
> A saw requires a controlled slow slope and a very fast one, that 
> usually has less control (because of the short time).
> I like the above mentioned circuit, because it minimises dead time,
> this is one of the high frequency errors of a saw osc.
> I think this is possibly the fastest arrangement for discharge
> *SENSING* you can find. 
> 
> One should not forget that even CMOS is beeing used, the schmitt input
> impedance is dominated by the ESD-Protection junctions inside
> the chip. Leakage can be a topic there. Most manufacturers guarantee
> <1uA over temp. This is a whole lot. One should test the circuit for
> leakage
> therefore (just disconnect the expo converter and see what happens).
> If a good cap is used, the observed droop or rise makes it easy to
> calculate
> the leakage current. 
> 
> A shortcomming could be the diode parasitic resistance as well as the
> fact that the diode resistance as such varies wideley as the current
> through the element changes (e.g. ladder filter). Ie. I would expext a
> very fast charging in the beginning, but then when approaching the upper
> level the charging gets slower. However, the schmitt will switch off
> about 2.5V or so, far away from the usual 5V. It would be interesting,
> if the diode resistance has a LARGE influence on charge time.
> The voltage across the diode influences the switching behaviour,
> that may be undesired, eg. for sync.
> 
> Maybe a faster diode with better recovery time and also low leakage
> could be used.
> 
> I had similar plans (but now I dropped sawtooth ideas completely),
> but I wanted an active switch for charging. Leakage problems get worse
> for BJT , because of beta.
> 
> OTOH you might also like to have sync. In this case you'll need a second
> diode arrangement, I want to sync charge to a predetermined level,
> in this case an active switch, BJT JFET or MOS would be good.
> If this sync switch takes also normal charging, even better.
> 
> m.c.
> 
> 
> Since the diode is charging, the expo source has to discharge, you can
> use npn, that's good.


____________________________________________________________________
More than just email--Get your FREE Netscape WebMail account today at http://home.netscape.com/netcenter/mail