<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD>
<META content="text/html; charset=us-ascii" http-equiv=Content-Type>
<META name=GENERATOR content="MSHTML 8.00.6001.23562"></HEAD>
<BODY>
<DIV dir=ltr align=left><SPAN class=788334901-07032025>Hello
Spiros,</SPAN></DIV>
<DIV dir=ltr align=left><SPAN class=788334901-07032025></SPAN> </DIV>
<DIV dir=ltr align=left><SPAN class=788334901-07032025>Yes, I have only designed
and built triangle core VCOs before now, for the very reasons to which you
allude -- the "Dixie" and the "Rubicon" were the results of my efforts --
both of which track very well.</SPAN></DIV>
<DIV dir=ltr align=left><SPAN class=788334901-07032025></SPAN> </DIV>
<DIV dir=ltr align=left><SPAN class=788334901-07032025>The bidirectional current
source is very easy to achieve, and back-to-back zeners (or, in my case, a
"zener bridge") ensure accurate levels. Also, if you apply balanced AM to
the square wave, you can achieve through-zero FM.</SPAN></DIV><BR>
<DIV dir=ltr lang=en-us class=OutlookMessageHeader align=left>
<HR tabIndex=-1>
<FONT size=2 face=Tahoma><B>From:</B> Spiros Makris
[mailto:spirosmakris92@gmail.com] <BR><B>Sent:</B> Wednesday, March 05, 2025
10:57 PM<BR><B>To:</B> David G Dixon<BR><B>Cc:</B> Roman Sowa; synth-diy mailing
list<BR><B>Subject:</B> Re: [sdiy] Transistor questions<BR></FONT><BR></DIV>
<DIV></DIV>
<TABLE style="PADDING-LEFT: 0px; BORDER-COLLAPSE: collapse">
<TBODY>
<TR>
<TD><FONT
style="PADDING-BOTTOM: 3px; LINE-HEIGHT: 1.6; BACKGROUND-COLOR: #ffecb3; FONT-STYLE: normal; PADDING-LEFT: 3px; PADDING-RIGHT: 3px; FONT-FAMILY: Arial, sans-serif; COLOR: #000000; FONT-SIZE: 12px; FONT-WEIGHT: normal; PADDING-TOP: 3px">[<STRONG>CAUTION:</STRONG>
Non-UBC Email]</FONT></TD></TR></TBODY></TABLE>
<DIV dir=auto>
<DIV dir=ltr>An interesting observation in regards to the sawtooth core:
<DIV>Its operation has an inherent non ideality that can be optimised but never
eliminated (you will never switch "instantaneously" no matter what); the
apparent simplicity comes at the cost of balancing transistor capacitance,
conduction resistance, trace inductance, comparator hysterisis and delay
etc. The reset portion of the waveform will last a constant amount of time,
which eventually becomes comparable to the adjustable ramp portion, and that
messes your tracking, as well as the waveform itself (although I'd argue you
won't be able to hear that difference in texture).</DIV>
<DIV>In contrast, the triangle core oscillator does not have this inherently
unachievable feature, since the discharge/charge currents are not required to
ever become infinite. Of course, the comparator propagation delay, frequency
dependence of hysteresis, capacitor leakage will create a set of
constraints in operation. However, it is admittedly easier to make a very
fast accurate comparator than making an adequate reset circuit for a "wideband"
tracking sawtooth core. The cost you have to pay is a significantly more complex
bidirectional current source, which cannot be realised easily with discrete
components.</DIV></DIV></DIV></BODY></HTML>