[sdiy] BODE Frequency Shifter

[Bernard Arthur Hutchins, Jr]

It is good that Atsushi Maeda found the difficultly to be only that of misidentifying a component.  90-degree PDNs are very INSENSITIVE to component tolerance variations (because they have only real-valued poles).   A moderate amount of care should assure a near-nominal performance – they just work – no tweaking.

An exception to this rosy picture occurs when we use an extremely large bandwidth to minimize end phase errors – say 5000:1.     The poles have a corresponding spread and so do the RC products of the individual sections.  In attempting to keep resistors in a range of perhaps 5.1k to 100k, capacitors may spread outside a corresponding desired spread range of perhaps 330 pfd to 0.05 microfarad (too small means stray capacitance matters, too large that caps are physically too large).  Happily I found some 30 years ago that by pairing the poles, largest to smallest, using 2dn-order secions nsead of first, the spread range is reduced to near unity.  See:

http://electronotes.netfirms.com/EN168-90degreePDN.PDF
LARGE SPREAD OF CAPACITOR VALUES; - Netfirms<http://electronotes.netfirms.com/EN168-90degreePDN.PDF>
electronotes.netfirms.com
THE DESIGN OF WIDEBAND ANALOG 90° PHASE DIFFERENCING NETWORKS WITHOUT /\ LARGE SPREAD OF CAPACITOR VALUES;-by Bernie Hutchins INTRODUCTION: The design of wideband 90° phase differencing networks is fundamental to the



This discussion along with several related entries is on the Electronotes site as misc. items.   As a bonus, you cut the number of op-amps  in half.  Only reasonable way to design these – or so it seems to me.

Bernie

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://synth-diy.org/pipermail/synth-diy/attachments/20181102/b8b4f7be/attachment.htm>