From: Winfield Hill
Subject: Re: low distortion oscillator
Date: 9 Nov 2002 06:03:08 -0800
Organization: Rowland Institute
X-Newsreader: Direct Read News 4.10
Martin Griffith wrote...
> MIKE CONNELL scribbled:
>> There is a schematic for a cheap low distortion oscillator based on a
>> Wein bridge, page 30 figure 40 in http://www.linear.com/pdf/an43.pdf
>> The distortion is below .003% ...
> go for wire wound 10 turn pots, as Phil said, also look at this Walter
> Jung circuit, http://martingriffith.users.btopenworld.com/LDO.HTM
Good advice, Martin. Walter's approach, using inverting amplifiers
rather than the more common non-inverting form, is preferred for low
distortion, because it eliminates opamp input common-mode voltages.
He also uses a linearized FET level regulator (see AoE page 139).
For even lower distortion composite opamp stages can be used, e.g.
see Linear Technology Magazine, Feb 1994 for Dale Eagar's amazing
1ppb oscillator. Of course we don't all need 0.0000001% distortion
levels, achieved with his triple-opamp "Super opamp," but a simple
dual-opamp composite can perform wonders. Here's one that should
work well. (Note the first opamp's inverted polarity.)
. 3.3pF 10k
. __ ,--||--/\/\--,
. ---|+ \ 10k | __ |
. | >--/\/\--+--|- \ |
. ,--|-_/ | >-----+----
. | gnd --|+_/
. gnd OPA627 OPA134
First, the loop gain increased, especially useful at high oscillator
frequencies, where the opamp's GBW normally limits loop gain. The
composite stage above has 118dB of gain at 10kHz.
Second, the first opamp's input stage has lower distortion because its
output stage is driving a higher load impedance. This is an important
issue for input-stages at low frequencies, where distortion can arise
as a small thermally-induced change in offset voltage each cycle.
To design a composite amplifier suitable for use as an integrator, as
is needed with a Wien-bridge oscillator, simply choose the 2nd opamp's
gain-setting components to insure the composite has a 6dB/octave slope
at unity gain. For example, the 1st-stage OPA627 has a 16MHz GBW, so
I chose a 5MHz integrator bandwidth for the 2nd-stage OPA134. That
means the gain slope is -6dB/octave above 5MHz for good phase margin,
but increases to -12dB/octave below 5MHz for high loop gain at the
frequencies of interest.