Distortion and Magic Resistor Ratios

[Grant Richter]

Hi All,

Just came off a very manic month.
Prototyped 9 new filter designs.
A only few keepers.

Most interesting was actually running the
numbers on "religious" filter designs.

Some background: Old analog computers
contain comparator circuits which flash an LED at
any excursion above or below + or - 10 volts.
A sensible design based on op-amp saturation
somewhere around the +/-15V supply rails.
TL074s will run to about +/-13.5 volt and no more.
(individual op-amps will do better)

Magic Numbers: The 100K/220 ohm input structure
for the CA3080 is practically religious. But simple
math shows a +22 DC mv excursion for a +10DCV input.
Or 44 mv p-p for a +/-10p-p for 20dB input.
All the classic filter designs (including ones carried
over to LM13x00 datasheets) assume max 10V p-p for
1% distortion, but actually 220 ohms should be 100 ohms
for full +/- 10 volts channel headroom at a maximum of 1% distortion.
(assuming the Bernie rule of max 10mv p-p into a CA3080
with a worst case of +/-20v p-p and no more than 1%THD).

It has been pointed out numerous times that newer devices offer
better signal to noise ratios. So, a few number determined by personal
listening tests.

CA3280: Still the most chic OTA. Use the datasheet value of 62K
for Id and 10V p-p into 10K ohms. 68K (who stocks a 62K?) pushes the
clipping down to + and - 8 volts. Conclusion: Special order 62K.
Use 1% 10Ks for improved THD. Should be a lot better for US$4 (Digikey).
Still worth the extra $ (buy lots of them so Harris doesn't make them
obsolete).

LM13x00: Magic number is 220K/1K for 100mv p-p at 1% distortion and 
Id at 1 ma (15K from +15v) into the current mirrors. Not as audiophile
as the CA3280, but nearby. Have pushed 1V p-p through at 10%THD
(22K instead of 220K), you might like it, sounds as distorted as a Moog
ladder.

Best Regards,

Grant

"So, the CIA decided, they wanted Roland dead,
that son-of-bitch Van Owen, blew off Roland's head"
					-Warren Zevon