using average oscillators for low distortion measurements

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
having had difficulty both with the sine voltage source of LTspice, and in finding a low distortion oscillator for prototype testing, i've designed a 1khz resonant low pass filter. in radio, the reason a class C amp can be used with the output showing no harmonic distortion, is because of resonant circuits and filters that are used to get a pure RF sine wave.

i was having problems with LTspice's oscillator, as it has distortion products only -60db down from the fundamental, which is not low enough for my purposes. i also plan on using my sound card as a signal source for testing, but also need a low distortion source for distortion testing of prototypes. since distortion tests are usually only done at a handful of fixed frequencies (similar to the radio world where devices are usually used at fixed frequencies), it's more cost effective to filter the oscillator output, than it is to buy an expensive oscillator (and hope it's still in calibration).

the filter consists of 5 series resonant sections made with 100mH inductors and capacitors selected according to frequency. the capacitors have odd values, but are more easily "built up" from smaller capacitors than would be the case with inductors.

since i generally use only 3 different frequencies, i only need to find 3 different capacitor values. i have simmed the 1khz filter, and even created a symbol for it, so i can use it in amp design tests as a 3 terminal device. it has a 5 pole chebyshev response, and a -100db/octave slope above cutoff. real world components probably won't perform the same, but will still be quite useful.
btw, you need to use resistive loading with the sim filters, unless you like starting at 1mv output and having your voltage climb to thousands of volts within a simulated second (it makes a great demo of how resonance works, though)

i found 250 ohms in series with the input, and a 1k load at the output to work ok, but you may want to measure a real world 100mH coil and use that resistance in series with each inductor in the sim version.

for 1khz, the cap value is .93uf, and after i finish this post, i'm going to finish the 100Hz and 20khz versions of this filter and report the cap values.
 

Attachments

  • 1kfilt.jpg
    1kfilt.jpg
    23.9 KB · Views: 220
Hi unclejed,

I am wondering if there is a special reason why you wouldn't use a sound card. 24 bit sound card generated sinewaves have a very low distortion. Usual THD+N is -100db or better.

And there are freeware programs that do that, and also generate a sweep tone if you want.

About LTspice I have no idea, but for prototyping it would seem a PC sound card would be ideal, no? Or am I missing something?
 
ok, values for 100hz are 93uf, and 20khz is 2.33nf

series resistance for 100mH is probably going to be about 200 ohms or more for each coil.

this probably can be done in the form of an active filter, but an op amp might introduce distortion in place of what has been filtered out.
 
i do want to use a sound card as a signal source, but i'm running an older ibm laptop, and i'm not sure what the sound card looks like for distortion..... as a matter of fact, this laptop has one of those plugs with a standard 1/8" stereo jack surrounded by an rca ground shell, and IIRC the rca shell has something to do with using microphones, or something...... looking it up now....
 
tnx tim...... i didn't know that........

still going to build filters for real live amps, though..... tested audio test bench and spectrum lab last night. while spectrum lab seems to be more versatile, it's oscillators are limited to 2khz, which is ok for 1khz distortion tests and 2 tone imd testing (900 & 1000 hz), i think for other tests (such as 20 khz distortion, slew rate, etc...) i'll use ATB. i still don't know what the quality of the sound card in this laptop is (ibm thinkpad 600e).....
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.