Low-distortion Audio-range Oscillator

3 oscillator post filters, each with 4 or 2 + 2 inductors, so cw/ccw connection is possible.

The low pass is the steepest but has all different inductor values and inductors are in the direct signal path. Also hum goes through.

The low and band pass combination filters hum but still mostly different inductor values, 2 of them in the signal path.

The band pass has only resistors in the direct signal path, inductors all have the same value, the filter can be scaled by adding or omitting R-LC segments, it filters hum, is steep enough.
I guess this is still thee best.
 

Attachments

  • ANALOG_FILTER_125_OHM_7TH_ODER_X_1.JPG
    ANALOG_FILTER_125_OHM_7TH_ODER_X_1.JPG
    141.3 KB · Views: 520
3 oscillator post filters, each with 4 or 2 + 2 inductors, so cw/ccw connection is possible.

The low pass is the steepest but has all different inductor values and inductors are in the direct signal path. Also hum goes through.

The low and band pass combination filters hum but still mostly different inductor values, 2 of them in the signal path.

The band pass has only resistors in the direct signal path, inductors all have the same value, the filter can be scaled by adding or omitting R-LC segments, it filters hum, is steep enough.
I guess this is still thee best.

What is the input impedance as a function of frequency for these three filters? The existing reference for low distortion seems to be Victor's oscillator. It has an approximate output impedance of 200 ohms. Also, the distortion of that oscillator and many others begins to degrade for load resistance in the 1k ohm to 10k ohm region. Finally, if these filters require a precision source impedance, that would interfere with their use with different oscillators unless an active buffer amplifier is put in front.

Cheers,
Bob
 
Tried REW with a few old soundcards, the best so far M-Audio 24/96

My analyzer´s output without and with the simple 2-inductor filter.

The unshielded filter picks up a lot of hum but the area of interest stays clean and

the filter clearly adds a zero to the THD number.

Input adjusted to equal levels.
 

Attachments

  • M-Audio-14dB_ohne _Filter.jpg
    M-Audio-14dB_ohne _Filter.jpg
    309.4 KB · Views: 383
  • M-Audio-14dB_mit_ Filter.jpg
    M-Audio-14dB_mit_ Filter.jpg
    317.1 KB · Views: 382
Last edited:
For me, 48 kHz shows best results, I put a 2mm aluminum plate on top of both coils, this reduces inductor values and bandpass moves to the side a bit, no negative impact on distortion but also does not help with shielding.
 

Attachments

  • aluminum.jpg
    aluminum.jpg
    317.1 KB · Views: 355
What is the input impedance as a function of frequency for these three filters? The existing reference for low distortion seems to be Victor's oscillator. It has an approximate output impedance of 200 ohms. Also, the distortion of that oscillator and many others begins to degrade for load resistance in the 1k ohm to 10k ohm region. Finally, if these filters require a precision source impedance, that would interfere with their use with different oscillators unless an active buffer amplifier is put in front.

Cheers,
Bob

The bandpass can have any impedance, just higher impedance means higher loss. But as I want to measure power amplifiers, the gain will compensate for it.

If oscillator performance degrades a bit or it is necessary to add a uld buffer, I think it is no problem, afterwards the post filter will reduce harmonics by tens of dBs.
 
If you meant the balanced output of the serial 1kHz board like this (schematic of the old version):
https://content32-foto.inbox.lv/albu...ed/1kHzBal.jpg
Then no noise cancellation. Noise is little higher and also may distort little more (some dB).

Vic.

The picture is empty ?

I mean like in my attached schematic.
You have 1 gain device ?
2 gain devices back to back with the "tank" in the middle.
Maybe not possible.
 
Thanks, now works.

I don´t know much about oscillators and my schematic is no oscillator, just to demonstrate...

2 opamps coupled together, oscillating out of phase.
Symmetrical circuit like my crystal oscillator.

Maybe not possible because 2 feedback networks are necessary and values in feedback path can not be 100% matched.
 

Attachments

  • 00sy.jpg
    00sy.jpg
    59.2 KB · Views: 479
small news

Now I made a 3.1 mH inductor, wound on a HQ audio transformer core incl. MU shielding.

With the coil inserted in the signal path after my 16 bit generator / simple bandpass filter, the distortion is equal to "no coil". See two screenshots.

However... Measured with my handheld Multimeter the coil Rs is near zero but with my RLC meter it is 10 ohms

I don´t understand. 1 layer with 0.8mm wire, about 20 windings.

If I put that inductor in my filter will it behave like a 10 ohms coil ???
 

Attachments

  • 0mitSpule.jpg
    0mitSpule.jpg
    265.7 KB · Views: 423
  • 0ohneSpule.jpg
    0ohneSpule.jpg
    265.6 KB · Views: 424
Last edited: