0
\$\begingroup\$

I am experimenting with a square OCXO, dividing its output frequency to 1kHz using a 4017 synchronous counter and feeding the output to a sound card using Spectrum Lab software. I can then record it and pass it to another software to visualize the output.

I don't have a spectrum analyzer but my question is what can cause that noise around the 1kHz output frequency from the IC?

Here is the schematic of the circuit:

enter image description here

Here is the result of my recording of the signal:

enter image description here

Ideally it should be much less noise. I should not neither see the harmonic at 2 kHz, and on the spectrogram scale I would expect to see the 1kHz signal like this (red is the ideal output, blue is the real one.

enter image description here

Any idea on how this overall noise can be attenuated somehow? Can this be related to the noise of the OCXO itself?

edit - the OCXO is a square wave one, all ICs are with 100nF caps, and there is also a larger one for 5v DC. the output measured in Volts from the divider to the soundcard is 0.6 V and yes I am using a 10k trimmer before the sound card to fine tuning it. /edit

\$\endgroup\$
2
  • 1
    \$\begingroup\$ You are not showing the waveform of the signal. But obviously there is a 2 kHz component if the FFT says so. Based on the very zoomed-out waveform view, it looks darker in the bottom than on top so there must be some sort of asymmetry or clipping happening. Zoom in and show the data sheet of the OCXO. It may have sine wave output which is bad for divider, and if there's no bypass caps, it may worsen the jitter, spreading the peaks. I hope you are not directly feeding in a 5V square wave to a sound card input? \$\endgroup\$
    – Justme
    Commented Nov 16, 2023 at 7:53
  • \$\begingroup\$ the OCXO is a square wave one, all ICs are with 100nF caps, and there is also a larger one for 5v DC. the output measured in Volts from the divider to the soundcard is 0.6 V and yes I am using a 10k trimmer before the sound card to fine tuning it... \$\endgroup\$
    – Ken90
    Commented Nov 16, 2023 at 8:36

1 Answer 1

2
\$\begingroup\$

You are taking your output from pin 4 of the last 4017. This is a 10% duty cycle waveform, so you would expect a large amount of 2nd harmonic. In fact all harmonics would be present in large amounts until zeroes at the 10th and its multiples, but your spectrum only covers the first two. You need a square wave, so 50% duty cycle, to get zeros at even harmonics.

The floor of your blue trace goes down to about -80 dBFS, and -70 dBc, so I think you're seeing the skirts of your FFT window function. Hanning is a common default window as it has a narrow main lobe for good frequency selectivity, but it doesn't have a good noise floor. If you want to see lower, use a narrower window like Blackman Harris (4-term cosine), or a Gaussian truncated at 4 or 5 sigma.

Unless your setup is truly pathological, you can regard a crystal oscillator followed by a divider as perfect, especially at your frequencies. You are seeing the soundcard, its clock jitter and ADC noise, your FFT window, the decoupling on the divider.

\$\endgroup\$
2
  • \$\begingroup\$ Hi Neil, yes that is totally correct, but are you telling me that on others pins of these 4017 I should have a different duty cycle? Because If I remember I tested last time and they looks all like to having 10% Duty Cycle, every pin! I will check that again and update . \$\endgroup\$
    – Ken90
    Commented Nov 16, 2023 at 11:38
  • 1
    \$\begingroup\$ @Ken90 I was telling you there's a 10% duty cycle on the pin 4 that your schematic shows you using, which explains the 2nd harmonic on your spectrum. I was implying nothing about the other pins. Reading the 4017 datasheet, I am now telling you there's a 10% duty cycle on all the other outputs, so it's the wrong part to use if you want a 50% duty cycle. You might want to consider an HC390 for a /2/2/5/5 or an HC74 for a /2/2 if you want a 50% duty cycle output. \$\endgroup\$
    – Neil_UK
    Commented Nov 16, 2023 at 11:53

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.