# FFT on my oscilloscope is not detecting a peak which is a peak

I am having a strange problem and it's due to that lack of my info about signals.

I examined a PWM output from my MCU under sds1104x-e oscilloscope which is 656KHZ with a duty cycle of around 45.5% and converted the signal into FFT domain to examine the harmonics, RMS, and so on.

Anyway look at the first image, it represents Vrms versus frequency in the case of using DC coupling, as you can see there is a peak frequency at 15.25KHZ whose amplitude is higher than the amplitude at 656KHZ but still not detected as a peak point.

when I changed the coupling to AC, that first peak disappeared but the amplitude of the other 2 peaks increased.

I don't know why this happened, but hear me out. I am just trying to understand how to convert a PWM signal to an analog signal using a low pass filter and one of the things the tutorial looks at is the FFT of the signal, my problem is I concluded that the analog equivalent value of the PWM signal should be equal to the summation of Vrms of the composing signals, so my PWM signal shall produce a 1.5v analog signal, in first image the summation of the 2 Vrms of the signals is actually around 1.49v while in the second image, it's around 1.7v, so a lot of question and I really don't understand what's going on.

• is that at 15 kHz, or is that the DC peak of a signal with DC offset? I wouldn't be surprised if a spectrum analyzer ignored the DC peak when you use the peak-detect function. Oct 6, 2023 at 20:06
• @Hearth, I really don't know, it just shows a signal at 15KHZ with a high peak but not detected, when I re-run the oscilloscope again, it changes from 15KHZ to 2KHZ, so I guess it may be DC offset, I really don't know but what I know is the frequency of my PWM signal is 656KHZ, I don't know where that low frequency signal came from. Oct 6, 2023 at 20:12
• I assume the "15 kHz" at this frequency resolution is the first slot of the spectrum and represents the DC part of the signal.
– Jens
Oct 7, 2023 at 0:37