after getting nerdsniped into the whole power line frequency monitoring thing by @x44203 I decided that instead of monitoring the same power network as anyone else I'd hook up this thing and see how stable it is.
freq: 50.009522391407444, avg:50.010218546801035, stddev:0.001176184258215776
well, it's more stable than the mains frequency at least, so that's something!
you can see the frequency drop for quite a while as the thing slowly warms up before it kinda settles. it's still kinda noisy though and there are some interesting bursts of noise.
here's a look at one of the bursts of noise. it looks kinda… regular? not sure what's going on. it also starts at exactly 5pm. coincidence?!
Nov 11, 2026, 11:50 AM@f4grx I know, right? It's even better with the cover off!
Nov 11, 2026, 5:12 PM
@slyka Looks like data transmission via a rounded off set of square waves?
@fundamental transmitting on my… tuning fork?
@slyka Oh I thought you saw those blips when looking at mains instead of your isolated setup.
@slyka What's your external time reference? And are you in range of a chiming bell?
@penguin42 no chiming bells around I think…
time reference is just the sample clock in my audio interface
@slyka @penguin42 how are you measuring the frequency? Fixed threshold, then look for pairs of crossings in the same direction to get the full cycle?
@synx508 @penguin42 the signal is downconverted to 0Hz as a complex signal, then the derivative of the phase is averaged over one second. it's basically an FM demodulator.
@slyka @penguin42 Interesting. I'm guessing it's a beat between whatever the frequency is and that 1s window, or something *like* that, so triggered when the frequency hits a certain value and you don't end up measuring the frequency because it wobbles. Could be very wrong though. It looks like "a maths" rather than something that is happening.
@synx508 @penguin42 yep, that's possible. everything aside from the averaging is completely continuous though, so I could just replace the averaging step with a 1Hz lowpass filter or something to eliminate that possibility
@slyka @penguin42 I ran into something similar when I was making a 19kHz pilot level meter in gnuradio, i'm stuck under a cat at the moment and can't reach the machine that's on but I think I solved it with a low pass filter instead of using averaging.
@synx508 @penguin42 yeah, the problem is that the frequency response of a non overlapping rectangular window is just… kinda terrible
@slyka if it's an audio interface have you thought about putting a different reference clock in on the other channel?
@penguin42 I did a measurement of the output frequency of a function generator and got a standard deviation of around 3*10^-6Hz over the course of a few minutes, so I trust the clock in my audio interface. It's a decent studio interface and it should be a pretty good reference clock.
@slyka How well is it insulated from noise and vibration?
Could be some kind of motor or pump on a timer.
Nov 11, 2026, 6:41 PMNov 11, 2026, 6:53 PMNov 11, 2026, 7:06 PMNov 11, 2026, 7:08 PM@f4grx @marshray well, it's probably not what you expect…
(I was running it at a lower supply voltage here, at full supply voltage the output peaks at around 350V)
For the measurement I just divided that down, plugged it into my mixer through some capacitive coupling and turned down the highs and mids and boosted the bass on the EQ. I used a metrology grade EQ, of course.
@1000millimeter @f4grx @marshray using 60 year old electrolytic capacitors. it looked clean when I checked the the thing out, but since it's all class A amplifiers inside this thing the PSRR is basically 0
Nov 11, 2026, 6:39 PM@slyka This kind of waveform is typical of a chaotic behaviour, can be caused by non linearities, for example from overdrive and departure from linear harmonic approximations. Not all oscillators are well behaved.
Edit: I did not notice it was frequency variations. Then it is even stranger.
Nov 11, 2026, 8:20 PM@slyka That looks as if something in there would be oscillating mechanically in there.
I have experimented with the mains frequency and it helps to view the data with Audacity. It allows you to do a spectrogram which might be more revealing.
@casandro yeah, I might add something to automatically save chunks of incoming data to disk if it strays too far outside the standard deviation. might be interesting for glitches in the mains frequency too
Nov 11, 2026, 8:28 PM@slyka Well on the mains frequency I've experimented a bit, and there's some stuff that's easy to see.
For example you can easily see the 15 minute oscillations because of the electricity market. However one can also see much faster oscillations in the order of 1Hz probably caused by oscillations in the European power grid. After all thanks to all those rotating masses, the "speed of light" in the grid is very low, so you get a high group delay. I can get you preliminary data.
@slyka wondering if what's happening here is grid load/supply switching and a brief instability as supply comes online. Specifically that possibly the grid coupling goes like this:
This is sort of bullshit based on conjecture though and I'm not even sure all parts of that theory align?
Experiment:
Use another channel on the input to track grid frequency so that you can capture both with the exact same time reference and you can measure the two together.
(You can also try correlating against public grid frequency data but as this can theoretically dox someone I would be careful about posting too much detail)
@Lunaphied yeah, might give that a try along with all the other stuff suggested in the thread. I don't think it's locking to the grid since the grid is way more chaotic and all over the place, but maybe there is some beat frequency stuff happening if the two frequencies get close?
Nov 11, 2026, 9:55 PM@slyka I wonder if you are seeing zellweger / ripple control signalling coming through. Not sure how your device works or if you have the in your area
@xssfox hmm, good question. might have to do a recording of both the oscillator and the mains and see if there is a correlation between these oscillations and *something* happening on the mains.
The oscillator just runs off of an unregulated 15V supply, so it's not impossible for stuff from the mains to have an effect, especially cause they're so close in frequency
@slyka what if you shout at it
@ferrix my neighbors knock on my door and ask if everything is alright
Nov 11, 2026, 6:42 PM@slyka whats giving you this accurate measurement?
@f4grx most of those decimals are a lie, but I'm just doing direct frequency measurement by turning the signal into an IQ signal, differentiating the phase and averaging that over one second. when hooking my function generator to my sound card I get a standard deviation of around 3*10^-6Hz over the span of 5 minutes or so.
@penguin42 yes, they really knew how to polish a knob back then!