Why is FM restricted to 15khz when it can actually handle a lot more (as a mono signal)

I saw on here a few days somewhere that someone was running at 16khz without setting off the stereo light. Is that correct?

I can hear the difference between 15khz & 20khz. I’m actually debating running in stereo as iv never had issues with stereo hiss for years like it was in the 90s. Now days it just auto changes to mono if it’s weak.

You are exactly right, the audio on FM is rolled off at 15KHz to prevent interference with the stereo pilot. The 4KHz gap is to allow for the filter response to gradually roll off - if your filtering is excellent with a steep cut off then you can go higher. Running in mono you may notice some songs can set the stereo signal going with the high hats or cymbals - most of time that's all the audio content that's up there - and that's exactly the shit that'd screw with your stereo if the audio was unfiltered.

Would I be able to run at 18khz If I was to hard cut anything above 18khz off?

usually the receiver also has a better or worse 15KHz filter on the L/R output, so what you intended has no effect.

But if you also want to play with the demodulated MPX output of the receiver, then you can go far above the human ear!

XXL wrote: ↑Tue Dec 13, 2022 8:45 pm Would I be able to run at 18khz If I was to hard cut anything above 18khz off?

Theoretically, yes. However there's no filter that will work that steeply without other issues: The passband (the audio) has to be phase-accurate if the stereo image is to be effective, there must be no level "ripple" in the passband, and you need low distortion. All these criteria conspire to prevent you designing a filter that rolls off sufficiently between 18 and 19 kHz. The audio content needs to be at least 50dB down by 19 kHz (and a lower content than that is preferred). As usual, electronic design is a set of compromises!

Earlier designs for 15kHz lowpass filters used op-amps and high accuracy capacitors (expensive) and resistors. I used "gyrators" to emulate inductors, and got pretty good results. Other designers used a mixture of Chebyshev and Sallen and Key filters to achieve the same thing.
These days I use a switched capacitor filter with a crystal reference for my 15 kHz lowpass, and get a residual of just -62dB at 19kHz. The phase accuracy in the passband is excellent (up to at least 14 kHz), and the noise and distortion is vanishingly small.

The simple fact is that there is nothing of interest at those frequencies for the average listener!

Most "PRO" sound engineers will dramatically cut 16 to 20Khz frequencies in some cases to minimise distortion to boost wanted frequencies!

The same applies to the low end of frequencies often known as "Mud"

Low frequency inaudible "Rumbles" can severely distort the low end response of their "Masterpiece" as they would call it!

Anyone ever tried using a notch filter to cut 19khz instead of a low pass? You'd probably still need some kind of low pass to stop the mono and mpx frequencies interfering with each other, but you could get away with a much higher cut off frequency and/or a less sharp (cheaper) filter.

The good thing about a notch is you can get a pretty sharp response with a single opamp stage using the twin t circuit. Combine that with a single pole passive RC low pass and you could get similar performance to a circuit using 3 or 4 opamps for a fraction of the cost.

Yes, you can use a notch filter at 19khz but it will only work for mono transmissions to avoid triggering on the stereo decoder. In stereo, you need to continue supressing ultrasonics beyond 19khz up to at least 53khz which is where the L-R signal resides.
Anyway, a realistic notch filter will start to roll off at 15 or 16 khz, unless it has an unachievable Q factor.

sinus trouble wrote: ↑Wed Dec 14, 2022 12:13 am The simple fact is that there is nothing of interest at those frequencies for the average listener!

This is true, and also I suspect the majority of people who listen to FM these days are over 40 and less likely to be able to hear anything over 15khz anyway, particularly if they're ex-ravers who have abused their ear drums over the years. I include myself in this category, 44 and I can't hear a 15khz tone any more!

One other thing to consider, using these higher frequencies means a reduction in the perceived loudness of your audio - to most peoples ears anyway. It won't be a linear 20 - 25% you'd get in white noise, it'd be dependent on the program material - the more stuff is above 15K the less you have available for stuff people can actually hear. Personally I can't hear anything over 13KHz and that's in one ear, the other one is worse and has a whistle at about 4KHz and my heartbeat in it, too many gigs - but I can take my own blood pressure in a quiet room, which is handy for an old fella...

I would have to agree! I have abused my hearing over the years too!

Infact, now i find very low or very high frequencies irritating and sometimes nauseous!

Anyways, if nature intended humans to have a wider hearing range? I see no reason why multiplex signals could not be pushed higher into the frequency spectrum and still function correctly?

If you desire 16Khz? Most filters would likely still permit that? But you may need a little extra gain!

In experimental stereo coders, we pushed the pre-emphasised audio to 16kHz. The results were disappointing - it's really not worth the effort to get that extra audio bandwidth - It's not really possible to hear a difference between the two turn over frequencies.

Another question is 'why is deviation 75kHz?'

I read recently it's because 15 kHz x 5 = 75. Which would make some kind of sense other than the fact that stereo goes up to 53 kHz (58 with RDS).

Also, I would like to know if your right by your little FM TX and it plays mp3 so strong stereo signal would it be worth going from 128 mp3 to 192 mp3 because I have converted my mp3s to this on the sd cards. ?

The higher bit rate on MP3s gives lowered distortion. That's all.