NBFM feature request


I miss the NBFM quick jump feature from default to narrower like it does for USB/USN - LSB/LSN . I ran the Kiwi to the 2m band and with distant weak signals the jump from 12kHz to 6kHz significantly improves listening comfort. An additional module could be marked as NBFM/NNFM or 6FM/12FM to also fit on the control panel. Convenient especially on mobile devices.



  • Yes I was thinking about this too, and it's a good suggestion, especially now that conditions on the upper HF bands are improving, where NBFM tends to be used a lot.

    The incorrect De-emphasis value for NBFM is still a bit annoying though.



  • jksjks
    edited January 2023

    I could add that pretty easily. Is 6 kHz the right default value? (default is of course configurable on the admin/config tab).

  • Probably not nearly as easy an addition but <300 Hz post-detect NBFM roll-off to take out repeater tone would also help listening quality.

  • If it is configurable, the default 6kHz should be ok. Generally, in amateur bands, narrow FM (marked as NFM/NBFM) has a width of 2.5kHz (FM-LINK repeater networks in the 70cm band) and wide FM (WFM/WBFM) has a typical deviation of 12kHz. If that's not too much of a problem, you could create 3x NFM-MFM-WFM narrow-medium-wide buttons. Am I the only one who thought of the popular ESSB by the way? ;) too much luck at once

    Today was another day with elevated conditions in the 10m band and I was receiving on my stick, over 6700 km from me New York KQ2H repeater, with a signal of about s7 -90dBm its 146.2Hz subcarrier is not audible, but some amateur stations have 127.3Hz switched on by default and this can actually be heard.


  • Ah no, please don't add that without being able to tuRn it off.

    It's often the only way to identify some signal from their tone frequency.

    It would be nice if there was tone and DCS detection too, but that's asking for too much.

    Ideally the bandwidths would be 20kHz for 2.5 spaced channels, 12 for 12.5kHz spaced channels and maybe 6kHz for weak signals. But as the KiWi can't do 20KHz, unless running in 3 ch mode, I guess 12kHz and 6kHz would probably be the most useful.



  • Whilst on the subject, I thought I'd investigate the 'rough' sounding NBFM demodulation that the KiWi produces.

    I made some comparisons between the NBFM audio output from an Icon IC7000 amateur transceiver just tuned to a quiet part of the 10m band with the squelch open, and a similar level of audio from a KiWi.

    First the Icom



    And now the KiWi with de-emphasis on. Note the large spikes.



    The de-emphasis is having some effect, but the slope is not quite correct.

    De-emphasis off.


    Spikes still present but much larger.


    Note the flatter spectrum.

    I'm not sure if the high amplitude spikes are due to the incorrect de-emphasis curve, or lack of amplitude limiting / clipping that would probably be present in a dedicated NBFM receiver.



  • edited January 2023

    With a flat frequency response (i.e. no de-emphasis), IIRR white noise into an FM demodulator should show a triangular distribution on the output, i.e. output rising with frequency by 6dB per octave.

    Among other things, I use Kiwis in the FM mode to examine the modulation on FMCW radars. This is typically a sawtooth waveform and the frequencies can be as low as 0.1 Hz. So a flat response from DC upwards is important for me.

  • Hi Richard,

    Good point about the spectrum without de-emphasis. I'd forgotten that bit of theory.

    It does suggest that there may be something wrong.

    Incidentally, I've also been looking at radars using the KiWi on the lower VHF bands.

    These links may be of use to you and others.

    Soundcard Oscilloscope (free) is a handy tool for looking at both the PRF using cursors on the oscilloscope and also for observing and snap -shotting the sweep patterns. The FFT spectrum analyser can also help quickly ID strong spectral components in digital signals.


    Another useful tool is KG Tone (free) which will decode CTCSS and DCS tones.

    Webpage in Japanese


    Download link




  • I need access to a Kiwi with reliable NBFM reception. Preferably with an open ssh port. Please email URL, port and password info to support@kiwisdr.com Thanks

  • from the wiki:

    The most common FM transmitting applications use peak deviations of +/-75 kHz (100 or 200 kHz spacing), +/-5 kHz (15–25 kHz spacing), +/-2.5 kHz (3.75-12.5 kHz spacing), and +/-2 kHz (8.33 kHz spacing, 7.5 kHz spacing, 6.25 kHz spacing or 5 kHz spacing).

    Which is considered NBFM these days?

    When I was in the Land Mobile part of the radio biz, we'd called the +/-2.5 KHz deviation NBFM

  • The use of the terms WBFM and NBFM are purely a "short hand" method of distinguishing between two general groups of modulation deviations and the pre/de-emphasis curves that are applied.

    I'd suggest that anything broadcast related (including STL links) with a deviation of typically greater than 50kHz, and using a 50uS, 75uS or 100uS pre/de-emphasis curve would be considered to be WBFM

    Anything communication related tends to use NBFM (e.g. amateur radio and CB on the HF bands and also commercial 2 way radio on the higher bands) with a deviation of under 50KHz (old channel spacing) and uses 6dB/oct roll off above 300Hz, which would probably equate to a pre/de-emphasis value of somewhere around 500uS or 750uS if it was expressed in that manner.

    The KiWi legacy de-emphasis values seem to be a vestige from OpenWebRX which support WBFM Broadcast reception with much wider receive bandwidths (about 150kHz) in addition to the NBFM modes.

    If you search the forum there has been a lot of previous discussion about this topic, without any meaningful real investigations being conducted. But now there is better propagation and more activity on the upper HF bands, and NBFM is widely used, e.g. 10m amateur repeaters, then some of these long standing issues have become much more apparent.



  • My ham experience with FM dates back to when hams used things like the Motorola 41V and deviation was 15 KHz ! :-)

  • I think the high amplitude short duration noise spikes could also be the reason that the KiWi's NBFM squelch is so 'choppy'.

    It seems to require a much larger value to be set beyond the initial point of muting, before it stops opening up every few seconds.

    For example it may start to mute at 20, but it takes a setting of 35 to fully stop the occasional 'chuffing'.

    In addition there is often some form of 'hysteresis' incorporated. This is to help prevent signal from 'chopping' when they drop in an out of the squelch threshold, so that when the squelch has opened it remains open of a short duration, in order to help reduce the annoyance when listening to rapidly fading signals.



  • I have a fix ready. But it needs to be tested against real signals which I don't have reception of here.

    So no software release with these changes until I can test them..

  • This is more complicated than I thought. There are two separate passband concepts here. The Kiwi passband that you see in yellow/green on the frequency scale, and in the waterfall etc, is applied to the FM signal with its corresponding deviation. This is the signal going into the FM demod.

    This is entirely separate from the output of the FM demod and what kind of audio characteristics you want to hear, including passband and de-emphasis.

    So when you say you want an 6 kHz "NNFM" mode in addition to the existing 12 kHz NBFM mode what are you saying exactly? You want a narrower audio response to reduce fatigue etc? Because that would involve modifying the demod output and not changing the input passband (which would otherwise limit the deviation going into the demod).

    Or are you saying there is another class of FM signals with smaller deviation that should really have a small demod input passband to limit the noise going into the demod?

  • Ironically I was just looking at the newest GNUradio and saw this which might help.

  • The "clicks" (spikes, roughness or "crackling" sounds) are caused by not having a clipper after the demod. Both CSDR and SDRAngel have this. The Kiwi uses the CSDR demod and I failed to notice the clipper the first time I looked at the CSDR code.

    PA3FWM (WebSDR author) has an excellent description of why the clicks occur here: https://www.pa3fwm.nl/technotes/tn24-fm-noise.html

  • edited February 2023

    Hi John,

    Thanks again for taking a look at this, as it's been bothering me for a long time.

    I think that all that was originally asked for was the ability to quickly toggle between an RF bandwidth or 12kHz and 6kHz, so that it is easy to swap between them when listening to stations using 5 or 2.5kHz deviation.

    It would be nice if the audio level could be brought up slightly to compensate for demodulation of the narrower deviation, but I don't think this is essential.

    I suspected the lack of clipping, but wasn't sure if or how it was implemented in DSP.

    I have sent you an email with sdr logins etc. as requested, and have temporarily setup a switched VHF down-convertor and a local nbfm signal source for you to play with, but you could also listen to the local 27MHz CB'er' or the 29MHz amateur repeaters who use NBFM too. KQ2H in New York was quite strong in the UK today.



  • Okay, so NBFM/NNFM should control the existing Kiwi passband filter which for FM is a filter on the FM demod input. That's what I have now. I can work on adding a correct de-emphasis filter next.

    I found the 400/1k Hz FM mod function on my sig gen, which I have never used. That was very useful in helping me to understand some things.

  • jksjks
    edited February 2023

    @G8JNJ No email yet Martin. Please try again. Thank you.

    Edit: got it. Thanks.

  • OK I've forwarded the original emails..

    If your generator will accept an external AF modulation input, you can use the audio generator in Soundcard Oscilloscope.

    This can also generate audio sweeps, which may be useful in conjunction with the scope itself.



  • I'm still working on the de-emphasis filtering.

    I can now do FIR filter design in Octave (MATLAB). Something I couldn't do before and never really understood.

  • Hi John,

    Thnaks again fo doing this, you wouldn't believe how much of a pleasure it now is monitoring NBFM stations with the KiWi. In fact I'm considering permanently adding the 2m convertor (and possibly others) to my home KiWi.

    I don't think you need to massively over engineer the de-emphasis, as long as it is a reasonable approximation it should be fine. Even the current set of values is of some use.

    My original thought was to locate the existing filter coefficients, and simply increase them by a factor of ten.

    This would be easy in the analogue world, but these digital "whatnots" are a mystery to me.



Sign In or Register to comment.