I have noticed that when changing the noise floor threshold value the signal level in registration tab increases significantly.

1- Is this just a reading or does it ACTUALLY have effect as if TX power is increased?
2- Does it affect CCQ, data rate and latency?
3- Is it a best practice to change noise floor threshold on PtP links to achieve a better signal if the above is true?

I believe a good answer from Mikrotik staff would help all wireless users out there.
Thanks.

i think noise floor or snr says little about the real situation

the real and important measurement its signal vs interference, only can be measured with spectrum analyzer

i think noise floor or snr says little about the real situation

the real and important measurement its signal vs interference, only can be measured with spectrum analyzer

But you didn't answer my 3 questions.
Please I need a clear answer.

Thanks.

bump

I also noticed that increasing noise floor threshold value will increase the signal level.
SNR is also affected.
very strange

i think noise floor or snr says little about the real situation

the real and important measurement its signal vs interference, only can be measured with spectrum analyzer

But you didn't answer my 3 questions.
Please I need a clear answer.

Thanks.

We all would like a clear answer! But just maybe Mikrotik and others just work off application notes and modify to suit,
as a lot of technical questions like this go unanswered and the same on many other forums !

According to the wiki

This property is only effective for cards based on AR5211 chipset.

Maybe changing this setting just "alter" the calculations as in reported signal strength levels when using connect-list or access-list (Reported signal level is exponentially weighted moving average with smoothing factor 50%.)

Would like also to hear definitive word from Mikrotik!

Acording to the docs and other threads on the forums, this setting only affects Atheros 5211chipset radios.

It basically works like a squelch. Any signal received that is weaker than the threshold does not count as a signal.
(Imagine a police band scanner that is quiet except whenever someone is talking on the frequency. If there is someone very far away that you could hear, but with a lot of static, you can disable squelch to hear these transmissions, but with squelch on, you would miss them)

It should have no effect on your wireless link's signal strength, or SNR. I can see how it might improve ccq if the radio is only listening to stronger signals, but it shouldn't improve SNR.

SNR is your key statistic - you need around 20dbm of SNR in order to get full modulation rates.

I noticed this setting isn't available on AC chipsets, however it is present on N chipsets (atheros AR9342); don't know if winbox just "filters out" AC chipsets, or in fact any AR N chipset could be considered "AR5211 based".

Just tested this on a temporary RB912-5HPnD PTP link with borderline values. Whether this actually made a difference in the (AR9342) card operating parameters, or just altered calculated measurements, I don't know, though it "looks like".

If the threshold value is increased from the default setting (-110) to -105 we are "raising the bar", raising the threshold as ZeroByte pointed with the squelch analogy, thus effectively decreasing the Dynamic Range the radio amplifier has to work with.

The perceived signal strength decreases in the same dB, e.g. if -110 gets a -75dBm signal, -105 gets a -80dBm signal, thus signal decreases, guess due to the amplifier changing its gain to achieve this.

However, although signal strength was lower, and Signal To Noise lowered too from 40's to 20's (expected, as it's almost a 6dB change), CCQ increased and stabilized from 75%-95%'s to steady 95%'s, whereas effective bandwidth just decreased slightly from about 22Mbps to 20Mbps TCP.

Decreasing the threshold value (i.e. -110 to -120), i.e. increasing the Dynamic Range, will in fact report higher signal strengths, and Signal to Noise, at the cost of higher interference pickup (higher gain) and worse CCQ, so this would only be useful (I guess) in absolutely clean spectrum environments to extract the last ounce of performance from the link.

So... the proof is in the puddle, IMHO a TCP bandwidth test has the last word, we're working with disruptive technology chipsets here and should take measured SNR values (or any, for that matter) as a reference point only; as chechito pointed, accurate and reliable SNR measurement equipment cost several orders of magnitude more than any Atheros/Mikrotik based radio.

Definitely a word from Mikrotik would clear things out

I couldn't understand.

If this should adapt the "hearing" of the radio....
Lets pick the Basebox5 as example.

We have a -96 as limit for the sensitivity.
https://routerboard.com/RB912UAG-5HPnD-OUT

So, how putting -105 as threshold would change anything ?

If the objective was to decrease the value to be able to reduce the "listening" of signals weaker than -105....
how come the -105 would be relevant if -105 was never "hearable" in first place.

Sensitivity is not a limit is a radio characteristic, a measure of its maximum "hearing" abilities.

What this parameter impacts is the gain/dynamic range on the radio amplifier, making it to lower its "hearing" abilities so that less noise is heard: what you are providing here is a treshold value, which is used by RouterOS driver to autoadjust radio amplifier gain/dynamic range according to that threshold.

Has this topic got any action lately?

I've recently been doing some testing with this parameter using the 5GHz RB912 bd.

The question I posed to Mikrotik that was never answered, was "what is the numerical setting of 'default' ". Above, it's mentioned several times that the default setting is -110, however, if you enter -110, you get different results than if you enter default. This is based on the RSSI readings that are altered by this setting.

On the board I am using, I've found that if I set noise-floor-threshold to -115, I get similar RSSI readings as to when it's set to default.

Has this topic got any action lately?

I've recently been doing some testing with this parameter using the 5GHz RB912 bd.

The question I posed to Mikrotik that was never answered, was "what is the numerical setting of 'default' ". Above, it's mentioned several times that the default setting is -110, however, if you enter -110, you get different results than if you enter default. This is based on the RSSI readings that are altered by this setting.

On the board I am using, I've found that if I set noise-floor-threshold to -115, I get similar RSSI readings as to when it's set to default.

According to the Wiki

"By default, it is dynamically calculated"

https://wiki.mikrotik.com/wiki/Manual:I ... e/Wireless