Receiver Sensitivity -- Do I convert to the typical channel bandwidth?

[Sam Jester]

Greetings,

I am attempting to do some RF analysis for interference and was curious how I apply receiver sensitivity.

I have a receiver that was given a minimum sensitivity of -83 dBm and the typical IF channel bandwidth is 450KHz. In this scenario, what does -83 dBm represent? Normally when I see this I just assume "dBm/Hz" but was told I should be applying the bandwidth to get the real receiver sensitivity to Hz (-83 dBm - (10*LOG10(450e3/1) == -140 dBm/Hz)

Is this correct?

The issue I am having when applying this however is that there are other receivers (specifically GPS L1) with a given receiver sensitivity of -130 dBm and a IF channel bandwidth of 2.046MHz (it actually gave 20 MHz however I see this as an error as it should've been a channel bandwidth), as this equates to -193 dBm/Hz receiver sensitivity there has to be something wrong.



Main goal: I want to put receiver sensitivities in dBm/Hz for easy comparison. As I am doing RF co-channel interference I would like to evaluate my output power (in dBm/Hz) to the receiver sensitivity and see if I am above or below the sensitivity of the "victim" receiver.

Thank you!

 

[chuckey]

I think the important words are " receiver sensitivity". i.e. the Rf amplifiers/If amplifiers and DEMODULATOR. So once you have set a limit for a "good" receiver in terms of a BER of 10 -7 or 10 db S/N ratio for a certain amount of AM or FM, then you can test to see what input is required to achieve this is. I suspect it is implicit in the above statement the input sensitivity is -83 dBm, but it should be specified. Else you can just use noise factor, which takes into account the IF bandwidth etc. but does not relate to what sort of IF bandwidth you require.
Frank

 

[Sam Jester]

I think the important words are " receiver sensitivity". i.e. the Rf amplifiers/If amplifiers and DEMODULATOR. So once you have set a limit for a "good" receiver in terms of a BER of 10 -7 or 10 db S/N ratio for a certain amount of AM or FM, then you can test to see what input is required to achieve this is. I suspect it is implicit in the above statement the input sensitivity is -83 dBm, but it should be specified. Else you can just use noise factor, which takes into account the IF bandwidth etc. but does not relate to what sort of IF bandwidth you require.
Frank

Hi Frank,

Thank you for the reply. Just so I am understanding correctly Noise Factor would be the input sensitivity (-83 dBm) into account of IF bandwidth (in this case 450KHz, so thus the noise factor would be -83 - 10*LOG10(450e3)?

Am I to assume then that receiver sensitivity is normally specified in dBm/Hz (or is this distinction even relevant?). Apologies but I am just not familiar with how receiver sensitivity is derived nor what units it is supposed to have so it is probably a very basic misunderstanding.

 

[davenn]

receiver sensitivity is commonly specified in uV ( microVolts) at a 10dB S/N ratio, 0.1 to 0.4 uV is a reasonably sensitive receiver
tho just plain dBm is also common so a FM receiver with 12dB SINAD, -110 to -120 dBm is quite sensitive for general purpose communications receivers

your -83 dBm receive spec. is relatively insensitive

dunno if that helps ?

cheers
Dave

 

[Sam Jester]

receiver sensitivity is commonly specified in uV ( microVolts) at a 10dB S/N ratio, 0.1 to 0.4 uV is a reasonably sensitive receiver
tho just plain dBm is also common so a FM receiver with 12dB SINAD, -110 to -120 dBm is quite sensitive for general purpose communications receivers

your -83 dBm receive spec. is relatively insensitive

dunno if that helps ?

cheers
Dave

Thank you for the response Dave,

It does help me get a better understanding of receiver sensitivity and how it applies.

I think I have discovered how this -193 dBm/Hz makes sense when converting -130dBm on a 2.046MHz channel, what's added on, but not implied, is that there's a coding gain that is added in. As what I looking at is specifications for co-channel interference to a receiver it makes sense to include coding gain in this metric of "receiver threshold" for victim receivers. For GPS L1 I have seen that coding gain can be up to 45 dB!!

So -83dBm would be -83 dBm/450KHz == 140 dBm/Hz receiver threshold.

Thank you for the responses I hope this thread can be used for future for questions that might be similar.

Best regards

 

[vfone]

For co-channel test you have to let the system sensitivity in dBm/IF-BW, in your case dBm/450kHz.
Have to remember that in receivers used for digital modulation the co-channel test is not an RF or a receiver test (front-end/IF/demodulator), compared to other RX tests as: adjacent channel rejection, blocking test, intermodulation test, BER/PER sensitivity, etc.

In this situation the co-channel test is verifying only the ability of the BB/DSP processor to decode a signal in presence of an interferer situated on the same channel with the desired signal.

The receiver front-end, IF filters, IF amplifiers, etc, cannot help you at all to improve co-channel performance, because the desired input level is well above the sensitivity level but also far away from compression.

 

[jiripolivka]

Receiver sensitivity is a wrong concept to compare receivers.
The correct approach is to define the thermal noise floor from Pn = kTBN.
The easy to use formula is for Pn in dBm : Pn = -174+ F +10 log B, where F is noise figure in dB and B the RF bandwidth in Hz.

Then one needs to define the input S/N ratio, the signal power above the Pn as defined earlier. S/N is again in dB and differs for various modulation methods for the received signal.

Note that S/N is defined a receiver input, and the investigated receiver must be LINEAR. No detectors, limiters or demodulators allowed.

After a detector, limiter or demodulator is connected, the "input" S/N ratio is changed by non-linear effects. Sometimes to one's advantage: FM can have the "demodulation gain", for wideband FM it can be ~18 dB, so the FM receiver with input S/N of 20 dB can achieve the "output" S/N of >38 dB. PSK and other modulation types also have similar advantages.
For digital communication, S/N is converted to Eb/N ratio or into the BER. Bit error rate can be calculated but in real systems it significantly deviates from calculations. Correction coding can help, too.