Noise power is proportional to bandwidth

At room temperature, a resistor will develop -174dBm of thermal noise or -174dBm/Hz   In a 2.5kHz SSB bandwidth, the total power is therefore -174 + 10.LOG(2500) = -140dBm

When using an SDR + waterfall with 0.3Hz bins, the thermal noise per bin, and  the minimum possible sensitivity is -174 + 10.LOG(0.3) = = -179dBm.   Using QRSSS with a 0.01Hz bin, that becomes -194dBm

Using the bin size of an FFT as the bandwidth is not strictly the case as FFTs are windowed to increase dynamic range, which has the effect of a signal appearing in adjacent bins at reduced power.  So a good average to assume if you don't know the exact windowing function is that  bandwidth is twice the bin size.

Andy  G4JNT

On 22 January 2018 at 20:06, Martin Evans <[email protected]> wrote:

Alan - thanks for this.
Being a bear of very little brain ( I wouldn't recognise a Newton if I tripped over one!) much of your explanation goes over my head, but I understand well enough your conclusion that "that is not bad".

One thing that I don't understand is the relevance of bandwidth to an SDR receiver.

If I short my receiver input, my baseline stands at -140dB ( that's the lowest that the spectrum display goes); connect my antenna and my background noise level around 137kHz is around -120dB; DCF30 is around -70dB at the moment, and if I increase my scan width, R4 on 198kHz is -40dB. All these measurements are taken from the spectrum display in the Elad software.

I don't see where bandwidth comes into the picture(!!)

No doubt one of these days some of this will begin to make sense....

Thanks again,

Martin GW3UCJ.

********************

Hi Martin, I would estimate the field strength from DCF39 in your area is about 0.5mV/m (from measurements done by PA0SE in the UK) and from your data that gives about 1.5uV at the RX input as the level of the noise. If that is in an SSB bandwidth that is not bad. I remember the noise plotted on my CFH measurements in around 2002 was in this area but in a 300Hz bw, on the AOR7030 (Remembering WSPR s/n is related to a 2.3kHz bw if I remember right)
My minimum signal detection using a waterfall on the audio was 5nV (rx bw not relevant) but 0.3Hz bin size using a good sig-gen. I guess this was little more than 1 to 1.5dB above the noise. I think it was probably quieter here in East Anglia 15 years ago than it i now.

I hope that helps (saves you ripping your station apart :-))  )
Alan
G3NYK

----- Original Message ----- From: "Martin Evans" <[email protected]>
To: <[email protected]>
Sent: Monday, January 22, 2018 4:51 PM
Subject: Re: LF: 136kHz - this is how badmy system is!

Watching Roger's adventures on 136kHz, I thought I'd go take a look and I wondered -

DCF39 is about 50dB above the noise at 16:44UTC.

I'm using a 150ft inverted L, untuned, stuffed directly into an Elad FDM-S2 SDR.

Decoded G3XIZ, G4FTC, G8HUH & G4GIR between 16:12 and 16:44.

What does this say about my setup?

Is it deaf/average/ok?

How far above the noise is DCF39 in an optimised "good" setup?

Anyone tell me?

Martin GW3UCJ Swansea.