Pages: [1]   Go Down

Author Topic: Measuring local noise floor with an SDR  (Read 390 times)

AK4YA

  • Member
  • Posts: 196
Measuring local noise floor with an SDR
« on: November 03, 2022, 03:06:41 PM »

How do I know what to trust here?  I have a fairly reputable SDR a QS1R and using its SDRMAX-V software I see a sideable difference on the noise floor of any band based on the sampling rate I pick.  If I pick a wide sampling rate I may get -80db for the noise floor and if I pick a narrower one I'll see -100db for the noise floor.

Is this normal?

And then if I use HDSDR software I'll get 2 completely different set of numbers.  Not sure what to think here.
Logged

KD7RDZI2

  • Member
  • Posts: 689
Re: Measuring local noise floor with an SDR
« Reply #1 on: November 03, 2022, 03:54:52 PM »

How do I know what to trust here?  I have a fairly reputable SDR a QS1R and using its SDRMAX-V software I see a sideable difference on the noise floor of any band based on the sampling rate I pick.  If I pick a wide sampling rate I may get -80db for the noise floor and if I pick a narrower one I'll see -100db for the noise floor.

Is this normal?

And then if I use HDSDR software I'll get 2 completely different set of numbers.  Not sure what to think here.
I think It is normal. I use other software and it behaves similarly. There are theorems behind that, which I leave others to understand. -100db is relative to something. It is not an absolute measurement. It might relate to the dynamic range of the ADC. So for a 16 bits receiver it would be something  like 96 db. But using decimation as far as I understand it can go lower. To have an idea of the noise floor / sensitivity of the receiver I would use a known S9 (=-73dbm) signal generator. You could see it might peak to -43db, which in turns tells that the 100db you see is close to -130 dbm.
Logged

AI5BC

  • Posts: 456
    • HomeURL
Re: Measuring local noise floor with an SDR
« Reply #2 on: November 03, 2022, 04:33:06 PM »

You cannot measure the noise floor with a a VNA. You cannot limit the Receiver bandwidth down to 10 0r 5 KHz. Need a professional Spectrum Analyzer and noise source.
Logged

K6AER

  • Member
  • Posts: 7159
Re: Measuring local noise floor with an SDR
« Reply #3 on: November 03, 2022, 06:49:29 PM »

Noise floor in todays HF receivers is mostly irrelevant.

Typical Ham receivers have anywhere from 97 to 107 dB dynamic range.

S meters reading on a HF receiver are not linear by any imagination. S0 to S3 could be only 1.5 to 3 dB in range. Not until you get to S8 will they approach 6 dB per S unit. Above S9+10 the typical HF receiver goes into compression with S9 +20 being almost 15 dB per S unit.

Your noise floor is determined by your location and the antenna. If you disconnect the the antenna and your S meter drops to S zero your reeceiver is not the issue.

Typical dynamic range of a HF receiver is only 60 dB from the noise floor to S9 +30 dB.

I run an amplifier most of the time for hams have avery high noise floor. Sometimes approaching S9+10. With out the amplifier we would not have a QSO.


« Last Edit: November 03, 2022, 06:54:44 PM by K6AER »
Logged

K4FMH

  • Member
  • Posts: 681
Re: Measuring local noise floor with an SDR
« Reply #4 on: November 04, 2022, 08:58:23 AM »

Noise floor in todays HF receivers is mostly irrelevant.

Typical Ham receivers have anywhere from 97 to 107 dB dynamic range.

S meters reading on a HF receiver are not linear by any imagination. S0 to S3 could be only 1.5 to 3 dB in range. Not until you get to S8 will they approach 6 dB per S unit. Above S9+10 the typical HF receiver goes into compression with S9 +20 being almost 15 dB per S unit.

Your noise floor is determined by your location and the antenna. If you disconnect the the antenna and your S meter drops to S zero your reeceiver is not the issue.

Typical dynamic range of a HF receiver is only 60 dB from the noise floor to S9 +30 dB.

I run an amplifier most of the time for hams have avery high noise floor. Sometimes approaching S9+10. With out the amplifier we would not have a QSO.

Hi Brad,

Good suggestion. Checking out Rob Sherwood's website in his disclosure of how he measures noise floor for his table of measurements or Bob Allison's book on how the ARRL Lab (at least, before Bob retired) measured NF is a good idea.

73,

Frank
K4FMH
Logged

N6YWU

  • Posts: 362
    • HomeURL
Re: Measuring local noise floor with an SDR
« Reply #5 on: November 06, 2022, 12:35:34 PM »

My SDRs include a direct sampling RTL-SDRv3 and a Hermes Lite 2 (among many other SDRs). 

I measure my local noise floor (including the combined contributions from QTH RFI, antenna, and receiver noise) by comparing the spectrum magnitudes of the received IQ data from my antenna against received data from a calibrated signal source (an S9 RF frequency generator), and also an uncalibrated (but roughly known power level) noise source as a double check.  I then use a set of step attenuators to check my magnitude measurement linearity (set to 6 dB per S unit) down to S0.
Logged

KD7RDZI2

  • Member
  • Posts: 689
Re: Measuring local noise floor with an SDR
« Reply #6 on: November 06, 2022, 12:48:39 PM »

My SDRs include a direct sampling RTL-SDRv3 and a Hermes Lite 2 (among many other SDRs). 

I measure my local noise floor (including the combined contributions from QTH RFI, antenna, and receiver noise) by comparing the spectrum magnitudes of the received IQ data from my antenna against received data from a calibrated signal source (an S9 RF frequency generator), and also an uncalibrated (but roughly known power level) noise source as a double check.  I then use a set of step attenuators to check my magnitude measurement linearity (set to 6 dB per S unit) down to S0.

I use the elevator as a calibrated noise source... When somebody uses it, the noise floor jumps exactly to S9, wideband!!! It helps to understand the sensitivity of the receiver and on top of that, during contests, it helps training the ears ;D :-X
Logged

AK4YA

  • Member
  • Posts: 196
Re: Measuring local noise floor with an SDR
« Reply #7 on: November 06, 2022, 06:59:47 PM »

Thanks all.  Can you tell me what noise floor your different sdrs show with no antenna?
I see s3 on mine which seems odd
Logged

KD7RDZI2

  • Member
  • Posts: 689
Re: Measuring local noise floor with an SDR
« Reply #8 on: November 07, 2022, 11:20:51 AM »

Thanks all.  Can you tell me what noise floor your different sdrs show with no antenna?
I see s3 on mine which seems odd
The basic 8 bits RTL stick is around -55 db but using different  decimation it can go lower. The Softrocks receivers and transceivers between  -115 and -120 depending on the bands. It depends on the soundcard as well and input levels. The software I use is Quisk under Linux. Injecting an S3 signal level (-107dbm) is seen almost 20 db above the noise floor at around -95 -100.
Logged

AK4YA

  • Member
  • Posts: 196
Re: Measuring local noise floor with an SDR
« Reply #9 on: November 07, 2022, 11:34:16 AM »

Thanks all.  Can you tell me what noise floor your different sdrs show with no antenna?
I see s3 on mine which seems odd
The basic 8 bits RTL stick is around -55 db but using different  decimation it can go lower. The Softrocks receivers and transceivers between  -115 and -120 depending on the bands. It depends on the soundcard as well and input levels. The software I use is Quisk under Linux. Injecting an S3 signal level (-107dbm) is seen almost 20 db above the noise floor at around -95 -100.
OK I think I amy see where Im out in the weeds here... you say you inject a -107dBm signal, and then say what I interpret to mean the noise floor is from -95 down to -100 but then say the signal is 20dB above the noise, but in my mind I see the signal (-107) is already well below the noise (-95 to -100).  Can you please correct my thinking?  I have got to be off here somewhere.
Logged

KD7RDZI2

  • Member
  • Posts: 689
Re: Measuring local noise floor with an SDR
« Reply #10 on: November 07, 2022, 01:00:23 PM »

Thanks all.  Can you tell me what noise floor your different sdrs show with no antenna?
I see s3 on mine which seems odd
The basic 8 bits RTL stick is around -55 db but using different  decimation it can go lower. The Softrocks receivers and transceivers between  -115 and -120 depending on the bands. It depends on the soundcard as well and input levels. The software I use is Quisk under Linux. Injecting an S3 signal level (-107dbm) is seen almost 20 db above the noise floor at around -95 -100.
OK I think I amy see where Im out in the weeds here... you say you inject a -107dBm signal, and then say what I interpret to mean the noise floor is from -95 down to -100 but then say the signal is 20dB above the noise, but in my mind I see the signal (-107) is already well below the noise (-95 to -100).  Can you please correct my thinking?  I have got to be off here somewhere.

The scale you see in the SDR programs are relative. Do not confuse the db on the software to dbm. I use a source signal, a carrier calibrated to -107 dbm. -107 dbm are an S3 signal. The software I use with a Softrock receiver show a -115 db noise floor. It's not the real noise floor of my receiver. If the real noise floor was just -115 dbm, I would see the carrier just 8 db above the noise floor. The fact that I see the -107dbm carrier 20 db higher than the noise floor means that the scale in the software is simply traslated. Instead of reading -107db, it shows -95 db. Of course I am happy for this because the noise floor of the receiver is better thaN -115. It is probably around 12 db better, -127dbm.

With an RTL stick I can see just -55 db . You might think it is not sensitive. NO! The same signal source at -107 dbm is seen peaking at -35 db, 20dB higher than the noise floor. Evidently the real noise floor of the RTL sticky is around -127dbm. All receivers have this sort of sensitivity, even cheap portables. They are all about the same.
Logged

W9IQ

  • Member
  • Posts: 8866
Re: Measuring local noise floor with an SDR
« Reply #11 on: November 07, 2022, 03:19:58 PM »

Or even dB and dBm.

- Glenn W9IQ
Logged
- Glenn W9IQ

God runs electromagnetics on Monday, Wednesday and Friday by the wave theory and the devil runs it on Tuesday, Thursday and Saturday by the Quantum theory.

AK4YA

  • Member
  • Posts: 196
Re: Measuring local noise floor with an SDR
« Reply #12 on: November 07, 2022, 04:17:22 PM »

ah ok then you've got me sorted now I believe thanks very much.  This would explain the very crazy differences in levels I see with differing SDRs and different software on the same antenna.

My only problem now is these different SDRs (and even differet superhet radios) reports weird differing  base noise S levels.  on the same antenna
Logged

KD7RDZI2

  • Member
  • Posts: 689
Re: Measuring local noise floor with an SDR
« Reply #13 on: November 08, 2022, 01:31:54 PM »

ah ok then you've got me sorted now I believe thanks very much.  This would explain the very crazy differences in levels I see with differing SDRs and different software on the same antenna.

My only problem now is these different SDRs (and even differet superhet radios) reports weird differing  base noise S levels.  on the same antenna

With the antenna not connected con did er the noise floor as S0. The real noise floor is usually around -130 dbm for all receivers. The level you see on the software will differ as the upper limit is not calibrated. Rather it is related to the dynamic range of the receiver. However connecting the antenna the noise floor should increase with the same magnitude for all receivers above the noise floor with the antenna disconnected. To check sensitivities of the receivers I use http://www.norcalqrp.org/s1s9generator.htm which I ugly built using a copper board and spare components. It is one of the most useful device to have. Elecraft sells https://elecraft.com/products/xg3 .
Logged
Pages: [1]   Go Up