Noise figure for a sytock RTL-SDR

[Jan Lustrup]

Any know the noise figure for a normal stock RTL-SDR?

Jan Lustrup –LA3EQ

[Alex P]

     Air Spy mini

Technical specifications

• Continuous 24 – 1700 MHz native RX range,
  
• 3.5 dB NF between 42 and 1002 MHz
• 12bit ADC

[Alex P]

But, as you divide that number by the preceding system gain, the number is small compared to the first elements. 

https://www.microwaves101.com/encyclopedias/noise-figure

On Tuesday, May 21, 2024 at 6:48:43 AM UTC-4 Jan Lustrup wrote:

[James Abshier]

Data Sheet Here:

https://rtl-sdr.com/wp-content/uploads/2013/04/R820T_datasheet-Non_R-20111130_unlocked.pdf

The data sheet gives NF of 3.5 dB, but this is apparently with 75 Ohm input impedance.

On 5/21/24 06:48, Jan Lustrup wrote:

Any know the noise figure for a normal stock RTL-SDR?

Jan Lustrup –LA3EQ

--
--
You received this message because you are subscribed to the Google
Groups "Society of Amateur Radio Astronomers" group.
To post to this group, send email to sara...@googlegroups.com
To unsubscribe from this group, send email to
sara-list-...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/sara-list?hl=en
---
You received this message because you are subscribed to the Google Groups "Society of Amateur Radio Astronomers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sara-list+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/sara-list/008901daab6c%246f0fb910%244d2f2b30%24%40no.

[Marcus D. Leech]

On 21/05/2024 13:00, James Abshier wrote:

Data Sheet Here:

https://rtl-sdr.com/wp-content/uploads/2013/04/R820T_datasheet-Non_R-20111130_unlocked.pdf

The data sheet gives NF of 3.5 dB, but this is apparently with 75 Ohm input impedance.

Somewhat higher outside its main operating range.  Down at the high end of HF, it's about 5.5dB, and also above
  about 1200MHz, about 5dB (from memory).

But, for radio astronomy, the receiver noise figure just doesn't matter very much, unless it's horrific.  It will be swamped
  by the gain of the LNA ahead of it.

For example, let's say the Treceiver is equivalent to 1000K, but there's 30dB of low-noise gain ahead of it. That Treceiver
  will get reduced by a factor of 1000.  So will only add about 1K to the overall Tsys.

To view this discussion on the web visit https://groups.google.com/d/msgid/sara-list/cee8f7e9-833f-4496-8fdd-4af4a9e69e1a%40ameritech.net.

[Dimitry UA3AVR]

From my experience with RTL-SDR and noise figure meter CANFI its NF noiticeably depends on the gain and may be 5 dB and more for lower gains. So, its usage in devices like CANFI can be problematic without high gain LNA before RTL-SDR.

вторник, 21 мая 2024 г. в 20:05:02 UTC+3, Marcus D. Leech:

[b alex pettit jr]

With the 8 Bit  RTL-SDR,  

its   Quantization Noise   may be of greater significance than Noise Figure

              6 Bit  vs  8 Bit Noise Level Comparison

-------------------------------------------------------------------------------

https://classes.engineering.wustl.edu/ese488/Lectures/Lecture5a_QNoise.pdf

Quantization (signal processing) - Wikipedia

Quantization (signal processing) - Wikipedia Quantization, in mathematics and digital signal processing, is the process of mapping input values from a large ...

( I use the 12-Bit  AS mini )

Alex

[Peter East]

The important choice for amateurs is to choose the total RF gain to fully utilize the A/D dynamic range. To get the RF gain such that the noise floor occupies a few digitizer increments should be the aim.

Peter

[b alex pettit jr]

Hello Peter,

" .... fully utilize the A/D dynamic range. "     agreed .

This was a system characterization curve  using a random noise RF source ..

Distanced the source from the antenna so the displayed level was just below saturation, then added attens in front of the SDR

( w/  AirSpy mini's 3 gain stages set to Max )

Alex

==================================

[Dimitry UA3AVR]

Sorry, i didn't mention the RTL-SDR in my experience with significanlt NF variations vs gain was in 8 bit version.

вторник, 21 мая 2024 г. в 22:26:21 UTC+3, Dimitry UA3AVR:

[Marcus D. Leech]

On 22/05/2024 03:21, 'b alex pettit jr' via Society of Amateur Radio Astronomers wrote:

With the 8 Bit  RTL-SDR,  

its   Quantization Noise   may be of greater significance than Noise Figure

I'll note that at large modern observatories, ADCs of more than 8-bits are fairly rare.

For many-channel, high-speed sampling, 3 or 4 bit ADCs are quite common.

The CHIME telescope (1024 RF channels) uses 8-bit ADCs.

The reason one might want more bits is dynamic range, so that RFI can be eliminated digitally prior to further
  down-stream processing.  That's why CHIME has "so many bits" in the ADCs.  They operate in a frequency range
  where RFI is common, so need more bits in the ADC to deal with it.

[Peter East]

Alex,

Don't forget you have Friis on your side. In practice providing you ensure sufficient gain for the first LNA to dominate the theoretical implications of the A/D noise then by ensuring the noise exceeds a few A/D bins you maximize the A/D dynamic range. Also we tend to average a lot of data which tends to increase the final resolution.

The RTL2832U limitation of 8-bits has never posed a problem for me in all my H-line and pulsar work.

Professional correlaters sometimes go down to 1-bit A/D converters for speed.

Peter

[Richard Flagg]

Hi Alex,

I dont understand how you are getting a "linear" response down to 10 dB below 15 kelvin.

As you add attenuation to the rcvr input you are adding noise generated in the attenuator.

Richard

As you add attenuation

--
--
You received this message because you are subscribed to the Google
Groups "Society of Amateur Radio Astronomers" group.
To post to this group, send email to sara...@googlegroups.com
To unsubscribe from this group, send email to
sara-list-...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/sara-list?hl=en
---
You received this message because you are subscribed to the Google Groups "Society of Amateur Radio Astronomers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sara-list+...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/sara-list/1719360913.1562289.1716375620568%40mail.yahoo.com.

[Marcus D. Leech]

On 22/05/2024 19:19, Richard Flagg wrote:

Hi Alex,

I dont understand how you are getting a "linear" response down to 10 dB below 15 kelvin.

As you add attenuation to the rcvr input you are adding noise generated in the attenuator.

Richard

As you add attenuation

Reminds me of the time I queried Hugh Chivers about his calibration process for his riometers.
  "I just use a precision step attenuator in the lab".  Yeah. State of sin, etc.

To view this discussion on the web visit https://groups.google.com/d/msgid/sara-list/c7fcd800-618e-4906-a363-93866a38586d%40hawaii.rr.com.

[b alex pettit jr]

Richard,

The attenuators were on the Input to the SDR = a high level RF signal. not the Receiver

Alex

[b alex pettit jr]

Richard,

The attenuators were on the Input to the SDR = a high level RF signal. not the LNA

Alex

[Richard Flagg]

Alex,

It looks like you have inserted up to 30 dB or so of attenuation at the input of the SDR - hence raising its noise figure by that amount,  and possibly negating the benefit of going thru a low noise preamp.

Or maybe I just don't understand your plot and procedure.

Thanks

Richard

--
--
You received this message because you are subscribed to the Google
Groups "Society of Amateur Radio Astronomers" group.
To post to this group, send email to sara...@googlegroups.com
To unsubscribe from this group, send email to
sara-list-...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/sara-list?hl=en
---
You received this message because you are subscribed to the Google Groups "Society of Amateur Radio Astronomers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sara-list+...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/sara-list/1533395677.1746809.1716422444656%40mail.yahoo.com.

[b alex pettit jr]

The test was to verify the SDR was linear over a signal range.

I then added the values obtained when it is actually used to the plot.

<Fixed_RF_Source>    <Attenuators>    <SDR_Input>

Alex

[b alex pettit jr]

Richard,

This may better explain the test. 

Once a Source<>Antenna distance which did not saturate the electronics was found,

it remained in this location and a series of attenuators were added to the input of the SDR and the displayed amplitude recorded.

It shows that with the particular gains selected on the SDR, 

the system has ~ 6 dB headroom above 290K and that 

the system noise floor is well below 10K cold sky

Alex

[b alex pettit jr]

It shows that with the particular gains selected on the SDR, 

the system has ~ 6 dB headroom above 290K and that

the SDR noise floor is well below 10K cold sky

.

[Richard Flagg]

Hi Alex,

Sorry but I am still confused.

What is the vertical scale?  For example what does -45 dB mean - what is the reference?
I would normally think of the vertical axis in such a plot in terms of power.

Can you say a few more words about what you mean saying the system noise floor is well below 10 kelvin.

Are you saying the system temperature looking into the first LNA is below 10 kelvin?

Thanks

--
--
You received this message because you are subscribed to the Google
Groups "Society of Amateur Radio Astronomers" group.
To post to this group, send email to sara...@googlegroups.com
To unsubscribe from this group, send email to
sara-list-...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/sara-list?hl=en
---
You received this message because you are subscribed to the Google Groups "Society of Amateur Radio Astronomers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sara-list+...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/sara-list/137372592.1818256.1716459420878%40mail.yahoo.com.

[b alex pettit jr]

Hello Richard,

This is a typical data plot  from the AirSpy Studio software where the 10K sky level is ~ - 60 dB ( power )

( I have never calibrated the absolute values of the scale, but the attenuators verify it corresponds correctly with signal level )

I used the Noise Source and by using attenuators, simulated equivalent (voltage/power) signal levels higher and lower than that into the SDR ..

 

It does not correlated to an actual measurement, but an electrical signal  into the SDR.

The noise floor of the SDR is lower than the 40 dB LNA Amplified output corresponding to the 10K sky ..

Alex

===================================

[b alex pettit jr]

               The purpose of the measurement sequence was

To verify the SDR was linear over the Radio Telescope's signal level range.

Alex

[Jim Sky]

SDRplay.com has data sheets for each of their products.  The noise figure will vary with frequency. Below is a sheet for the RSP1A.

On Tuesday, May 21, 2024 at 6:48:43 AM UTC-4 Jan Lustrup wrote:

[Marcus D. Leech]

On 24/05/2024 20:50, Jim Sky wrote:
> SDRplay.com has data sheets for each of their products.  The noise
> figure will vary with frequency. Below is a sheet for the RSP1A.

Meh, the only time you might care about the noise-figure of just the
receiver might be if you want to directly connect it to
  your antenna, and even then, you'd probably only do that at HF and
low-VHF frequencies.  We operated my riometer design
  for several months at a field site with no amplifier at all between
the antenna and SDR receiver.  Worked fine.   The inherent
  noise temperature of the sky at 38MHz is about 5000K, which is
considerably higher than the inherent noise figure of
  the receiver...



In all other cases, in radio astronomy and "radio science", you'd use an
LNA, which immediately renders the noise figure
  of the receiver almost completely irrelevant, or of only minor
consequence.

Keep in mind that at HF the noise temperature of the *channel itself* is
likely higher than the noise figure of your
  receiver, which is why you don't see anyone busting their brains
trying to make a super-low-noise HF receiver.
  It wouldn't help.  Of more importance at HF is dynamic range, etc.

[Jim Sky]

"Meh, the only time you might care about the noise-figure of just the
receiver might be if you want to directly connect it to
  your antenna, and even then, you'd probably only do that at HF and
low-VHF frequencies" 

I missed that Jan was specific to RTLSDR which he could conceivably use 1.4 GHz.  My SDRPlay reference was inappropriate. I would still like to know the noise figure of the receiver when selecting front end gain blocks.  So it isn't an unreasonable question.  but the answer is 5 to 8 dB depending on the manufacturer and frequency. 

[b alex pettit jr]

It is also important to evaluate the portion of the ADC range being used for the acquisition of the signals...  

Quantization Noise is not an insignificant parameter and should be  'reasonably'  minimized.. 

Understanding the Dynamic Range Specification of an ADC - Technical Articles

Understanding the Dynamic Range Specification of an ADC - Technical Arti... This article will discuss the dynamic range specification of an ADC.

Alex

I missed that Jan was specific to RTLSDR which he could conceivably use 1.4 GHz.  

My SDRPlay reference was inappropriate. I would still like to know the noise figure of the receiver when selecting front end gain blocks.  

So it isn't an unreasonable question.  but the answer is 5 to 8 dB depending on the manufacturer and frequency. 

...

[Dimitry UA3AVR]

Also note, please, following FFT and averaging widens the dynamic range of ADC and moves the noise floor down. For FFT it depends on how many FFT points (bins) are used for spectrum. The noise floor (and consequently the dynamic range) improvement is 10*log_10( M/2 ), where M - the number of bins. Because of this, i think, low bit ADC are still effective for using in telescopes. Don't matter that low bit ADC has narrow natural dynamic range; the FFT makes it wider anycase.

See also enclosed papers form Analog Devices about FFT.

воскресенье, 26 мая 2024 г. в 13:31:20 UTC+3, b alex pettit jr:

[Jan Lustrup]

’Yes Jim, I was wondering about the genuine RTL-SDR dongle stick that I use on 21cm for radio astronomy.

. -Jan

--

--
You received this message because you are subscribed to the Google
Groups "Society of Amateur Radio Astronomers" group.
To post to this group, send email to sara...@googlegroups.com
To unsubscribe from this group, send email to
sara-list-...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/sara-list?hl=en
---
You received this message because you are subscribed to the Google Groups "Society of Amateur Radio Astronomers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sara-list+...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/sara-list/cda8e344-471f-4b47-a3c6-b78085e50a8dn%40googlegroups.com.

[James Abshier]

Jan,

I recently came across some more information on the NF of an R820T2-based SDR. The attached plot was apparently generated from NF measurements. It shows that at low gain settings the NF can be rather poor. Even at the highest gain setting, the data sheet value of 3.5 dB was not quite attained.

Jim Abshier

On 5/21/24 13:00, James Abshier wrote:

Data Sheet Here:

https://rtl-sdr.com/wp-content/uploads/2013/04/R820T_datasheet-Non_R-20111130_unlocked.pdf

The data sheet gives NF of 3.5 dB, but this is apparently with 75 Ohm input impedance.

On 5/21/24 06:48, Jan Lustrup wrote:

Any know the noise figure for a normal stock RTL-SDR?

Jan Lustrup –LA3EQ

--

--
You received this message because you are subscribed to the Google
Groups "Society of Amateur Radio Astronomers" group.
To post to this group, send email to sara...@googlegroups.com
To unsubscribe from this group, send email to
sara-list-...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/sara-list?hl=en
---
You received this message because you are subscribed to the Google Groups "Society of Amateur Radio Astronomers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sara-list+...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/sara-list/008901daab6c%246f0fb910%244d2f2b30%24%40no.

--
--
You received this message because you are subscribed to the Google
Groups "Society of Amateur Radio Astronomers" group.
To post to this group, send email to sara...@googlegroups.com
To unsubscribe from this group, send email to
sara-list-...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/sara-list?hl=en
---
You received this message because you are subscribed to the Google Groups "Society of Amateur Radio Astronomers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sara-list+...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/sara-list/cee8f7e9-833f-4496-8fdd-4af4a9e69e1a%40ameritech.net.

[Marcus D. Leech]

On 08/06/2024 13:27, James Abshier wrote:

Jan,

I recently came across some more information on the NF of an R820T2-based SDR. The attached plot was apparently generated from NF measurements. It shows that at low gain settings the NF can be rather poor. Even at the highest gain setting, the data sheet value of 3.5 dB was not quite attained.

Jim Abshier

The way gain adjustment is implemented is via a variable attenuator -- this is standard in RF architectures, so as the gain
  is lowered, the attenuation is increased.  I don't know whether R820T2 has the attenuator in front of the first gain stage
  or after.  From the plots, it looks like it's in front of the first gain stage.

Regardless, for science applications, the back-end receiver noise figure is largely irrelevant, since you use an LNA
  "up front" which more-or-less defines overall Tsys.

If you're finding that you have to run the receiver (whatever it is) at a very low gain (high attenuation) setting, there's
  something wrong in your overall architecture between antenna and receiver.

To view this discussion on the web visit https://groups.google.com/d/msgid/sara-list/4090afbe-cea6-4d5d-8164-8a6bc75453d3%40ameritech.net.

[Jan Lustrup]

Thanks Jim.....Just what I was looking for!

-Jan

To view this discussion on the web visit https://groups.google.com/d/msgid/sara-list/4090afbe-cea6-4d5d-8164-8a6bc75453d3%40ameritech.net.

[Lamar Owen]

On 5/22/24 09:14, Marcus D. Leech wrote:
> On 22/05/2024 03:21, 'b alex pettit jr' via Society of Amateur Radio
> Astronomers wrote:

>> With the*8 Bit*  RTL-SDR, *
>> *
>> *
>> *
>> *its /  Quantization Noise /may be of greater significance than
>> /Noise Figure
>> /*

> I'll note that at large modern observatories, ADCs of more than 8-bits
> are fairly rare.

To add a data point, the DIRV project that was done here at PARI in the
2010's used 8-bit 1Gs/s oscilloscope ADCs on boards built by Berkeley,
specifically the IBOB plus two ADC2x1000-8 (
https://casper.astro.berkeley.edu/wiki/IBOB and
https://casper.astro.berkeley.edu/wiki/ADC2x1000-8 ).  Three of these
were used; one for a 200MHz bandwidth centered at 2.2 GHz, and the other
two for two 200MHz bandwidth segments in X-band, with the filters
centered at 8.375GHz and a filter bandwidth of 525MHz.  The two segments
were chosen to bracket the X-band downlink segment at 8.4Ghz.

The standard way of getting more dynamic range than what your ADC can
natively do is to oversample (
https://www.analog.com/en/resources/technical-articles/increase-dynamic-range-of-sar-adcs-using-oversampling.html
).  Effective Number of Bits (ENOB) can be increased this way since
quantization noise is shaped noise, with its highest amplitude at the
Nyquist frequency.  Raise the Nyquist frequency, filter, and decimate,
and the quantization noise is greatly reduced.

But in radio astronomy, bandwidth is critically important for
sensitivity, even more important than ENOB.  The only reason to have
high dynamic range is to deal with RFI for most radio astronomy use;
since you can use stepped attenuators post-LNA to bring the continuum
flux within the dynamic range of the ADC (the original DIRV receiver
design, done by Charles Osborne, used some really nice digitally
controller Mini-Circuits stepped attenuators, but they were found by the
team to not be necessary in practice).  If the ADC input is driven to
saturation by RFI then sensitivity is very negatively impacted.

Even the very brightest continuum sources outside the solar system only
need 3-6 dB of dynamic range, even with moderately large antennas.  Just
make sure the noise floor of the system at the ADC input is above the
noise floor of the ADC so that it can be detected, and make sure your
calibration noise source for Dicke switching or other noise source
calibration method has a level less than the full scale of the ADC.  And
then adjust based on expected RFI.

Now, in the real world case, at S-band, for instance, you do indeed NEED
a pretty large dynamic range, thanks to Sirius XM satellites and 4G LTE
that bracket the satellite downlink band from 2.2 to 2.3GHz.  XM5 and
XM6 in particular are incredibly strong signals; 73dBW EIRP.  Not a
typo: 73 dBW.  But the 4G LTE signals ate 2.125, 2.150, and 2.175 GHz
are even stronger.  We have to be careful with the dynamic range of the
analog fiber links from the feed assembly down to the downconverter
rack, even, and those are 40dB rated.  But at X-band the story is
completely different.  But that also means that if I have only 40dB of
dynamic range at the fiber link any ADC dynamic range beyond 40dB isn't
very useful.  An ENOB of 7 bits gives 42dB dynamic range (there's a
chart in the page at
https://www.prosoundtraining.com/2010/03/17/more-bits-more-footroom/
that, while built for audio ADCs, is still valid regardless of sampling
rate).