Re: FFT Size and Signal & Vector Sources Amplitude Unit

[Marcus D. Leech]

On 09/03/2023 10:14, Wolfgang Wilde wrote:
> In fact, in real world and with real measurement devices, the units 
> are not related to "instantaneous voltages" and even lesser "voltages 
> seen at the antenna". Any form of antenna may deliver different 
> voltages. Is it some Yagi antenna? Or is the active element a closed 
> loop? Both forms would differ totally in "antenna voltage" and both 
> forms will transform the "antenna signal" over impedance 
> transformation to for example some 75 Ohms impedance cable for 
> consumer products or some 50 Ohms impedance cable when you have some 
> TRX system. Real RF _measurements_ are all related to RF-power. Even 
> the often used db/µV targets at a given power, as it only gives valid 
> information when you also provide the impedance of your RF system. 
> This means: talking about a signal of 40 dB/µV could be both, enough 
> for FM receiption or to less for it, depending on what impedance (50 
> Ohms/ 75 Ohms/ 240 Ohms?) you are refering to!
>
> Never the less you do not have any SDR-Sticks out there with given 
> sensitivity nummbers. Even the more expensive devices like HackRF and 
> all the Ettus Research devices are not "measuring" some RF field 
> strength. All you get is somehow a resulting numeric factor of how 
> good the SDR can detect some signal. The sensitivity does vary widely 
> over the frequency ranges of this devices and it is by no means really 
> directly proportional to some RF power or even a voltage at the 
> antenna port or your antenna. You may to some degree use SDR-Sticks or 
> devices like HackRF, USRP's and so on for qualitative informations 
> about a radio signal, but it won't tell you anything about the real 
> power of a signal, as opposite to RF measurement systems! With all my 
> devices (ranging from RTL2832 based sticks over HackRF and others) you 
> get jumps in signal strength when you do a full band power scan. This 
> happens for example, when the SDR hardware switches to some other 
> oscillator setting. All of my devices can do a full band scan of at 
> least 1 GHz or even more, but none of them can do it by just stepping 
> up the Synthesizer PLL without for example using harmonics of the base 
> PLL Frequency for mixing down to the IF band or Baseband. And for each 
> time switching to another harmonic you get signal strength jumps. And 
> with each other PLL frequency you might get other mixing products, 
> ghosting from other frequencies and so on. So, the kind of SDR we 
> refer to here can not give any numeric factors to RF power nor RF 
> voltages. It is just a numeric value, somehow more or less 
> proportional to the quality the SDR can receive the signal. No units, 
> no absolute values at all.
>
>
> Regards
I'll repeat this again, because you clearly didn't "get" it.   The 
numbers coming out of the signal-processing chain are
  *linearly proportional* to the instantaneous (as in at the moment the 
sample was taken by the ADC) voltage as seen
  at the antenna input to the receiver.   This says *NOTHING* about the 
antenna itself, nor whatever cabling and other
  bits-and-pieces are between the receiver input port and the antenna.

YES, the calibration constants will, absolutely, vary over the 
tuning/bandwidth/sample-rate capabilities of the receiver.
  But if they *ARE NOT* linearly-proportional (or mostly-so) to the 
instantaneous antenna-*PORT* input voltage, then we
  might as well all go home.

The samples that arrive into your flow-graph aren't some handy-wavy 
random thing that kinda-sorta represents the
  real world.  They are very-definitely a *linear proxy* for the 
instantaneous voltage as seen at the antenna input port
  at the time that it was sampled.  Again, if they aren't very close to 
this, then we might all just as well go home and
  take up basket-weaving.

What IS true about actual laboratory measurement instrumentation is that 
such instrumentation is *calibrated* over
  it's operating range (in as many steps as seems reasonable) to 
produce results that are directly-tied to physical units.

You can do EXACTLY THE SAME THING with even a cheap receiver like the 
RTL-SDR, HackRF, USRPs, LimeSDRs, etc, etc.
  In fact, USRPs (some of them) now have a *CALIBRATION INTERFACE AND 
API* that allows you to use them in the
  same way as you can use a laboratory instrument.

In MANY actual cases, you'd like your radio to be calibrated over some 
much-smaller range of its operating parameters--
  you'll be using it for perhaps a single application, where 
understanding what is appearing at the antenna input ports
  in terms of power or (by a bit of simple math, voltage) may be important.

In MY case, I calibrate in degrees-K of noise power, because that's 
relevant to my usage of these types of radios for
  radio astronomy.

Your post makes it seem like SDRs are delivering samples that bear only 
the weakest relationship to the physical world,
  and that just isn't true.  IT CAN'T BE.