Why would I want to measure Noise without a Noise Figure Meter (NFM)?
Regarding point #1, the Measurement Bandwidth of your NFM is listed somewhere in the documentation for your test equipment. Find it. If it is > the bandwidth of your DUT, then you have a problem. Look at the figure above. The NFM will take some average value for amplitude since it can't zoom in on the exact frequency you are interested in; only a range of frequencies equal to the Measurement Bandwidth. Oops!
(By the way, measurement bandwidth is important when using all sorts of test equipment. Keep this in mind when measuring very narrow frequency-response devices or when searching for spurious signals)
What you will need:
For accurate measurements, you need to know what the Net Gain and Noise is from the output of your DUT to the input of your Spectrum Analyzer (LNA plus all RF cabling). This will not be a "narrow" frequency response so you can use a classic Noise Figure Meter to measure this.
What if a NFM is unavailable?
You will have to look up the Noise and Gain of your LNA on its data sheet, measure the loss of your cables, and cascade them together to figure out the Net Gain and Noise Figure; this will be less accurate.
**Use this calculator to determine Net Gain and Noise of a cascaded RF network! (coming someday)
Theory:
An ENR noise source works in the following manner. It applies a Noise reference at two noise temperatures: Thot and Tcold into our DUT. We want to measure the resultant output power into the Spectrum Analyzer when Thot then Tcold are applied; their ratio will yield Net Noise through our system (from DUT input to Spectrum Analyzer input).
From the Net Noise and knowledge of Gain & Noise of the other components in our system, we can determine the Noise of our DUT by re-arranging the Noise Cascade equation.
Why do I need an LNA?
Since the ENR source will be injecting such a small amount of power into your DUT, the LNA needs to have enough gain to bring this power above the noise floor of the Spectrum Analyzer. The Noise contribution of the LNA must be removed later when determining the Noise of the DUT.
- Your Device Under Test (DUT) may have a narrow frequency response. So narrow that it is less than the Measurement Bandwidth of your NFM. Think of a narrow Surface Acoustic Wave Filter.
- You cant afford a NFM on top of all the other fancy equipment you purchased.
- You want to know how a NFM works.
Regarding point #1, the Measurement Bandwidth of your NFM is listed somewhere in the documentation for your test equipment. Find it. If it is > the bandwidth of your DUT, then you have a problem. Look at the figure above. The NFM will take some average value for amplitude since it can't zoom in on the exact frequency you are interested in; only a range of frequencies equal to the Measurement Bandwidth. Oops!
(By the way, measurement bandwidth is important when using all sorts of test equipment. Keep this in mind when measuring very narrow frequency-response devices or when searching for spurious signals)
What you will need:
- Spectrum Analyzer with a Resolution Bandwidth << Bandwidth of your DUT (not hard to find) and "lots" of dynamic range.
- Excess Noise Ratio (ENR) source.
- Low Noise Amplifier (LNA) that operates at your Frequency of Interest (FOI).
- DC power supply to turn your ENR on/off.
- Amplifier with known noise and gain. (optional).
For accurate measurements, you need to know what the Net Gain and Noise is from the output of your DUT to the input of your Spectrum Analyzer (LNA plus all RF cabling). This will not be a "narrow" frequency response so you can use a classic Noise Figure Meter to measure this.
What if a NFM is unavailable?
You will have to look up the Noise and Gain of your LNA on its data sheet, measure the loss of your cables, and cascade them together to figure out the Net Gain and Noise Figure; this will be less accurate.
**Use this calculator to determine Net Gain and Noise of a cascaded RF network! (coming someday)
Theory:
An ENR noise source works in the following manner. It applies a Noise reference at two noise temperatures: Thot and Tcold into our DUT. We want to measure the resultant output power into the Spectrum Analyzer when Thot then Tcold are applied; their ratio will yield Net Noise through our system (from DUT input to Spectrum Analyzer input).
From the Net Noise and knowledge of Gain & Noise of the other components in our system, we can determine the Noise of our DUT by re-arranging the Noise Cascade equation.
Why do I need an LNA?
Since the ENR source will be injecting such a small amount of power into your DUT, the LNA needs to have enough gain to bring this power above the noise floor of the Spectrum Analyzer. The Noise contribution of the LNA must be removed later when determining the Noise of the DUT.
Procedure:
- Measure the Gain (GDUT) of your DUT with a Network Analyzer at your FOI .
- Measure the Noise (FLNA) from output of DUT to input of Spectrum Analyzer.
- Connect the chain together as indicated in the Test Setup.
- Set the Spectrum Analyzer to the following:
a. Center: FOI
b. Span: (3dB Bandwidth of Device) / 10
c. RBW: (Span) / 100
d. VBW: (RBW) / 100
e. Averaging: 30
f. Amplitude/ div: 1 dB
g. Attenuation: 0 dB
Note: These settings are a starting point; you will probably have to do some tweaking until you get sensible data. - Measure output power at Thot: Turn on the DC Power Supply to stimulate the ENR source (usually 28V, but depends on model of ENR source). Re-start the Spectrum Analyzer Averaging and note the output power: Phot.
- Measure output power at Tcold: Disconnect the DC Power Supply. Re-start the Spectrum Analyzer Averaging and note the output power again: Pcold.
The Net Noise Power (linear) of the chain is:
where Y (dB) = |Phot - Pcold| (dBm), and ENR (dB) is listed in a table on your ENR source as a function of frequency.
By re-arranging the Noise Cascade equation, the Noise Power of the DUT, FDUT, is:
*Note: All components of this equation are linear, not dB.
where Y (dB) = |Phot - Pcold| (dBm), and ENR (dB) is listed in a table on your ENR source as a function of frequency.
By re-arranging the Noise Cascade equation, the Noise Power of the DUT, FDUT, is:
FDUT = Fnet - (FLNA-1)/GDUT
*Note: All components of this equation are linear, not dB.
In review, this is what you need to measure:
- GDUT
- FLNA
- Phot
- Pcold
- ENR
Replace the DUT with an amplifier with a known NF and Gain. Do your measurements make sene?
...Coming someday: a calculator for Noise Figure
No comments:
Post a Comment