I am using the Agilent E4446A to take a noise floor measurement. The spectrum analyzer is connected to a bow tie antenna via 3 low-noise RF amplifiers made by Mini Circuits. The antenna is contained in a completely enclosed Faraday cage and the amplifiers are being run with batteries (as to not add noise from a power supply). The amplifiers seem to work fine when connected to the spectrum analyzer and antenna by themselves and in pairs. However, when I connect the 3 together I see a huge spike in the noise floor measurement. I am seeing nearly 60dB from the noise floor at approximately 182MHz consistently. My noise floor measurement is taken from 100MHz to 300MHz. I was wondering if the cascade of the 3 amplifiers results in some kind of noise resonances that are causing this "noise spike"? I would like to use the 3 amplifiers for future measurements to increase the level of the smallest signal that I can detect.
It appears that one of the amplifiers is breaking into oscillations when cascaded together. There is no need to connect so many amplifiers together. The noise floor of the system is primarily driven by the noise figure of the first amplifier. What are the noise figures and gains of the amplifiers used? By adding excessive amplifiers, you increase the noise floor because of the added noise from each amplifier stage. The best amplifier to use with the E4446A is the option 110 amplifier, since the instrument is calibrated with the amplifier to provide accurate power measurements. Also, the amplifier has a low noise figure and sufficient gain. Adding any amplifier ahead of the optional built-in preamp wouldn’t improve the sensitivity of the analyzer.
Regards -
-------------------------------------------------------------------------------------
Assuming this instrument is not equipped with option 110 - it does appear that a retrofit is available.
Alternatively, a cost effective external amplifier choice which should result in excellent system sensitivity would be the 11909A:
http://cp.literature.agilent.com/litweb ... -1373E.pdf
For a do-it-yourself, low-budget solution, the LNA-1000 from RF Bay Inc has specs similar to the 11909A:
http://www.rfbayinc.com/LNA/LNA-1000.pdf
Hi Tabbott,
I agree with your excellent considerations, except the last statement that says: “Adding any amplifier ahead of the optional built-in preamp wouldn’t improve the sensitivity of the analyzer”. I believe that in many cases the sensitivity of a SA can be improved, yes, by adding a LNA ahead the built-in preamp.
According to the Agilent’s specifications for the E4446A, the typical DANL (displayed average noise level) with the 110 preamp ON is –164 dBm, for 50 to 500 MHz and BW of 1 Hz. On the other hand, an ideal noiseless SA would have a noise floor of –174 dBm/Hz, what means the overall noise figure of the E4446A plus preamp is 10 dB. This can be adequate for many applications but certainly a NF of 10 dB isn’t spectacular by any measure. It is true that the specifications for a SA are a compromise between low noise, linearity, wide bandwidth and wide dynamic range, etc. That compromise explains why the noise figure of most SA is “bad” even with a “low-noise” built-in preamp.
Back to the E4446A, it should be clear now that the sensitivity can indeed be improved by using an external preamp. It is enough that the external preamp have a NF better than10 dB. In the specific case where the source is an antenna (which typically generates much less thermal noise than a resistor of the same impedance), the increase in sensitivity by using a good LNA can be dramatic. It is true that an external LNA could compromise the dynamic range but in many measurements an ultimate sensitivity is more important than a large dynamic range.
Regards,