What is residual noise (vs noise figure) and flicker of RF/IF Gain blocks

I would suggest thinking about it a bit differently. Start with the equation.

Noise Spectral Density (dBm/Hz) = -173.8 dBm/Hz + Gain(dB) + Noise Figure(dB)

So an amplifier with a noise figure of 10 dB and a gain of 10 dB will have an output noise floor of -153.8 dBm/Hz. If you output signal level is 0 dBm then the noise spectral density in dBc/Hz is -153.8 dBc/Hz.

We generally don't characterize flicker noise since these devices are generally ac-coupled at a high-pass corner frequency that is well above the 1/f corner. If they do have flicker noise it will be independent of whatever signal is present.

Hi Nash

Thank you for the reply.

From the Noise Power Density equation I would do this experiment at T=25C:

I would drive ADL5536 with 0dBm signal having -177dBc/Hz phase noise at 1kHz.

Should I assume that I will get ~18dBm output power with a phase noise of -174dBc at 1kHz offset (NF=2.8)?

( I considered that FM/AM noise are not correlated, resulting in the PN noise of -177 instead of -174dBc/Hz).

I wonder if I use an extremely good source of +10dBm and -185dBc/Hz, what power&noise I would get at the output?

I guess the 1/f corner can be measured by either building an oscillator and measure where the slope of the PN changes from 20dB/dec to 30db/dec. Alternatively, a very goon OCXO can be used to find the flicker of devices with larger flicker corner.

Regards,

Dorin

If you apply a 0 dBm signal with phase noise of -177 dBc/Hz at 1 KHz offset, then the absolute value of the phase noise is equal to -177 dBm/Hz. The output noise will be the sum of the input noise (that is the KT Noise of -173.8 dBm/Hz and the -177 dBm/Hz from your source; this is approximately -172 dBm/Hz) plus the gain plus the noise figure.

Thanks Enash,

The previous answer may satisfy some customer, unfortunately ;-( I have to be more on the high-maintenance side  because of my application.

For example: I have measured in the past a GaN amplifier which had a NF better than 0.5dB but a flicker corner higher than 20MHz. With this I measured an output phase noise of -180dBc at 30MHz offset but the noise at 10kHz was ~-130dBc.

I created a kind of test-case to exemplify why flicker is important and why input power level is important.

I understand if these measurements are not available.

Assuming I have access to an 100Mz OCXO similar to the following:

(http://www.pascall.co.uk/products/OCXO-series-and-OCXOF-series.asp)

Output power = 13dBm with Output Phase noise -180dBc@10kHz

(This means that the signal is at +13dBm and noise is at -167dBm.)

If I attenuate this signal by 13dBm, I will hit the noise floor and have +0dBm signal with -173.8dBm noise.

If I apply this +0dBm signal of -173.8dBm noise at 10kHz offset to the ADL5536

=> (simple assumptions)

Output= ~ +19dBm ; Noise =-173.8+20db(gain) +2.6dB(NF) = -151.2dBm => SNR=-170.2dBc

However, the ADL5536 is fabricated on a GaAs HBT process which most likely has a flicker corner between 300kHz and 10MHz. If I count for a 1MHz flicker, my noise figure degrades 20dB between 1MHz and 10kHz, resulting in a SNR of -150.2dBc.

Now, if I apply the full +13dBm power I could assume that

Output > ~20dBm Noise = -180+20+2.6 = 157.4dBm => SNR = -177.4dBc.

However, because I drive the amplifier in compression, it may well happen that the noise is higher and the flicker corner is at higher frequency.

This is why is important for me to know the power level of the input signal used to measure the NF.

Regards,

Dorin

The math for the case with +13dBm input power is incorrect.

Here's the corrected version:

Output : 20-33dBm Noise=-167dBm +20 +2.6 = -144.4dBm

SNR = -164.4 to 177.4 depending on the compression level.

Dorin

The power level of the signal is not specified because of the nature of the measurement. Typically, the noise figure analyzer measures noise and specifies the result at a level below compression but the specific level is not known. I believe you are looking for a noise figure measurement with a blocker, and unfortunately we don't have that measurement for this device.