AnsweredAssumed Answered

FMCOMMS6 noise analysis

Question asked by dmarcosgon on Oct 14, 2015
Latest reply on Oct 19, 2015 by bsam

Hello,


First of all sorry if this is not the correct forum, I wasn't really sure wether to post this here or in eval boards forum.

 

I'm designing a radio receiver and from the IQ demodulator to the ADC the design is very similar to FMCOMMS6, which we are using as reference design. We use the same IQ demodulator, the same ADC and we're choosing the ADC amplifier.

 

There's a noise analysis spreadsheet provided with the FMCOMMS6 and I was trying to adapt it to our design to see how noisy the ADC driver can be before it's a significant total noise contribution. Here's the spreadsheet:

 

https://wiki.analog.com/_media/resources/fpga/xilinx/fmc/fmcomms6/noiseanalysis_adl5380_adl5566-ad9652.xls

 

I've found what I think it's a big mistake in the spreadsheet but being such a mistake I'm not sure if It's actually a mistake or if I'm missing something very important. The problem comes with the RF noise contribution calculation.

 

The spreadsheet converts from noise figure and conversion gain to voltage noise spectral density, multiples it with the noise bandwidth of the demodulator and gets the RMS noise voltage contribution referred to the output of the IQ demodulator. All ok for now. Then, when calculating the total RMS noise referred to the input of the ADC (cells K30 to K38) it just quadratically sums the previously calculated noise as another term (cell K32 has =D50 formula) without doing anything else. I think you should first apply the amplfier gain (16 dB) to that noise contribution and also an attenuation factor to take into account the divider formed by the 50ohm diff output impedance of the ADL5380 and the 160ohm diff input impedance of the ADL5566 to refer the noise to the input of the ADC, and then you can sum all the terms referred to the same point.

 

Am I right or am I missing something very important?

 

Thanks in advance for your help

Outcomes