we usually measure output spectrum of DAC AD9776 with spectrum analyzer.

It seems that SNR is too low for a this DAC (about 10dB !). But I don't know what's wrong ?

thanks

we usually measure output spectrum of DAC AD9776 with spectrum analyzer.

It seems that SNR is too low for a this DAC (about 10dB !). But I don't know what's wrong ?

thanks

Can you explain what mode you are using the AD9776 ( interpolation and filter modes used) and give some details on data sample rate, output frequency of itnerest what results are you seeing and under what operating conditions.

We spec SFDR and NSD but not SNR so maybe there is some confusion in how SNR relates to the SFDR and NSD specs we include. If you are seeing too high of noise you could do better with the AD9778 or AD9779. If you are seeing too high of distortion then you may be limited by the analyzer or our specs may not be clear on our data sheet.

Can you send us a scren shot of the spectrum analyzer to look at?

Hi

I made a mistake. my DAC is AD 9779A. also:

sig. power: Full scale

Sig. freq. : 2.27MHz

DAC clock Freq. : 100MHz

UpSample : on (8X up sample rate)

Spectrum RBW : 30 KHz

- measurements with spectrum:

- peak signal = +2

- noise floor = -60

SNR = Psig - Noise floor - 10Log{(Fs/2) / RBW }

= +2 - (-60) - 10Log{50MHz/30KHz} = 30 dB !!

Regards

Your equation desen't seem correct

SNR = Psig - Noise floor - 10Log{(Fs/2) / RBW }

As RBW decreases this equation predict the SNR gets worse which is just the opposite

of what happens. When you narrow RBW the noise floor gets lower

We believe the correct equation is

SNR = Psig - [Noise floor - 10Log{(Fs/2) / RBW } ]

When you use this the SNR = 94dB. If you reduce RBW it predicts lower SNR as it should.

I am curious if you are testing with your own PCB or using our evaluation board.

If you are using our latest evaluation board the 2nd harmoic would be very large at the 2.26Mhz and would be coming from our transformer that was picked for using with output frequencies above 10Mhz. When I set up this test I got a lower noise floor using the Agielnet PXA analyser but the noise floor is also very dependent on the Attenuation used.

Similar to the last explination the noise floor decreases as you reduce resolution bandwidth.

so in this equation SNR= noise floor – 10Log10 [fs/2/BW] , the noise floor must be expressed as -

x dB and as the BW decreases the 10Log10 [fs/2/BW] will increase and effectively increase your SNR.