I'm testing the ADC12D1620QML-SP in DESIQ mode with a sample clock of 1.28 GHz, which provides a sample rate of 2.56 Gsps real. The sample clock is generated by an Agilent signal generator. The 1.28 GHz tone passes through an analog bandpass filter to remove harmonics produced by the signal generator. A second signal generator is used to generate 90, 1000, and 2400 MHz tones to be digitized by the ADC. Each of these tones is also passed through an analog bandpass filter to remove harmonics produced by the signal generator. The two signal generators are phase locked so the frequency of the digitized time series exactly matches the frequency on the signal generator front panel.
I measured ENOB at the three frequencies. The measured ENOB is fairly close to the ENOB specified in the data sheet at 90 and 1000 MHz, but is significantly lower than the data sheet at 2400 MHz. The 2400 MHz tone falls in the second Nyquist region. So, this tone shows up at 2.56e9 - 2.4e9 = 160 MHz in the digitized time series.
I waited until the ADC reached a steady state temperature before performing my measurements. I used the ADC Calibration feature before every data collection. I adjusted the "Q Channel Offset Adjust" register to minimize the spur at fs/2. I adjusted the "DES Timing Adjust" register to minimize the spur at fs/2 - fin.
Do you have any ideas why the measured ENOB at 2.4 GHz is much lower than the ENOB specified in the data sheet? Can you provide me with recommendations to increase the measured ENOB at 2.4 GHz?
|
Continuous Tone Freq (MHz) |
Data Sheet ENOB |
Measured ENOB Siggen Sample Clock |
|
90 |
9.3 |
8.8 |
|
1000 |
8.5 |
8.4 |
|
2400 |
6.7 |
4.9 |
Regards,
Scott
