Using ADS5400 for undersampling

Hi,

As the input bandwidth for the device is listed at 2.1GHz while the max sample rate is listed at 1Gsps, yes - the device will support undersampling.  in fact, the typical AC performance specifications on pages 8 and 9 of the datasheet encompass the 1st Nyquist zone (@ 125NHz) through the fourth Nyquist zone (@ 1700MHz).

Regards,

Richard P.

Hi Richard,

Thank you for the reply.

Our signal is 2 GHz +/- 200 MHz.  Since the bandwidth is 2.1 GHz, we cannot use it to sample this signal; am I right?

Best,

Mark

Hi,

Well, you could if you take into consideration the attenuation that the signal would encounter in your band of interest.    See figure 24 of the datasheet, which is where the 2.1GHz number came from.   Input bandwidth is normally specified at the -3dB point, and on Figure 24 that would be at about 2.1GHz.  Your band of interest, being about 400mHz wide, would be more attenuated at the higher end of that band than at the lower end of that band, by a few tenths of a dB in loss of flatness across your band.  Maybe about half a dB loss of flatness across the band, looking at that figure.   

Also, since your signal spans from about 1800MHz to 2200MHz, you can't leave the sampling rate at 1000Msps, as then the Nyquist zone would split your bandwidth and you'd have your signal alias onto itself.   I'm guessing you would have your sample rate at just under 900Msps so as to put your bandwidth into the 5th Nyquist zone. 

Another option would be a much newer (and more complicated with its serialized JESD204b IO) device such as the ADC32RF45 that could be used to direct sample a signal up at your bandwidth and then digitally down convert and decimate so as to output just enough of a sample rate to faithfully represent your 400MHz bandwidth.  I could have others recommend some other devices from our portfolio that might be of interest as well.

Regards,

Richard P.

Hi,

Does TI have any device that has the same foot print, but higher bandwidth?

Best regards,
Mark

Hi,

No, that is the only device we have in that footprint. 

For the higher sample rates it becomes more impractical to use LVDS for the output as the pin count gets so large.  Not that input bandwidth has any direct tie to sample rate, but in general a process that would support a higher bandwidth would maybe support a higher sample rate too, and LVDS tends to cover about 1Gbps per LVDS pair or a bit more.   That is why many of our newer devices are going to the JESD204b output format, requiring the FPGA or whatever the data is going to to support the serial format.  I have asked around, and I was pointed to the ADC12J1600 as maybe the best device for your application if you can support the JESD204b interface.

Regards,
Richard P.

Hi,

Thanks for the reply!

If we want to stay with ADS5400 for the 2 GHz +/-200 MHz input signal, do we know to pay attention to other issues, apart from what you have mention regarding the flatness across the band and the sample rate?

Best regards,
Mark