0
\$\begingroup\$

I’m looking to generate a RF signal sampled at around 10 GSPS operating above the 3GHz mark. In order to maximize data throughput, the signal will be modulated at that frequency (rather than mixing in a lower frequency baseband).

I have a working system at around 2.5 GSPS using a single AD9915, the parallel input of which is driven using an FPGA (which defines the desired DDS output for each sample on each clock cycle). The AD9915 allows synchronization of output across multiple ICs with a maximum sample rate of 2.5 GSPS.

I’m looking at the most cost efficient possibilities to get to a sample rate closer to 10 GSPS with at least 12-bit resolution (less is not an option). Firstly I’d be open to any IC which can already do this, I haven’t found one. Secondly I have another proposal to do this which would require a different type of multiplexing IC for which I’m not sure one exists...

The idea is to synchronize multiple AD9915s (supported, P34 of datasheet), independently drive their parallel inputs for modulation using an FPGA and then use some kind of very high speed ‘analog multiplexer?’ to sample each AD9915 independently to create a higher resolution output. The theory of operation is shown below, but the question is whether anyone is aware of an IC which can accomplish multiplexing these signals at these speeds while maintaining sync?

Illustration of Multiplexing Idea to increase sample rate:

Proposed Multiplexing Solution

Note: the above diagram shows using two AD9915s, I’m looking to use four but the same theory would apply.

Notice the the multiplexed output (lower chart) is at twice the resolution of the first due to the multiplexer running at 5 GHZ (GSPS).

I’m open to solutions to get the sample rate up. Perhaps another way to think about this problem is creating an arbitrary waveform generator. Most high-end OEMs seem to use custom silicon, this isn't an option for me due to low volume and a commensurately small budget.

Does anyone know how I could create the combined waveform (in the diagram) reliably at the sample rates specified, or does anyone know of any other solution which would get closer to 10 GSPS?

Footnote 1: I’m not looking for bench equipment, I’m looking to incorporate this into a relatively small device with control over how it is designed and produced.

Footnote 2: I’m aware that the AD9914 can go to 3.5 GSPS, but it only supports synchronization across ICs at 2.5 GSPS, and I’d like to get above 3.5 GSPS.

\$\endgroup\$
9
  • \$\begingroup\$ I do not know about the programming of the AD9914 to get them to be synchronized as you show in your first graph. However, the DAC outputs of these chips are current sources, so, when operating into low impedances, they can simply be combined in parallel to get the waveform shown in your second graph. There is no need for a multiplexer. The currents just add up. \$\endgroup\$
    – user69795
    Feb 9, 2021 at 18:34
  • \$\begingroup\$ Thanks user69795. However, adding them would not increase the resolution in the way desired. Yes I'd get a combined waveform, but not the desired one with the higher GSPS. Page 34 of the datasheet describes the synchronization, seems possible. analog.com/media/en/technical-documentation/data-sheets/… \$\endgroup\$
    – monibius
    Feb 9, 2021 at 18:38
  • \$\begingroup\$ Sorry, I misread your question. I thought you had arranged the digital stuff to be half a sample period off for the two DACs. Then just adding them would get you the kind of waveform you want. \$\endgroup\$
    – user69795
    Feb 9, 2021 at 18:44
  • \$\begingroup\$ It's a fair point, I've added a bit more clarification to the question. Thank you! \$\endgroup\$
    – monibius
    Feb 9, 2021 at 18:46
  • \$\begingroup\$ You can adjust the sync_in and sync_out delays of the slave devices. This will effectively interleave the samples. \$\endgroup\$
    – Lior Bilia
    Feb 9, 2021 at 18:48

1 Answer 1

Reset to default
2
\$\begingroup\$

Old thread, but the AD9166 can do what you want. It has a high speed DAC going up to 12 GS/s and the NCO can be digitally modulated with I/Q at several GS/s. So the data stream is not that much faster, but you can access higher frequencies through the upconversion.

Alternatively, both Analog Devices and Texas Instruments have high speed DACs at that rate, or you can go RF SoC.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.