Hello Giovanni, we are pleased to receive you today. You are FPGA developer and along with what has been exposed lately on this blog, some serious innovations were brought in and around the on-board data processing units (DPU) or more commonly called FPGA’s that the M9703A carries (the M9703A counts no less than 4 Xilink Virtex-6 FPGA’s each unit supporting the on-board signal processing of 2 channels).
Giovanni, before we start could you please briefly describe your background?
Well, I have a Master of Science in Digital Electronic from University of Pisa in Italy. I have since the beginning worked with FPGAs and embedded systems at first in Italy in a start-up company where I was in charge of the development of sensor interfaces, then in the Netherlands at European Space Agency (ESA) mainly focusing on designs for space applications. I joined Agilent in Geneva in 2007 working on our first generation PCIe (PCI express) board and then on the new PXIe and AXIe platforms. Today I am mainly working on the deployment of new on-board processing functions looking to systematize this process across our high-speed digitizers.
Could you explain to us why Agilent ADC team considers the role of the DPU’s more and more important?
The DPUs are responsible for controlling the module functionality, data flow, and obviously real-time signal processing. On-the-fly processing is becoming more and more important, as requirements on both ADC resolution and sampling rate are continually increasing, whist the available data readout bandwidth remains limited. To address this problem, techniques may be employed at the digitizer level to reduce processing time and optimize data transfer.
How has Agilent then addressed the problem?
Real-time data processing is reached by either getting of-the-shelf (OTS) commercially available firmwares, or by implementing custom algorithms. Agilent provides a variety of IP firmware real-time processing, such as fast Fourier transform (FFT), averaging, peak detect, or digital down converter (DDC) option on the M9703A.
So it seems that Agilent has actually developed quite a few IPs. Why is that? Wouldn’t be easier to buy an existing signal processing IP’s?
Well the answer is simple. It’s impossible to find a commercially available IP which contains the right set of functionalities which address the requirements of our customers when using our digitizers and which is easily adaptable and customizable to fit our systems and the frequencies involved. If we want to guarantee the best real-time performance at the multi-GS/s data rates our digitizers platforms are capable and at the same time fully leverage their exceptional hardware capabilities in terms of trigger positioning or signal fidelity this is the only way to go.
We heard that you are really proud of the latest firmware Agilent developed. What does it do and what are the main advantages of implementing it?
The DDC allows tuning and zooming on the signals to be analyzed, improving the dynamic range, reducing the noise floor, extending the capture time, and accelerating the measurement speed. This exclusive IP algorithm provides very powerful and flexible digital downconversion on all 8 channels. The filters and local oscillators (LO) are synchronized to maintain constant phase and timing relationships to allow phase-coherent post processing. The DDC allows isolating the signal of interest from other signals in crowded spectrum, and improving the dynamic range as the integrated noise is reduced, increasing the SNR and effective number of bits (ENOB).
Does Agilent consider to open their digitizers’ FGPA’s to customers signal processing algorithms?
Of course, Agilent high-speed digitizers’ team does. Any customers who would have interest in doing so should contact us. As I mentioned above, we believe that the on-board processing is the future of high-speed digitizers. Now we need to make sure that the deployment of real-time processing algorithms does not deteriorate the hardware performances. That’s where I spend most my energy and the successful DDC deployment on the M9703A AXIe digitizer is the proof that we can do it.
Thanks a lot for your time Giovanni, we are looking forward to hear about the next FPGA’s developments on high-speed digitizers with you in a near future.INTERESTED TO LEARN MORE ABOUT THE CUSTOMED ALGORITHMS DEPLOYMENT ON AGILENT DIGITIZERS? WRITE US AN EMAIL TO GET IN TOUCH AND GET MORE ON THIS IS EXCLUSIVE INFO. Tags: m9703a real-time processing share