Posted on 16 April 2021

AD9363, Part 1

I’m designing a prototype Mode-A/C/S 1030MHz receiver so I can start testing demodulation algorithms (with real signals) and implementing them as gateware. In a previous post I mentioned multiple candidate receiver topologies and for simplicity’s sake, I’ve decided to go with the AD9363 approach. The AD9363 might not give me enough dynamic range to meet the standard’s requirements, but it’s enough to get started with algorithm development/testing.

I’m planning to use an ECP5 FPGA for the baseband processor because they’re well-supported by the yosys/nextpnr open-source FPGA toolchain. I already have an ECP5 evaluation board (the one from Lattice without the PCIe edge connector), so I’m currently designing an AD9363 board that can plug into that evaluation board’s pin headers.

Today I planned out the power supplies for the AD9363. I’d just copy what the reference manual suggests, but I can’t do that exactly because this ECP5 board lets me set the FPGA’s VCCIO to 2.5V but not to 1.8V. Honestly I don’t know to develop a preference for VCCIO voltage; given the AD9363 and the ECP5 both seem pretty flexible about it. Is a smaller voltage better? I don’t know. Also I want to tap off the 12V because the 3.3V/2.5V regulators on the ECP5 board don’t seem to have lots of margin for added loads.

I’m thinking of feeding that 12V to a two-output low-noise switching regulator LT8653S to generate 1.8 and 2.5V. The 2.5V gets used for the interface and GPO (general purpose output) supplies on the AD9363, which aren’t that critical in terms of noise sensitivity.

The 1.8V doesn’t get used directly but instead I feed it into two separate ADP1754 low-noise LDOs, and each of the LDOs feeds a different set of power supply pins on the AD9363, like the reference manual suggests. Also I’m thinking of doing what’s done in the ADALM-PLUTO and using an ADM7160 ultra-low-noise LDO to power the oscillator.

Actually, I’m sort of reconsidering using that evaluation board due to signal integrity reasons (this Analog Devices webpage states that the CMOS drivers in the AD9363 are purposefully weak and recommends using LVDS instead, and I have no idea how to use LVDS) and instead making my own board from scratch with an ECP5 and an AD9363. It’s not like there’s a lot more to keeping the ECP5 happy than there is for the AD9363 – it’s just power supplies and decoupling capacitors. I think I’ll do it, it looks like a fun challenge!

Also I’m using a draft version of DO-260A (which I found on the internet) as reference for all of this. If you can get me DO-260B and/or DO-181E (or the EUROCAE equivalents) I’d be very happy. RTCA charges a lot for these documents – to the point that it’s a lot cheaper to actually pay membership fees (which lets you have the documents for free) which I’d absolutely do except they say “membership is restricted to organizations that are doing business in aviation” and I’m not sure I qualify as that.