EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: LordLynx on February 03, 2024, 12:27:55 pm
-
Hi everyone,
I'm currently designing a PDA type of style multiuse device for on the go (prop with Pi CM4 as base).
One part/extension for that is planned to be a USB dongle/device that can function as SDR, LA or oscilloscope depending on active "mode" and the software running on the connected device. I hope this idea seems doable, if not im open for advice and further ideas.
My current plan for parts looks like this:
- Cypress FX3 for USB 3.0 interface and Logic Analyzer (there are a few tutorials to be found online)
- complimented with a FPGA and some DDR3 memory to get the data rates for usage as an oscilloscope
- VNA functionality should be possible using the SDR hardware
Maybe you noticed the SDR missing on the list.
I basically don't have an idea on where to start designing one. While I would describe myself as good (enough) with PCB design and electronics I'm a total newby when it comes to RF. I read a few things online about the basics on SDR but I dont have any experience with it (yet ;)).
From what I understand i have two major "paths" to go when designing one: quadrature mixer (propably ADL5380, I'm still looking for a matching sine wave LO) and two dedicated ADCs or something like the AD936x/LMS7002.
What are the pros and cons of either one? I know I lose a bit of bandwith with the second option, but does it matter that much? Prices on the integrated components like the AD926x series are pretty high though. I would personally avoid a 120+€ IC if possible.
For my use cases I would want to cover the FM band as well as up to 2.4GHz. 5GHz+ would be nice but isn't a must.
Same with transmitting. I currently dont need it that much but useful to have.
If you have any advice/criticism or if I'm missing out on anything please share it.
Since it will propably be my daily driver for the next couple of years I dont want to use a "janky" solution with a RTL-SDR and the downconverter mod. It works and would propably be the easiest and cheapest solution for my current needs but i dont really know what I need in a few years.
And YES i want to design that by myself. Not primarily to save money, but to learn new things.
Thanks in advance for any answer.
-
That's a massive project you've laid out, especially if your writing all the software for it and don't have experience working with RF layouts before. I would strongly recommend trying to take it in as small, modular sections as possible. An example would be start with one of those devices and make it as a proof of concept - maybe using those cheap modules as building blocks to allow you to reconfigure, modify, and start with a functional base - and then get a feel for what demands you have placed on yourself in terms of hardware and software. If the modules you buy include the FX3 or an FPGA, even better, since you can get some experience working with them that should carry over more readily.
So aside from breaking it into several, much smaller, much simpler projects that lead up to considering the big one, let my advice be this: there is a number of reasons these devices don't really exist. You can do a lot with a fast digitizer and a lot of software tricks, but your concerns for layout, signal paths, signal generation, phase coherence and stability, and understanding of the characteristics of RF transmission lines and calibration are going to be critical to any sort of success with the VNA side (and will be critical to linearity of response of your oscilloscope and SDR.)
This is a BIG project, and to my knowledge, there are only a couple of model lines with commercially available instruments that can be both an oscilloscope, an RF receiver, an RF transmitter, and a logic analyzer with any good degree of usability in every mode, and even then only with costly option packages.
-
Thanks for your advice... I already thought it may be not working that good with all parts on one PCB.
I ordered a FX3 dev board yesterday, I will start from that with the logic analyzer (seems the easiest one to start with xD).
Do you have any recommendation on which path to go with the SDR?
The integrated systems like the AD9364 seem well suited and all, I just wouldn't prefer spending around 200€ on one chip in case it fails.
-
I suggest lowering your bar. Just to evaluate how to hook up a high speed ADC to the high speed MCU interface, and get a reasonable effective bit count out of the ADC at 50% of the data sheet’s ENOB.
-
Honestly, I would start with an FPGA development board that uses an FX3 and has an ADC/DAC built in (not specifically for a LA, but still a good starting point). You'll have some examples to get basic transfer working and then you'll have known-good hardware to start developing software that can display the data in a meaningful way, because even with tested hardware, the software and configuration are not a trivial task, even for basic functionality.
From there, if you were to start with a digitizer and then move towards a scope, you'd need to implement a hardware triggering system (could be a module to the dev board), and then eventually a frontend for ranging, protection, and preventing aliasing. At that point, you'd need to characterized the response of the system (phase and frequency) and then be able to apply that calibration to what you see on the output, and then you'd have the first semblances of a usable scope.
Honestly the scariest part of your project idea is that you want this device to be your daily driver. The engineering that goes into making the hardware as broadly reliable and fully characterized as it needs to be for general tasks is massive, and it will require both a lot of understanding of the complexities of the systems and a lot of accurate references to calibrate against, and that's only half the battle. The other half is to take all of this data and calibration information and make an interface that can display it conveniently, manipulate it in understandable ways, and perform the sorts of measurements that are most useful to whatever task you have. If you take a look through the menu systems of a benchtop version of any one of the devices you want to incorporate, you'll find dozens upon dozens of different ways to display and manipulate the same data coming through the same hardware, and getting those interfaces to be legible, manipulable, and actually useful is a whole team's task in a multi year commercial development cycle.
Making a device that can capture data and display it in a meaningful way to do some of the functions you describe is difficult, but can certainly be done. Making that same device work over a variety of input levels, display modes, supply meaningful measurement readings, keep stray signals sufficiently out, and all those other details that are demanded when it's the one tool you're going to be using for analysis.... it's in the ballpark of dozens of years worth of work for a single person.
-
Thank you for your answers.
@dobsonr741: I am trying... xD
@DaJMasta:
Is there any FPGA you would suggest?
Would a (small) Artix-7 be ok performance whise or do I need something like a Cyclone IV or V?