Hacker News new | ask | show | jobs
by cedivad 1177 days ago
I designed this system (not the chip, clearly), so feel free to ask any questions!
8 comments

What's the software story for the VSPA DSP? Is that going to run user code, or a binary blob that does some filtering and sends I/Q samples to the iMX or USB3?

NXP doesn't share many details on the LA9310. Could just be too early, but it smells like a "you must convince a sales rep you're going to be a qualified customer and sign a stack of NDAs before you dare ask for a datasheet or instruction set manual" situation, which is unfortunate.

NXP are open sourcing a lot of the support stack for the LA9310 to go along with this product release, but I'm afraid they consider the VSPA DSP to be highly proprietary, yes.

Example: https://github.com/nxp-qoriq?q=la93

Seeing the spec I thought there is a Zynq but iMX6 looks even better. Can I get this board as an individual hobbyist instead of a company?
*i.MX 8M Plus, please :-) The 8M Plus is not that different from the RPi4!

Yes, of course, this will be sold to anybody and be very, very competitive, like in the order of ~$550 including the Granita board ($100 less for GranitaLite).

I stand corrected. IMX 8M plus. 550 that is a nice price compared to say LimeSDR. I will be sure to grab one when its available.
Can this device work as a medium/short wave receiver? How low can the frequency go? Like, if one has this device, would it also replace something like RSPdx?

What is the compatibility with existing SDR software?

It mentions multiple ADCs, can it work as a coherent receiver, similar to kraken?

The Granita board should already work down to 10 MHz. If you need to go lower, I (or someone else) would need to design a custom RFNM daughterboard for low frequencies. You could do direct conversion up to 150 MHz, so it should be a simple task and come with good performances.

All existing SDR software should be supported via OsmoSDR.

Yes, it should work as a coherent receiver. There are some questions about unmatched I/Q pair trace lengths, but from what I heard from the experts as long as the clock is coherent (which it is), we should be able to correct those in software.

> If you need to go lower, I (or someone else) would need to design a custom RFNM daughterboard for low frequencies. You could do direct conversion up to 150 MHz, so it should be a simple task and come with good performances.

If do ship oscilloscope daughterboard that you have on your site, would that work also as SW radio?

You'd probably want something way more sensitive than that, right? Something that goes well into the micro/nanovolts rather than millivolts. But it should be reasonably simple to do. Also the RTL board when designed with a good frontend should be a viable option.
are the chips available anywhere? https://www.nxp.com/part/LA9310X7S11AA#/ "Lead Time: 52" doesnt sound encouraging.
You need to work directly with NXP. Our first big batch will be here in September. We could have had them earlier but there are plenty of different moving parts, things to do and software to write, so September is a good compromise.

But yes, the long lead time is the reason for the waitlist!

If you can provide a board to Alan, the lead dev of SatDump (https://github.com/SatDump/SatDump) to ensure compatability I'll 100% buy this board.
We should have a fair number of boards going around in a couple of months (10-20), when the first batch is done and basic software is complete. Mostly as loaners at first as there will be more requests than boards, but everyone that contributes should definitely get one from the first mass-production batch :-)
Fantastic news. SatDump and SDR++ compatibility would be sweet
Will this be open source hardware / software like the HackRF One?? I see no mention of this so I assume not?
The schematics will be released publicly when the product is shipping, but it won't be OSHW. I don't want to end up like the HackRF, where one small company is burdened with supporting an ecosystem of what is mostly clones. The HackRF designer himself said he wouldn't do it again.
The board layout triggers my OCD :)

Why the weird angle of those main ICs?

Usually it's 45 degrees or parallel to the board.

I routed the i.MX block like a normal human being, then rotated everything 23 degrees to (1) avoid the fiber weave effect and (2) not collide with any of the mounting holes, as that placement needs to be symmetrical under the slots for the two daughterboards.

I get that it's triggering people, I get PTSD thinking about needing to touch that routing as well. Altium doesn't support arbitrary-angle differential pairs, so this was a huge mess. Next time I should just ask the PCB manufacturer to rotate the fiberglass sheets by a few degrees instead.

I love the curvy traces! It gives me the warm feels about old school handmade rubylith films. Don't change a thing!
Things that are placed non orthogonally is a sign of high speed black magic to me.

Edit: the other thread shows me writing a reply at 2am this morning... Pretty sure I didn't dream walk just to post on HN...

Any chance to get a direct PCIe connection to plug this into an x86 host?
It would need to be a different design entirely.