I’m not sure if you could accurately synchronize then over USB, but if you had a common clock signal split off to each one you could certainly sync up the signals after capturing them. That’s a feature that exists in some logic analyzers already, but it will probably be a while before mine does that (seeing as I haven’t even started building hardware!)
For that many signals you’re probably better off using a dedicated logic analyzer connected over USB rather than a framework laptop with mini logic analyzers in three of its expansion slots.
And leaving out alott of other folks. Theres always a “better” way, or faster, or more sync options. I could also just save up, and buy a certified, calibrated, official well known brand logic analyser. But having small dongles would be cool. Not saying no to a pcie version, but would that not also complicate the interface and driver requirements?
PCIe is really not necessary. The expansion cards support Thunderbolt 4, which means you can theoretically squeeze 40Gbps out of one of them. That would be enough for up to 40 channels @ 1Gsa/s. That being said, squeezing a 40 channel logic analyzer into an expansion card is not going to happen any time soon.
after checking rp2040 documentation, I confirmed that rp2040 does have support for USB2.0 (12Mbps) just that it is not “PHY” (I still don’t understand this term, btw) and it can be done on a GPIO using the Programmable IO (PIO) controllers. example
I am interested to something like this since I plan to run the rp2040 as a pure GPIO controller and handle all protocols & applications on the framework, effectively offloading any computation from the mcu. I am currently working on a cpp+arduino version but certainly interested in remaking it in cpp+PIO assembly for best performance.
This would therotically provide ~750KSa/S on 16 pins and 6MSa/S (I doubt if this is possible as rp2040 runs 16MHz stock, so at that speed every cycle matters) on 2 pins which is more than enough for me who are really only using a logic analyzer for stuff like I2C.
*As a disclaimer, I am not undermining your project but rather trying to add a low cost alternative to it.
Might be easier to just repackage the good old fx2 as an expansion card if a really low cost version is required, the software support for that already exists
I won’t decide in a final price until I build a few prototypes and finalize the hardware design. But it will likely be around $160 for the logic analyzer + test leads, and another $60 or so for an optional set of high quality clip on probes.
Yes you are right, I was going off the crystal frequency. Though I have to point out that is the core frequency and does not necessarily mean IO can be done at that frequency, I haven’t read the clock section of the documentation yet.
Also, if anyone is interested in this direction, I ordered parts for 15 of those rp2040 cards for about $120 (it is hard to approximate cost as component price very greatly to quantity and what component is used, surely I haven’t optimize my component choice yet)
Side note: if you are to make 1-2 of this yourself, I recommend you to buy a pi pico and salvage parts of it, that would probably be cheaper than buying new parts from, for example, mouser
And even the 133 are just a sugestion, rp2040s run just fine at 240hz and above and with the pio block you could do sampling at close to that and dma it to a buffer.
But you know the 8$ fx2 logic analizers allready do that and have native sigrock support and stuff.
Yeah there are definitely cheaper ways to build this than what I have done. I mostly ordered the component that I know works (by matching component (flash&LDO mostly) to an existing board) rather than the cheaper ones since it is way more expensive to fail, redesign, and reorder it when I am expecting to producing less than 10 of these
Also, this is not BOM cost but rather an approximation of how much I spend for it. In fact, I brought a bunch of extras (including enough part for 30 rp2040, lots of extra passives and more then $50 of other pcbs) for other stuff totaling to around $240. Because the price of these thing just don’t scale linearly, I choose to give $120 as a safe estimate/budget for whoever want to attempt this.
a bit of rant: just shipping for the pcb and the components alone accounts for $35+, which is more or less fixed no matter how much boards you order (as long as it is under 50 i guess). This is usually leftout or averaged in BOM cost but matters way more for people who just want a handful of these boards, which is why I would prefer a estimate like this.