On the other hand you have only one single strong argument: community that knows PC-programming languages better than VHDL/Verilog. - True. But is it really showstopper?
I don't view it as a "showstopper" exactly. Simply put, the more people an open source project appeals to the more likely it is to be successful. You can argue that a software defined scope would be less attractive to EE types like yourself, and that could certainly be true, but the numbers are what matters. The head count of people who know how to program in C or Python is a much bigger number than the head count people who know how to program an FPGA, so if only 1 in 1000 (0.01%) of C/Python people are interested, that's still far more individuals than if 1 in 20 FPGA programmers (5%) are interested.
Hi-performance 1Gsps sampler scope is not a kid's toy. Someone shall implement PCIe bus PHY with all the bells and whistles... and not only that. In my opinion/experience PCIe is not that easy task at all. Not to mention PCIe-capable FPGA price. Sample processing and storage is much more simpler task than PCIe PHY.
Probably not as expensive or hard as you think. When I think of hardware to implement a PCIe capable scope, I'm thinking of something like an Artix-7, so between about $100-$200 US. Xilinx has a PCIe "integrated block" supported on this series, so it's possible to just use that with supporting components for PCIe... basically a shrink wrapped solution. An even quicker alternative would be to use an FPGA module like a Mars:
https://www.enclustra.com/en/products/fpga-modules/mars-mx2/... in which case only a carrier board needs to be designed with the supporting chips like the ADC and an SODIMM socket for the module. Single unit price on these is more like $250-$300 US, so it's more expensive than a complete DIY, but it's also easier.
Probably the best way to go for a digitizing card is to specify hardware capabilities and interface methods for the software and permit multiple hardware implementations anyway
Kind of the path that Sigrok takes, being hardware agnostic. That way is even better for a software defined project, because you can completely toss the previous generation of hardware if some great new technology comes out and keep most of your feature set.
Well... huge pool of linux programmers is not enough. They shall be interested in the final product too.
Yes, but if the project is defined mostly in software then the card will attract developers for applications other than oscilloscopes.
The less people who could actually use product (laptop/minipc owners and those who do not like PC on the desk facing backwards) - the less support it will get.
That's true if you assume that the card wouldn't be usable in a laptop or mini PC... if the card is low profile it probably would work in mini PCs, and even if it's not, then a solution like what gets used for GPU crypto coin mining could be used - essentially a flex based extension for the PCIe slot that runs to an add-on enclosure for the card, like this:
http://www.overclock.net/t/1427731/pci-express-extender-cables-benchmarkedAs far as the PC facing backwards, it's not all that hard to get extra long probe cables (this is the solution HP used for their mainframe based Oscilloscopes) or a front facing breakout box could be designed to provide front BNC connectors. It's extra cost, of course.
Thinking about it, though, it probably makes sense instead of the above to use an external PCIe interface like Thunderbolt-3... it already exists on many systems, has plenty of speed, and if you use USB-C over TB3 you even get power delivery over the cable. The scope digitizer "Card" would be a module in a box connected via thunderbolt to the PC. One small box with BNC connectors and one cable to the PC up to 2 meters long.
None is suitable for PCIe sampler. MiniPC does not have PCIe slots nor Thunderbolt, but dekstop tower PC is standing on the floor too far away for scope leads to reach my table. I am not going to put that huge box on the table facing its back to me just because there is/could_be open source scope PCIe adapter which costs 2x my standalone 1Gsps scope, Sorry.
So you'd be unwilling to buy a $50 US PCIe card to put in your desktop tower PC to allow use of an external digitizer in a box on your table to which you attach leads, costing something like $400-$500? Maybe you'd need to wait until the price comes down?
Reality check: does anyone besides erikg want open source PC-based scope which costs more than "made in china" standalone? Please share your opinion.
Opinion gathering is ok, but note that we're not really voting on something here. I'm just writing this to learn about peoples' opinions and thoughts on this subject. If I kick off a development project for a digitizer like this, you can be sure you'll see an announcement from me. Right now I'm really only capable of writing the software for it, though. Still learning hardware.
Thanks!