Can HNers point toward a purpose-built system-on-a-board for data acquisition tasks? I keep seeing improvement in compute capability with Arduinos, RPis, BeagleBoards, etc. , but no movement in ADC precision.
Arduinos are wonderful for handing off to a student for a quick DAQ job, but invariably we require higher precision once they get things working, which requires additional engineering from our end.
Nothing at that price I can recall, but you're looking for a watered-down Red Pitaya? (A single board doesn't have enough 14 bit inputs, but in all other ways, particularly frequencies, it exceeds those specs?)
Really watered-down, yeah. It might be more appropriate to think, "arduino with higher-resolution inputs". It's really the Arduino ADC resolution that is a major headache for us.
Interesting new board from the BeagleBoard.org Foundation that designs the BeagleBone Black and other embedded Linux boards. Has a dual core ARM Cortex A15 processor and what appears to be a lot of peripherals like USB ports, ethernet ports, etc.
More good info here too: http://www.cnx-software.com/2014/11/07/beagleboard-x15-devel... Looks like the SoC is quite full featured and includes a couple Cortex M4 cores, presumably for realtime logic like the PRU's in the BeagleBone Black, and even a DSP.
The PRUs are a waste of perfectly good silicon until there is a C compiler for them. I haven't seen any project outside of TI's demos that uses these in any advanced fashion.
It sounds like an excellent idea. Completely separate processors on the same memory bus that allow higher IO speeds than the main processor. But since it can only be coded in assembly, Altera SOCs are much more attractive. Instead of cortex M4s, there is an FPGA on the same chip. Instead of making the interprocessor feature a crippled afterthought, the FPGA/ARM interface is central to the system, and the bus can be made as wide as practical. Since you can create the precise peripheral to suit your application on the FPGA, its kind of a blank slate.
Edit: As it turns out, there is now a C compiler for the PRU, 3 years after the hardware was released.
This has always kind of been TI's thing. I worked with a group that wrote H.264 codecs running on the IVA of the old OMAP2/3 and Hollywood chips. The IVA was straightforward DSP but the "bridge" method of getting that to talk to the CPU was just...awful.
Except that there is no way to scale assembly. You write every instruction. There are no companies that offer prewritten assembly for different common functions. Toga vendors offer all sorts of ip.
Understanding assembly and choosing to develop with assembly are two different things.
I see absolutely no correlation between learning assembly and learning verilog. They are about as orthogonal as you can get.
Often the IPUs will provide an interface to setup a PRU systeml... but, "so easily"?
While the functionality will appear on the spec sheet, it can be difficult to program. If you're using GCC, you'll want permission from your husband/wife/roommates before you start, because you're going to be occupied and accruing some significant brain damage. (I'm a hobbyist, not a professional.)
I mean, 10 UARTs? That just screams automotive. Can't think of anything else that would benefit from all this I/O. And, really, automotive is the only battlefield left for TI (and Freescale's i.MX). Apple, Samsung, and Qualcomm have locked up the handheld arena.
> Apple, Samsung, and Qualcomm have locked up the handheld arena.
Not really. Many low end to mid range devices use Allwinner, Rockchip, MediaTek or Amlogic SoCs. NVIDIA's Tegra series has also been somewhat popular and the two K1 SoCs are really nice.
Was describing the fates of TI and Freescale more than the second-tier SoC groups. The larger chipmakers aren't getting back into Apple or Samsung's shop any time soon. And making small design wins for third-world handsets doesn't interest them.
I have the impression that the competition of Linux implemented UEFI/SecureBoot just to lock Linux out from "their" boards. I wish the Linux community would stop to support UEFI systems completely. Linux pretending to be "Win 8.1" to outsmart the bootloader is not an acceptable solution.
The Linux community has a much better choice in embedded boards, and they should support the makers of those boards instead of buying UEFI products. There are a lot of boards which are powerful enough to be used as full blown office PCs. Soon we will even have ARM quad cores with 4 GB, dual HDMI and full SSD support. This is the way to go!
Arduinos are wonderful for handing off to a student for a quick DAQ job, but invariably we require higher precision once they get things working, which requires additional engineering from our end.
Price <$100, 4+ 14-16-bit ADCs at 1-2 kSamples/s, 2+ 10+-bit DACs, reasonably stable clock?