Am 07.01.2021 um 19:27 schrieb Kevin Grant: > > Hello Helge, > > I have been researching the LA2016 a little to see if I can help with > the Sigrok integration. In summary, I would say it cannot work on > Linux or Windows at present. > I agree. There are at least two problems, I found. I sent a pull request #112 with the fixes. I don't think the driver is now complete, but I was able to use the LA2016 on Windows 10.
> There are certainly issues in the protocol.c [1] file, for example > some FPGA SPI register addresses are wrong; CTRL_THRESHOLD should be > 0x68 and CTRL_TRIGGER should be 0x20. So, I don't understand how this > could have been tested, unless the manufacturer has recently changed > the fw and FPGA bitstream. > How did you found this info? I only found the problem with Set Configuration and the non-packed struct causing a wrong OUT transfer. I currently trust the implementation of Florian, but if there are further problems or bugs it would be good to fix it. > That brings me to the next issue; the bitstream extracted from the OEM > software using the python script [2] is different to the bitstream > downloaded to the LA by the OEM software. I have some experimental > code here to run the LA and found that while the LED would fade in/out > as normal the unit would not run correctly. I then extracted the > bitstream directly from the OEM software USB packets (it was > different) and my experimental setup ran fine with that. So, either > the python extractor code is somehow causing bitstream errors or the > OEM is obfuscating the on-disk bitstream. > > Adjacent to the board LED there is an SO8 IC, no useful markings. It > is a challenge-response authentication device, with one-wire UART > comms on pin 3 at 12900 baud. Some of the 'unkown commands' mentioned > in protocol.c are related to this device. One of those commands reads > a serial number as shown in the OEM software 'About' box. However, the > purpose of this device is to prevent their software working with any > cloned boards and shouldn't prevent any hardware functions. I > recommend the 'unkown commands' (bRequest 96) are ignored/removed. > This might be true. But it's more than a serial number in the "unknown commands/responses". The content chnages on the next run. There are also PWM output pins. As long as there is no way to control these outputs from sigrok, the corresponding enable and configure transfers should also be removed. > IMO using the OEM fw and bitstream is an interim workaround, and a > workaround should not be confused with a solution :-) It would be a > better user and developer experience if we had open solutions; the fw > is not too complicated (the FX2 is just passing bytes around, > controlling endpoints) but the FPGA is perhaps a little tricky. > A better interim solution would be a good documentation of the transfers currently used. > Anyway, if anyone on this list has an LA2016 (or LA1016) working with > PulseView, please let me know your setup and where you got the bitstream. > Well, I downloaded the OEM's Linux software and used the Python scripts to get the fw and bitstream files. Best regards, Helge _______________________________________________ sigrok-devel mailing list sigrok-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/sigrok-devel
