Open lobsteroh opened 8 years ago
update ... reporting on various suggestions on how to fix ...
add 100nF capacitor between VCC and ground - does not change failure
is libusb-legacy the issue, can i switch to libusb?
Hi, libusb-0.1 (which is what libusb-legacy is pretending to be) is what's needed for the current LinkM library. libusb-1.0 has an entirely different API.
I'm not sure what's requiring you to wait ~500msec. I've been looking into it and it does seem like a driver issue, but I've not found anything telling yet.
hi tod, thanks a stack, in playing with this, the faults look really random. even if i issue the read commands at full speed i sometimes get 15 reads until it fails. it almost looks like something not being ready to communicate between linkm and PC, rather than on the serial output end of the linkm. can you confirm that your setup is running into the same issue, or is my system screwing things up. let me know if i can be of help in testing, troubleshooting or developing. be well,
ok, looks like a usb3 vs usb2 issue. When i plug the LinkM into a usb2 hub, and that into the usb3 port on my mac, then the linkm starts to play a lot nicer.
I know this ancient thread, but just dropping in to add a data that this seems to be an issue with some USB3+ ports. When I had the linkm connected to the USB-C port on my Macbook Air (USB3.1), it reported I2C comms issues. Plugging it into a USB2 hub and things work great.
Can probably close this one, if you're so inclined, @todbot.
oh no, i am running into communication errors with any repeated i2c reads. I get the same with 4 different kinds of i2c chips, and even the BlinkMs. the following example code
runs for a few times then throws IOException
if I make the Thread sleep, successful completions go up. A sleep of 500ms is needed to have this task start to run somewhat reliably. Don't have the option of reading only every half second. Any idea on how to fix it? - happens in my hands on MacOSX 10.11.4, LinkM, BlinkM, Java (build 1.8.0_77-b03), Eclipse