Objective Development Forums

I'd like to have 99% data before the trigger and 1% after. The relevant things happen before the error is detected.

Yes, you can sense the USB power. If you check for USB activity, the LED would go off when the hosts goes into suspend mode and not go back on when it awakes because there's no new enumeration.

You should also turn it off when USB Reset is detected.

I use #define UTIL_INTERRUPT(signame) \ void signame (void) __attribute__ ((interrupt)); \ void signame (void) and declare the interrupt as UTIL_INTERRUPT(XXX_vect) { ... interrupt implementation }

Try to omit the zener diodes. Most hosts accept 5 V levels. If that does not work, add two series diodes in the supply to drop the voltage to approx. 3.6 V. That's the cheapest solution.

The first error indicates that Linux insists that you claim an interface. That's somewhat strange, since it should be possible to query the device's string descriptors (through endpoint 0) when the device is otherwise busy. But you can change the code accordingly: int retries = 1, usbConfiguration =...

I guess you want to send a feature report to the device. Since your device is a keyboard, a operating system driver is already attached to it. This may cause some problems. I have heard about problems with newer Linux kernels and HID. It seems that you must claim an interface in order to communicate.

You need to supply your own configuration descriptor since the default one does not include additional endpoints. See AVR-Doper for an example.

Have a look at http://www.csd-electronics.de/. Shipping is even cheaper than from Conrad and the prices are much lower.

Sorry for the delay, I was (and still am) busy with other projects. I finally had the time to look at the data you set up for download. There's one full packet visible before the trigger point. And I can see that your logic analyzer is capable of recording more than 1 millisecond worth of data at th...

Should work as well.

Would be great to have a link in our projects section when the project is ready! Please keep me updated!

Sorry, will get back to you later. Does this file contain some USB messages BEFORE the trigger? I'd like to have as much as possible. A sampling rate of 10 MHz should be sufficient and gives us more history.

It depends on whether you implement USB suspend or not. If you implement suspend, you need to wake up your CPU from sleep mode with the interrupt. This works only with INT0 - 3. Otherwise the additional synchronization with the I/O clock should not matter.

Did you try this on a different host? CDC mode may be unreliable, depending on MANY factors. The data length and content may be one of them.

If you can reproduce this with a USB 2.0 hub inserted and with a different host, then please send us the offending hex file to the feedback address.

You have two options: (1) You use custom USB requests with libusb-win32. This requires an installation of libusb-win32, though. (2) You use feature reports to transmit your data. In this case there's more than one type of report (the x/y/buttons report) and you will have to use report-IDs to disting...