[hpsdr] I need a little help here.

[Bill Tracey]

Hi Bob,

If PowerSDR is working then I'd say that's a pretty good indication 
all is well with the hardware.

I can't recall any real significant changes to the protocol or 
framing with one exception.  We did start having to force send a few 
command and control frames when we first come up to make sure Ozy has 
valid data on what clocks to use -- essentially priming the 
pump.  Basically what we do now is to send 4 sets of command and 
control frames to make sure we're telling the FPGA what clocks to 
use.  The relevant code is in JanusAudio\io-thread.c, function 
is  ForceCandCFrames.

That's about the only thing that comes to mind at the moment.

Cheers,.

Bill (kd5tfd)


At 04:16 PM 3/16/2010, Bob Cowdery wrote:
>***** High Performance Software Defined Radio Discussion List *****
>
>Hi
>
>I've done absolutely nothing radio wise for I think best part of a
>year.  I'm trying to find a bit of time to wind back in but the system
>is fighting back. It's amazing  how quickly knowledge dissipates when
>the brain is a tad old. I've had to use Windows 7 for a lot of my work
>recently and have to admit I quite like it. To ease myself back in I
>though I would tempt Ozy to talk to me on Windows 7, but it refuses. I
>think the HPSDR hardware has survived its hibernation as it works with
>PowerSDR which also proves my libusb installation is ok. I can load
>firmware and have updated Merc and Pen to the latest firmware. This is
>just a bit of C which was working on Ubuntu last time I used it which I
>have rebuilt on Windows.  I can claim the interface and write a few
>blocks to Ozy, or at least it says it wrote ok, but reading just gives
>me -116 which is a timeout, so Ozy won't give me any data. I even looked
>at the PowerSDR code and can't spot any real difference. I thought maybe
>the protocol had changed and my command bytes were wrong, but no, they
>look fine. I just have no idea why usb_bulk_read() just times out on me.
>Can anyone think of anything else that could be wrong? Oh, and yes I did
>find that the endpoint needs to be 0x86 as Linux lets you get away with
>missing out the direction bit.
>
>73's
>Bob
>G3UKB



 1268889245.0