[hpsdr] Question on Janus access with upcoming library code

[Bill Tracey]

Catching up on mail - comments in line below.

Cheers,

Bill (kd5tfd)

At 01:13 PM 2/20/2007, Philip Covington wrote:
>***** High Performance Software Defined Radio Discussion List *****
>
>On 2/20/07, Ray Anderson <ray.anderson at xilinx.com> wrote:
> > ***** High Performance Software Defined Radio Discussion List *****
> > A question on the following for the list software Guru's: would it be
> > feasible to write a Windows or Linux driver for Janus/Ozy such that the
> > hardware appears as a soundcard to the O/S.
> >
> > From Bill's comments it sounds like the Janus library he is alluding to can
> > be compiled into an application to allow that particular application access
> > Janus. Could this same library (or a variant) be integrated into 
> driver that
> > could conceivable be used with other non-modified applications (say Rocky,
> > or KGSDR)

I do not know what the requirements are for code to be embedded into 
a Windows driver.  The code I've got now to talk to Janus/Ozy is user 
mode C code -- talks to Ozy via libusb.

> >
> > Thanks for any info!
> >
> >
> >
> > -Ray     WB6TPU
>
>This is off the top of my head so take it for what it is worth...
>
>It should be possible to make OZY/JANUS appear as a soundcard to
>Windows by exposing certain interfaces that conform to USB sound
>devices.  This would have to be done in the OZY FX2 firmware.  The
>problem with this is that there are now certain command & control
>functions that are send in-band with the I/Q data stream.  This would
>make the above not possible since the data stream needs to be decoded
>and the Windows USB sound device driver knows nothing about this.  If
>the command & control data were moved out of the I/Q data stream and
>sent to/from endpoint 0, I think it would be possible.  Bill T. might
>want to comment on the feasibility of moving the command and control
>data out of the I/Q stream.  I think there is some concern about CW
>latency or PTT issues in doing this...

Moving the C&C data out of the IQ stream is probably feasible .. if 
CW latency then becomes a problem one could go with one of the 
alternate connection possibilities (eg serial port) for the 
key.    Moving the C&C out of the IQ stream would require changes in 
the FPGA, FX2, and PowerSDR code.

The approach Phil is talking about above I believe would make 
Ozy/Janus appear as a generic USB sound card -- essentially plugging 
into some base Windows USB sound device support. Not sure how 
flexible this support is in Windows.  Janus/Ozy would need to appear 
as a 4 in 4 out device, potentially with different sampling rates and 
bit depths on the input and output sides -- not quite sure if you'd 
be able to do it as an extension of the existing Windows sound support.

I'd also think it should be possible to write a custom Windows driver 
that surfaced Janus/Ozy as a sound card and kept the existing 
protocol with the C&C data embedded in the IQ stream -- it would end 
up being a total custom driver so would require more work I"d suspect.


>It probably would be easier to write a driver for Linux - maybe
>someone with experience writing USB sound card drivers for Linux will
>respond.

May or may not be easier for Linux -- can't say I know about writing 
sound card drivers for either Linix or Windows.  All of the code is 
available if someone want to take it on.

>73 Phil N8VB



 1172210218.0