RE: [Discuss-gnuradio] Hi

[Philip Mackenzie]

Thanks for that info.

I've tried your code with my hardware, it worked really well. I changed the
tuning parameters, etc in the microtune_fm_demo example and its running at
~30% CPU time on a P4 2GHz. Sounds good too!

Unfortunately I think you're correct about windows virtual memory control,
although I'm trying to see if theres an easy way to get around this. I'm
using the computer boards driver at the moment and its sort of restrictive.
E.g. you have to call a special function to 'copy' the data from kernel to
user space. Not good at all for swradio.

That Microtune unit sounds interesting.  My IF amplifier is limiting our B/W
to only 300KHz at the moment. This allows us to use bandpass sampling at
2MHz to obtain a nice signal, but its still quite restrictive. I think I'll
try to get a microtune unit instead. Are you able to fill the full 12 bit
range (1 or 5 volts) from the output of your tuner? Have you been able to
use an antenna on the unit to get live signals or are you feeding it a test
signal?

Philip.


-----Original Message-----
From: Eric Blossom [mailto:address@hidden
Sent: 31 January 2002 22:59
To: Philip Mackenzie
Cc: address@hidden
Subject: Re: [Discuss-gnuradio] Hi


On Mon, Jan 28, 2002 at 08:36:52PM -0000, Philip Mackenzie wrote:
> Hi,
>
> I just came across the gnuradio project - I think its great. I'm a
graduate
> student in Ireland and we've been working on our own software radio system
> for the past year. (http://ntrg.cs.tcd.ie/swradio.php)

Thanks for the pointer.

> Some quick details: For our hardware we are using an AR8600 and a 10.7 MHz
> IF amplifier we built. Without knowing about gnuradio we also chose the
> PCI-DAS4020/12 for acquisition - (good board but it can crash the system
> sometimes).

Nice board.  Best bang for the buck.

> On the software side of things we are developing on Windows
> 2000. Our system is written in C/C++ and we have our own component
structure
> for coding up radios. There are some early papers on our site which may be
> of interest and more substantial ones will be available soon.

Thanks.

> I'm going to try out your code with our receiver, etc. I got the
impression
> that some people are using some type of cable modem. Does anyone have any
> tech specs on this? Has anyone has used any portion of the code on
windows?

Let me know how it works for you.

We're currently using the Microtune 4937 DI5 cable modem tuner module
as our front end.  The module has the output IF centered at 5.75 MHz
and it's about 6 MHz wide.  We connect it directly to a 50 ohm
terminated input on the PCI DAS4020/12.  Works pretty well.  Tunes
from 50 MHz to about 890 MHz.  It's got an I2C bus interface which we
bit bang using the PC parallel port.  We control the RF and IF AGC
using the DACs on the 4020.

Being "free software" kind of guys, we haven't tried to build it under
windows.  I wrote a driver for the 4020 that DMA's directly into user
space under linux, and we're using that as the primary high speed
input path.

The GNU Radio code takes serious advantage of the linux/unix mmap
system call to create circular buffers by mapping a given piece of
memory in at two consecutive virtual memory locations.  This may be
hard to duplicate under windows.

Assume you want a 64K circular buffer.  We allocate a 64K chunk of
shared memory, and then map it in to the user address space twice, at
immediately adjacent virtual addresses.  This hack works on pretty
much every unix/linux box running on a machine that does not have a
virtual mapped cache.

Please keep me posted as to your progress.

We are currently building an off the air (H)DTV receiver.  In the San
Francisco bay area there are now about 8 stations broadcasting digital
TV.

Eric