[hpsdr] Mercury sampling rate

[Chuck Hutton]

Phil:

Just what do you do when the FIFO is full or empty? Although it's a bit of
trouble, in the codec world there are several options: (1) repeat the last
sample, (2) interpolate between the last good sample and the next good
sample, and (3) extrapolate from the last x good samples.

Using these techniques, it's quite easy to make holes of at least 10 ms be
fairly undetectable to the average ear.

Of course, this doesn't help data demods that use audio as input although I
don't know that Mercury needs to worry about that.


And now, if we could circle way back to my original question about  rates
for archived data. (Web page says 250 ks/s, previous discussions mentioned 1
to 2 Mb/s). In an off-list email, you wrote:

"Sorry your question did not get answered. At present the only software that
supports HPSDR is PowerSDR. Hence all the features of that software are
available in terms of sample rates, recording of data and archival. PowerSDR
will allow you to record either I and Q data at up to 192kHz or the
demodulated audio output.  Recording of I and Q at 192kHz allows you to
replay the original recording and tune over the entire 192kHz looking for
signals."

That's not good news to those of us that were counting on the ability to
capture all or most of a band. Nor does it match up with other platforms
like Perseus (currently records 800 kHz), QS1R (recording not implemented
yet but transfers 1 to 2 MHz worth of samples to the PC), or the SDR-IQ
which records 190 kHz of spectrum.


I do think it will be important for Mercury to improve the recording
performance. I understand that there are many user communities - everything
from people who record bats to pseudo spectrum analyzers to surveillance t
hams/SWL's/DXers. Not all of them can be satisfied, but I think this is an
area that has proven important to many communities in the products that
already exist.


Chuck
___________________________________________________________________

Date: Sun, 13 Apr 2008 13:23:28 +0800
From: "Phil Harman" <phil at pharman.org>
Subject: Re: [hpsdr] Mercury sampling rate
To: <hpsdr at lists.hpsdr.org>
Message-ID: <014d01c89d26$81ed9bd0$0301a8c0 at phil>
Content-Type: text/plain; format=flowed; charset="iso-8859-1";
	reply-type=original

Most interesting discussion regarding Mercury sampling rates.  As Greg 
pointed out, we actually don't require a sound card  for Mercury since there

is a D to A converter on the board. You can also use the D to A on Penelope 
or a Janus board.

There was another reason for using an VHF clock that would divide exactly to

48/96/192kHz in addition to compatibility with sound cards.

When developing Penelope we wanted the CW performance to be the best we 
could provide. Since the D to A and A to D clocks for the receiver audio 
were from the same phase locked source then we would never run into the 
buffer over/under run issued that others have mentioned.

However,  PowerSDR expects that  the CW timing will be based on a 48kHz 
clock.  In which case any clock departure from exactly 48kHz resulted in an 
accumalated error that would place a 'blip' in both the CW sidetone and 
transmitted signal.

We did consider a 'fix' for this with Bob and Frank but concluded that life 
would be much simpler if we used a VHF clock that had an integer divide to 
48kHz.

Perhaps there  is a technically elegant way of solving this issue but we 
chose the simplest one.

73's Phil...VK6APH




 1208132084.0