Charleston & Quisk - an update
tfox at knology.net
Fri Jun 18 17:37:46 CDT 2010
After yesterday's success, I've played some more with the software.
First of all, the 8201 sampling rate seems to max out at 192k for the
laptop I'm using with this. Yesterday, I mentioned that the audio
freezes periodically. Last night I accidentally noticed that if I go
to the "CONFIG" screen, those freezes stop, and the audio is clean for
long periods. The CONFIG menu overwrites the spectrum display, so it
hit me that the freezes are probably due to computer overload rather
than any data underruns/overruns related to rf samples. Then I found a
DEBUG in Jim's Quisk code, that when enabled shows the load on the
system. This confirmed my assumption. The loading was low when the
CONFIG menu was displayed, but as soon as I enabled any of the graphic
displays, the loading went to over 90%. My four year-old laptop is
choking on all the processing. This may partially be due to much of
Quisk being written in Python (semi-interpreter), but I thought the
speedy stuff was in C. I will research further. Needless to say,
faster sampling rates are out for the moment.
Secondly, James Ahlstrom (author of Quisk), is happy with my progress,
and has offered to assist. I asked him about interfacing between some
buttons that I've added to the GUI, and my Charleston-specific code. He
has given me a couple of ideas, and he sent me sample code to add a
slider to the GUI. Today I temporarily implemented what I believe to be
the poorer (but faster/easier) version of his suggestion, and now have
all the Charleston receiver hardware under Quisk control. I can bypass
the LPF/Preamp, or have them inline. I can set the Preamp gain to
Hi/LO, set the preamp slope to POS/NEG, and adjust the preamp gain from
min to max (slider). I already had control of the PGA gain inside the
8201 chip. I'm thinking about the best "long-term" approach to
implement these controls.
One more thing that I would like to implement is to show when the 8201
chip is being overloaded. I'm not sure that there is a signal coming
from the chip to indicate this. I've thought about watching the 8201
sample outputs, and setting a flag whenever they are at the 16-bit
maximum, but that only shows overload WITHIN the filtered spectrum, not
what the A/D inside the 8201 is actually seeing. With an only 12-bit
A/D in the 8201, this is an important issue. More research here.
Anyway, great fun. I believe this will be a good Linux platform to show
off the Charleston board. Is anybody working on modifying any
Windows-based SDR software for the Charleston board?
Any discussion at tacos?
More information about the Tacos