[hpsdr] Recent PowerSDR versions and DPCs - curiouser and curiouser!!

PETER BUCK peteg3lwt at btinternet.com
Mon Oct 22 11:55:42 PDT 2012


Hello All,
I have been wondering for some time if anyone else who uses PowerSDR regularly has the following issues. My setup is Mercury, Pennylane, Metis, Alex, Pennywhistle and Excaliber with current firmware and PowerSDR dated 10/16/12. Computers are an older dual-core E6750 2.6MHz, and a dual-core mini-itx also 2.6MHz. Both have 4MB RAM and run XP. I have run dpclat while operating with either computer, and both systems show the same weird behaviour. With the display set as panafall, at irregular time intervals sometimes up to several hours, the panascope display freezes and receive/transmit audio drops out. As far as I can recall, this issue did not arise with versions of PowerSDR in use over a year or so ago.
As expected, dpclat revealed a huge (off scale) dpc peak when this freeze occurred. Sequentially killing unused drivers had no effect, but if PowerSDR was not running, these huge peaks never occurred. With PowerSDR running normally, the typical dpc was about 50microsec, rising sometimes to about 1000microsec for short periods after band change and/or mode change. Even so such changes did not give rise to significant clicks and bangs in the Rx audio or display jitters. However, sometimes when PowerSDR was started, or after a band change, especially in USB/LSB mode, dpcs would rise to some15000microsec and stay there. Normal operation could continue in this state, but I suspect that any additional polling while in this state could trigger the freeze-up. Band and mode changes when in this state were usually accompanied by audio clicks and bangs, and were much slower than normal.
Unfortunately these problems are not 100% replicable, but are annoying none the less. Is it possible there is an odd bug lurking either in the driver, or in the bandswitch/mode switch parts of the PowerSDR code? - or do I simply need to upgrade my computer?
I have also noted the recent change in behaviour of the bandswitches when saving last used frequency which was raised recently.
Pete, G3LWT



More information about the Hpsdr mailing list