[hpsdr] V3.2.7PS issues

[k2cb at comcast.net]

I have been experiencing some latency issues since upgrading to V3.2.7PS. On TX, the transmitted audio is delayed upwards of 500ms. RX tuning (qsy, frequency changes) also seemed sluggish, in that the hardware was delayed in changing frequency from when the actual software indicated a tuning change. 



In addition, when using standard (analog) mic audio from a microphone straight into the 3.5mm jack, there was an audible echo, as well as distortion, on the transmitted audio. If I switched over to  VAC transmit audio (using a Steinberg UR22 XLR to USB interface), the transmit audio was clean, albeit the delay was still present. 



I was not experiencing these issues with prior releases. 



Resetting the database did initially appear to clear the latency issues, and the tx audio echo, but I believe all the reset did was mask the underlying problem. 




After the database reset, I went through each setting one at a time, restoring values which I used in V3.2.1,  until I was able to duplicate the issue. 




What I found was that the DSP audio buffers for transmit and receive defaulted to 2048. I was previously setting them both to 16384.  As soon as I raised the transmit buffer to 16384, the transmit latency and echo returned. As I stepped the buffer setting down one increment at a time, the echo and tx audio latency was proportionately reduced. 




Next I tried the receive audio buffer. Similar results. At the default of 2048, the receiver would change frequency almost instantaneously while tuning up and down. When I increased the buffer to 16384, the latency in frequency tuning (qsy) returned. As I decreased the buffer size, the tuning latency reduced proportionately. 




Now the strange thing is that with V3.2.1, I can run the buffers at 16384 with no ill effects, and gain the performance advantage, but with V3.2.7, running the higher buffers results in the aforementioned issues. 


So at this point I would have to rule out the pc, unless I am missing something. 




Another two anomalies I noticed while experimenting this evening: 




1. The 20db mic preamp can be turned on, then turned off. But once you turn it off, you can no longer turn it back un unless you reset the hardware (ANAN). In previous versions, I could turn the 20db preamp on and off repetitively while testing mic levels. 




2. When running full duplex, the panadapter indicates the transmitted signal fine. When you switch to panascope mode (useful when experimenting with tx audio), the scope works, but the panadapter shows a frozen image of energy. If I turn off full duplex, the panascope functions properly, both the scope and panadapter portions. I don't recall if this occurred in earlier versions, but it does occur in both V3.2.1 and V3.2.7. 




I also ran the dpc latency tester, and the results are in the 115-130us range. 





Eric 

K2CB
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openhpsdr.org/pipermail/hpsdr-openhpsdr.org/attachments/20140129/d2774a2b/attachment-0003.htm>