I have noticed something 'interesting' in terms of the timing of IQ signals from the Flex 5000 with which I could use some advice.

I have been taking IQ data from the main Flex5000 receiver [RX1] and from the Second receiver [RX2] and feeding that data into the horizontal and vertical inputs to Linrad, in anticipation of dual polarity EME work with the Flex5000 used as the front end for Linrad.

I feed the main receiver FlexAudio IQ output to the H polarity Linrad Input, and the RX2 IQ output to V the polarity Linrad Input, using the PowerSDR "VAC with Direct IQ" function for the latter.  Sampling rate is 48000 for all channels.

I am doing this using Win XP SP3 with PowerSDR 2.0.16 with Driver version 3.5.2.8028 and PAL Version 3.5.2.8028.  VAC is version 4.9.0.1652.  I am using the WDM-KS driver on the PowerSDR VAC setup page.  Sampling rate is 48000 for both receivers, and for the Audio Repeater and the VACs.

There is a 540 msec delay of the VAC output from RX2 relative to the FlexAudio output, and so I use a VAC Audio Repeater between the FlexAudio output and the input to Linrad, in order to delay the RX1 output by 540 msec.  With this the RX1 and RX2 outputs are nicely time-aligned at time=0, and this 'baseline delay' between the channels is reduced to zero.

However, there is then a gradually increasing lag of the RX2 output relative to the RX1 output as seen at the input to Linrad, so that by 120 minutes there is a 130 msec delay of RX2 relative to RX1.  I have repeated this 'experiment' 3 times and within the accuracy of my 'scope measurements, the results are repeatable.

The way I measured the delay is by feeding a series of morse code '5's into the Flex 5000 in the 7 MHz band while everything is hooked up as above, and then feeding the outputs from Linrad [which will send RX1 to one speaker and RX2 to the other] into a dual channel 'scope, and measuring on the scope the delay between the RX1 and RX2 channels. I trigger the scope with the RX1 data stream.

After noting and measuring this delay, I wanted to see if the delay seemed to be occuring [1] in Windows 'after' the PowerSDR/5000 or [2] in the PowerSDR/5000, and so I made an IQ recording of the dits at time=0 and played separate copies of that back as a continuous loop simultaneously in 2 instances of CoolEdit, and fed the output of one instance of CoolEdit through the same VAC channel as I had been using for RX2, and I fed the output of the other instance of CoolEdit through the same Audio Repeater that I had used for the main Flex5000 receiver.  I left the Flex5000 running while I did this experiment, in case having it running might affect the results.  It was however no longer connected to either the Audio Repeater nor to the VACs.

When I did that, there was NO delay noticed in the output from Linrad for a period of just less than 2 hours.  Then a 30 msec lag in the channel that was not delayed in the previous experiments appeared.

I was not running other applications when I did these experiments.

I put a graph of the results for each experiment on the web at:
flex-vac-audio-delay.jpg
The xls file is at audiodelays.xls

I think the graph is well-enough labeled so that you can easily tell which line is for which set of data.  The Y axis is msec delay and the X axis is minutes after starting the experiment with RX1 and RX2 sync'd.

The striking thing to me is that for more than 100 minutes there was NO delay in the audio file experiment, whereas there was a gradually increasing delay in the PowerSDR/5000 experiment, with the increase in delay approximating 1.08 msec per minute when the PowerSDR/5000 was used.

I also did a separate experiment where I used the 7 MHz signal itself [rather than the RX1 data stream] to trigger the dual channel scope looking at the RX1 and RX2 data, and showed [as I expected] that the RX1 stream stayed locked to the RF signal, and the RX2 stream incrementally delayed, as noted in the text above.

Does anyone have thought on what might be causing this gradually increasing delay?  I tried the experiment both with and without the PowerSDR Diversity Control [Cntl-Alt-R] Enabled and it made no difference.  The two VFOs were of course always Sync'd in PowerSDR.

While the audio file experiment would suggest that the delay is in PowerSDR/5000, I can stop the Audio Repeater for a few seconds and when I restart it, the delay is back to the baseline 540 msec.  This would suggest to me that the delay is occurring AFTER PowerSDR/5000, somewhere in Windows.  My instinct is that the delay is a 'Windows Audio' thing.  But then why does the CoolEdit experiment produce no delay for nearly 2 hours when the instances of CoolEdit are run through exactly the same Audio Repeater and VAC paths as the two Flex 5000 receivers in the PowerSDR/5000 experiments?  I guess the effect may be an interaction between the PowerSDR RX2 IQ stream and VAC.

Thanks in advance for any thoughts.