Just to confirm same analysis here :
Stream is not ETR290 compliant. PCR accuracy is +/- 800us instead of +/-500ns. It is not very smart but should not be too disturbing.
Profession Dectek StreamXpert is used to do that, which is equivalent to your R&S for this kind of measure.
The main issue, is effectively the -500 ms of PCR/PTS (It is not measured by ETR290). This means that decoder is asked to display a picture which is not yet arrived (500ms late). Strategy of decoders to handle this issue depending on the decoder.
VLC for example, try to increase the input buffer in order to display later pictures :
"
main warning: picture is too late to be displayed (missing 281 ms)
main warning: picture is too late to be displayed (missing 259 ms)
main warning: picture is too late to be displayed (missing 250 ms)
main warning: picture is too late to be displayed (missing 230 ms)
main warning: picture is too late to be displayed (missing 210 ms)
main warning: picture is too late to be displayed (missing 190 ms)
"
SF8008 could have also this strategy..but it is a workaround. As clock is not precise, we could then have some issue between audio/video synchronisation.
Professional decoders (Tandberg, Scopus ...) doesn't allow for such errors and do not display a picture at all.
This issue has already been seen regularly with 2MS beacon, and was recovered by a reset of the encoder/modulator.
It seems that at 1.5MS the issue could happen more regularly.
Right now, it seems that the beacon is performed by Raspberry looping a video and output it to HDMI whish is then H264 encoded and modulated by SR System. The system could drift over time and could explain such errors (long term stability is not so easy in DVB).
We hope that we could make analysis just after a reset to confirm this analysis.
As Mike said, we are a community which include experts (DVB for example). Goal is to experiment and improve system..thus, could interesting to discuss and exchange with "The" team to share knowledge when we have such issue.
73 Evariste F5OEO