The main question is what the time for PCR error in DVB streaming ?
I am asking it because according to DVB standards (see additional information) PCR error retention period > 100 ms. But there is a lot of hardware which is logging PCR errors with PCR ~50 ms.
Aditional information
You can skip it, if you already know the answer, this information presents just as additional info about standards that I was found
ETSI TR 101 290 V1.2.1 (2001-05) - aka true DVB http://www.etsi.org/deliver/etsi_tr/101200_101299/101290/01.02.01_60/tr_101290v010201p.pdf
A PCR _accuracy_error occurs when a transmitted PCR value differs from what is expected by more >than 500 nanoseconds. The expected PCR value is calculated using an extremely stable internal clock >n the test device and previous PCR values. The calculated PCR is then compared to the transmitted >PCR values to check for accuracy. It is important to note that most receivers do not contain very >accurate clocks and therefore can be severely affected by this error.
ETSI TS 101 154 V1.9.1 (2009-09) http://www.etsi.org/deliver/etsi_ts/101100_101199/101154/01.09.01_60/ts_101154v010901p.pdf
Program Clock Reference (PCR) Encoding: The time interval between two consecutive PCR values of the same program shall not >exceed 100 ms as specified in clause 2.7.2 of ITU-T Recommendation H.222.0 / ISO/IEC 13818-1 [1]. Decoding: The IRD shall operate correctly with PCRs for a program arriving at intervals not exceeding 100 ms
ISO/IEC 13818-1 (Information technology — Generic coding of moving pictures and associated audio information: Systems) https://forums.xilinx.com/xlnx/attachments/xlnx/DSPTOOL/15095/1/iso13818-1.pdf
Thank you.
It's actually 40 ms, not 50 ms. But generally, both 40 ms and 100 ms are correct.
TR 101 290, section 5.2.2, note 2 to table 5.0b says:
NOTE 2: The limitation to 40 ms in the 'Preconditions' of 2.3 PCR_error and 2.3a PCR_repetition_error was removed from TS 101 154 [i.30] in 2005. The respective clause there now refers only to the 100 ms limitation in [i.1] which is recommended to be applied generally.
So I would conclude that the 40 ms has been relaxed to 100 ms, and that tools complaining about more than 40 ms have simply been implemented against an older version of TR 101 290.