TS discontinuities

[LubosD]

Hi,

I'm using a TBS6985 with VDR to stream TV channels to XBMC. Once I started using TBS6985 (replaced two separate DS3000-based cards), I see occasional picture corruption, even though the signal is VERY strong and femon is not reporting any corrupted data.

When playing the stream in VLC, I get these warnings whenever the picture starts breaking up:

Code: Select All Code

ts warning: discontinuity received 0x2 instead of 0x1 (pid=2010)
ts warning: discontinuity received 0xd instead of 0xc (pid=2010)
ts warning: discontinuity received 0xf instead of 0xe (pid=2010)
ts warning: discontinuity received 0x9 instead of 0x8 (pid=2070)
ts warning: discontinuity received 0x0 instead of 0xd (pid=2010)
ts warning: discontinuity received 0xd instead of 0x1 (pid=2010)
ts warning: discontinuity received 0xa instead of 0x9 (pid=2010)
main warning: not synchronized (-40011 us), resampling
main warning: buffer too early (-40011), down-sampling
main warning: resampling stopped after 10112258 usec (drift: -447)

I already use

Code: Select All Code

options saa716x_tbs-dvb int_type=1

to prevent IRQ: nobody cared errors.

Any ideas what might be wrong?

[LubosD]

I have now tried different software (Tvheadend) and it exhibits the same problem. The problem goes away once I disable "idle scanning".

To me it seems that parallel use of multiple adapters at the same time, or possibly just the tuning (my feeling), triggers the issue.

[JV16Bar]

Hello,

I had similar problems, and I was helped by the TBS support team (thanks again!).
What you should try:
1. Update your motherboard's BIOS to the latest version (this was the solution in my case).
2. Check your BIOS' PCI-E related options: no PCI-E overclock, set it manually to 100 MHz.
3. If you have the possibility, try your cTBS ard with a different MB.

[BeMax]

It seems that I have the same problem. I use one adapter (0) for SNR measurements. Every minute I collect SNR value for 5 frequencies with dvbsnoop. If I start tvheadend for the other adapters (1,2,3) the measurements will sometimes get corrupted due to the scanning.

It feels like a isolation issue between the adapters ?

[LubosD]

Hello,

I had similar problems, and I was helped by the TBS support team (thanks again!).
What you should try:
1. Update your motherboard's BIOS to the latest version (this was the solution in my case).
2. Check your BIOS' PCI-E related options: no PCI-E overclock, set it manually to 100 MHz.
3. If you have the possibility, try your cTBS ard with a different MB.

This issue is rather unpredictable for me... I thought it went away after updating BIOS, but I was wrong.

The corruption is definitely related to the amount of data being transferred, as long as I keep the transfers under 100 Mbit/s, then I'm on the safe side. By default, every tuner transfers exactly 100 Mbit/s of data regardless of the transponder's bitrate.

I find curious that the card accesses only 1MB of RAM for 4 tuners while my other single-tuner cards use 2MB, but I have no idea if that could affect anything.

Code: Select All Code

Region 0: Memory at f7a00000 (64-bit, non-prefetchable) [size=1M]

[LubosD]

I've disabled Spread Spectrum in BIOS and so far no glitches, BUT I've been wrong before, so I need to test it further.

[JV16Bar]

I've disabled Spread Spectrum in BIOS and so far no glitches, BUT I've been wrong before, so I need to test it further.

Is it working OK with the disabled Spread Spectrum?

[LubosD]

I've disabled Spread Spectrum in BIOS and so far no glitches, BUT I've been wrong before, so I need to test it further.

Is it working OK with the disabled Spread Spectrum?

It may have improved a bit, but I cannot say it's gone

[JV16Bar]

Can you debug if the problem occurs always on the same tuner (adapter), or it is not related to this?
I'm also experiencing some discontinuity problem, and it looks like that the problem appears mostly on the same adapter (/dev/dvb/adapter1)