[SOLVED] TBS6280: Poor SNR when tuned to HD channels

[zerogravitas]

I have recently upgraded from a Nova 500 DVB-T tuner card to a TBS6280 which has DVB-T2 capability. I am UK based (London) and have a pretty strong signal - Crystal Palace is a couple of miles down the road.

Unfortunately, I am getting much poorer reception on the HD channels than I am on the SD ones. I have adequate signal strength but femon -H reports much weaker SNR when tuned to an HD channel. I get picture breakup when watching or recording HD channels and I also get weird frame rate issues - looks like it is dropping every other frame making it really "choppy" and also audio/video sync issues - I have noticed it go out by anything up to ten seconds. I think this is just an issue of the tuner not being able to reliably capture a clean HD stream.

When tuned to "Dave" I get output like the following - looks fine to me.

Code: Select all

mythtv@poseidon:~$ femon -H
FE: TurboSight TBS 62x0 DVBT/T2 frontend (DVBT)
status SCVYL | signal  81% | snr  68% | ber 0 | unc 0 | FE_HAS_LOCK
status SCVYL | signal  81% | snr  69% | ber 0 | unc 0 | FE_HAS_LOCK
status SCVYL | signal  81% | snr  69% | ber 0 | unc 0 | FE_HAS_LOCK
status SCVYL | signal  81% | snr  75% | ber 0 | unc 0 | FE_HAS_LOCK
status SCVYL | signal  81% | snr  64% | ber 0 | unc 0 | FE_HAS_LOCK
status SCVYL | signal  81% | snr  71% | ber 0 | unc 0 | FE_HAS_LOCK

However, when tuned to "BBC ONE HD" I get output like this, which is less encouraging.

Code: Select all

mythtv@poseidon:~$ femon -H
FE: TurboSight TBS 62x0 DVBT/T2 frontend (DVBT)
status SCVYL | signal  80% | snr  29% | ber 0 | unc 0 | FE_HAS_LOCK
status SCVYL | signal  80% | snr  29% | ber 0 | unc 0 | FE_HAS_LOCK
status SCVYL | signal  80% | snr  29% | ber 0 | unc 0 | FE_HAS_LOCK
status SCVYL | signal  80% | snr  29% | ber 0 | unc 0 | FE_HAS_LOCK
status SCVYL | signal  80% | snr  29% | ber 0 | unc 0 | FE_HAS_LOCK

Setup is as follows

Back end: Intel i7, 48G RAM (it's not just a myth box), Ubuntu 12.04, MythTV 2:0.27.4+fixes, TBS6280
Frontend: Intel NUC Celeron 947, 2G RAM, Ubuntu 12.04, Kodi 14 + pvr.mythtv plugin
TV hardware: Loft mounted aerial, Labgear HDU681/S Home Distribution Unit - professionally installed in 2012

I've got plenty of horsepower and a gigabit network and MythTV all works perfectly on the SD channels - it's intermittently perfectly watchable on the HD ones even. It is noticeably worse in bad weather conditions so I am pretty sure it's an issue with the incoming signal rather than a hardware or software issue per se. I've tried using both the open source and the proprietary drivers for the capture card but it doesn't seem to make any odds.

Question is what should I do now? I'm OK with spending a bit of money to sort it out but I don't want to buy hardware that's not going to help. I gather than connecting multiple amplifiers in series is just going to introduce more noise but I am pretty clueless about RF generally. Anyone got experience of successfully solving this kind of problem?

[stuartm]

There are a couple of things to consider.

[*]A poor quality aerial, wrong type or simply in a bad condition; low quality coaxial cabling (no shielding) - all may give a high signal strength but a low signal quality. If you are using any signal splitters, try removing them. If the cable has insufficient shielding then local sources of radio noise - DECT phone base stations, mobile phones or just malfunctioning equipment nearby can all create interference. Try turn off or moving things away from the TV and the cable from the aerial.

[*]Too much signal. Being that close to the transmitter it's actually possible that the signal is too strong and it is overwhelming the hardware. If you have any signal amplifiers connected then remove them, also check whether the card has a built-in amplifier, if so it should be disabled. You may try fitting a signal attenuator which will reduce the strength.

[PhilB]

Any more hints here? http://www.mythtv.org/wiki/DVB-T_Reception_Problems

HD is transmitted at same power and same height as SD: http://www.ukfree.tv/transmitters/tv/Crystal_Palace

Is there a pattern with the multiplexes/ frequencies? Are all DVB-T2 poor and all DVB-T good? Or is it just (say) the higher frequencies? I would fully expect DVB-T2 would be less tolerant of issues than DVB-T because it uses QAM 256 not 64.

As a comparison, I also am 20Km from Crystal Palace – a blissful situation without the nasty reflections from mountains and tidal waters of previous houses. I have a high gain loft aerial, a longish downlead and a 3 way passive splitter at the end feeding a Hauppauge T500, a Hauppauge DBV-T stick (both DVB-T) and a TV. The TV has stable DVB-T2. Perhaps as has been suggesting you are indeed amplifying it too much?

Alternatively, are you sure it’s a reception problem and not a graphics one? It’s MPEG4 not MPEG2 and higher bit rates. Also, doesn’t BBC HD switch between interlaced and non interlaced in mid program which will be an extra challenge for your graphics?

Phil.

[jksj]

That close to the transmitter I would suspect signal over load.
If removing all forms of amplification from the signal path has not worked try one of these.
http://www.amazon.co.uk/Aerial-Coaxial- ... attenuator

[zerogravitas]

Thanks for the suggestions everyone.

The aerial is nearly new, high gain, professionally installed and the amplifier is supposedly of good quality. All the cabling is also new, good quality coax - I refurbished the house when I moved in a couple of years ago and it all got "done". Given that I am literally 3km from the transmitter (according to Google Maps) I really should have a decent signal.

I'm sure it's not a video decode problem, the NUC is using hardware acceleration and the x264 decode is handled by VAAPI (Intel SandyBridge chipset.) CPU utilisation on the NUC actually drops when it is playing back video as opposed to sitting in the Kodi menus - Kodi has a rather weird architecture due to its heritage as a games console application and it still does a lot of continuous redrawing. I've also got Raspberry Pi clients and even they can manage (with hardware acceleration), there's no way they could cope in software!

I've only got two DVB-T2 multiplexes - 546000000 and 570000000. All the DVB-T ones seem to be fine but both DVB-T2 ones are dreadful, the 570000000 in particular produces results like this - actually got some artifacting while I was running my test as you can see.

Code: Select all

status SCVYL | signal  83% | snr  11% | ber 134 | unc 134 | FE_HAS_LOCK
status SCVYL | signal  83% | snr  21% | ber 0 | unc 0 | FE_HAS_LOCK
status SCVYL | signal  82% | snr  19% | ber 0 | unc 0 | FE_HAS_LOCK
status SCVYL | signal  82% | snr   8% | ber 0 | unc 0 | FE_HAS_LOCK
status SCVY  | signal  83% | snr   6% | ber 256 | unc 256 |
status SCVY  | signal  82% | snr   8% | ber 256 | unc 256 |
status SCVYL | signal  83% | snr   8% | ber 1 | unc 1 | FE_HAS_LOCK
status SCVYL | signal  82% | snr  16% | ber 0 | unc 0 | FE_HAS_LOCK
status SCVYL | signal  82% | snr  14% | ber 0 | unc 0 | FE_HAS_LOCK

I'm coming round to the idea that I might have too much gain and there are a couple of things that I can do reasonably cheaply to try and sort that out. First one is going to be hooking the TBS6280 input directly to the aerial downlead rather than via the amplifier with a view to passively splitting the signal upstream of it - that will also take amplifier noise out of the equation - I just haven't got the right cable, annoyingly.

After that I guess I am into line attenuators - will certainly check out the suggested model. A little shopping is necessary now!

[zerogravitas]

It does indeed seem to have been too strong a signal - I bought the appropriate bits to put a passive split in upstream of the amplifier and I now have a much more stable HD picture and a much more encouraging SNR. This is the same multiplex that was previously dropping out all over the place with a single digit SNR reading.

Code: Select all

mythtv@poseidon:~$ femon -H
FE: TurboSight TBS 62x0 DVBT/T2 frontend (DVBT)
status SCVYL | signal  82% | snr  49% | ber 0 | unc 0 | FE_HAS_LOCK
status SCVYL | signal  82% | snr  49% | ber 0 | unc 0 | FE_HAS_LOCK
status SCVYL | signal  82% | snr  49% | ber 0 | unc 0 | FE_HAS_LOCK
status SCVYL | signal  82% | snr  49% | ber 0 | unc 0 | FE_HAS_LOCK
status SCVYL | signal  82% | snr  49% | ber 0 | unc 0 | FE_HAS_LOCK
status SCVYL | signal  82% | snr  50% | ber 0 | unc 0 | FE_HAS_LOCK
status SCVYL | signal  82% | snr  49% | ber 0 | unc 0 | FE_HAS_LOCK

Seems to be working well enough now, not going to bother experimenting further. Thanks for the tips.