most sensitive PC HDTV tuner card



Quote:
Originally Posted by SSpade19 /forum/post/0


I have an ATI HD Wonder tuner card and find its reception not to be as good as the built in tuner from my Panasonic plasma or my dish network dvr box. During certain times, the ATI card doesn't provide a stable signal for certain channels while the Panasonic tuner does.


What are some PC HDTV tuner card with the best ATSC OTA signal reception? Preferably one that can record 2 HDTV programs at once.

Click to expand...

Just go to the web and type in ATSC tuner cards. I have seen them as much as $200.00 , I myself have a Twinham tuner card. I have a great roof antenna so I get good reception, however stations in the other direction CANNOT be put in the memory without a rescan. I had a LOT of issues with this card freezing when I went to change channels. I have never seen tuner cards so far that could record 2 HDTV programs at once. I'm sure they can make then , but you know they will be big bucs.My card has a still capture feature which works well. I can change my channels now that I turned down my video card settings to the lowest rate, but I occasionally get an all pink screen. I close the program and have to start over again. They still have a lot of issues to deal with. I wish I spent the money on a OTA set top box instead.

We don't know all the variables but the OP's A/B test analysis that one tuner out performed another would only be valid if the two units were compared simultaneously sharing exactly the same antenna signal, which would need to be split. There's a possibility that the OP's two units were not in the same room, the comparison had the splitter in place (which causes a guaranteed signal loss of 3-6 dB) under "A" but not "B", or the OP is using different antenna feeds. Using two identical antennas with "identical orientation" also wouldn't be valid because they, by definition, must be in two different locations. Signal strength varies 10 to 100 dB based on antenna directivity selection, location (not room to room but rather inch to inch), including height and tilt, and orientation (compass bearing). Signal strength, or more precisely the sensitivity of a tuner to receive a signal, is unlikely to be even close to 10 dB in difference between brands within the same generation of tuners. Yes it is true newer generations have better and more sensitive tuners than older ones so I'll revise my statement:


Tuner reception quality, within the same generation of tuners, varies only very slightly compared to differences in antenna type/size and location/orientation.



Quote:
Originally Posted by m. zillch /forum/post/14161335


We don't know all the variables but the OP's A/B test analysis that one tuner out performed another would only be valid if the two units were compared simultaneously sharing exactly the same antenna signal, which would need to be split. There's a possibility that the OP's two units were not in the same room, the comparison had the splitter in place (which causes a guaranteed signal loss of 3-6 dB) under "A" but not "B", or the OP is using different antenna feeds. Using two identical antennas with "identical orientation" also wouldn't be valid because they, by definition, must be in two different locations. Signal strength varies 10 to 100 dB based on antenna directivity selection, location (not room to room but rather inch to inch), including height and tilt, and orientation (compass bearing). Signal strength, or more precisely the sensitivity of a tuner to receive a signal, is unlikely to be even close to 10 dB in difference between brands within the same generation of tuners. Yes it is true newer generations have better and more sensitive tuners than older ones so I'll revise my statement:


Tuner reception quality, within the same generation of tuners, varies only very slightly compared to differences in antenna type/size and location/orientation.

Click to expand...

Things have just gotten too complicated for those generalizations, though they used to work well in the analog world. Even the thoughts on db are beside the point, and the assertion "varies only slightly" deserves harsh criticism, as it's simplistic. No one denies that antenna placement and signal strength have dramatic and occasionally overlooked impacts on reception, so that's a dead horse. What's interesting (and beyond your expertise as well as mine) is the extremely complicated infrastructure of one ATSC chipset and another. There are amazingly contemporary advances in ATSC tuners that go well beyond analog tuners (and again, well beyond your experience as well as mine) -- the multipath corrections alone are a true innovation in digital technology. And there is layer after layer of more nuance in an ATSC tuner that goes far beyond the minor (but baseline) issue of having a good antenna and good signal strength!


The evolution of the ATSC chipset is confirmed, and furthermore is undergoing continuing evolution. I am anxious to see the next stage of improvements when a working group that involves LG and others hooks on a portable video component to ATSC that takes on the dominant virtues of MediaFLO. As that innovation unravels, you can expect that the nuances of ATSC reception will continue to improve.

The placement of the tuner section in an HDTV was chosen carefully by the designers of the product and tested as a prototype before production. The placement of an after market computer tuner card is done by the consumer and in a ton of different computer designs. One consumer may place it in a slot that is 1cm away from an RF noise generating hard drive, fan motor, or optical drive whereas another may have the luxury of placing it much farther away or have hard drives, fans, opt. drive etc. that generate much less RF noise in the first place. Two people could buy exactly the same card, have exactly the same antenna signal, yet get totally different results because of this. The little metal box that encases the tuner section isn't a perfect RF noise shield, it is thin metal and it has holes, also the power and other wires that lead into it may act as antennas accidentally themselves and also pick up RF garbage. Often bad reception isn't due to a weak signal but rather a noisy signal. The inside of a computer case is extremly noisy (regarding RF) and is about the last place on earth I'd want to place a tuner.


If I was in the market for one of these HD tuners I'd want it in a separate enclosure outside my computer case and if it was one of those that plugs straight in to a USB port, like the Hauppauge HVR-950, I'd also buy a USB extension cord to situate it several feet away, not adjacent to the RF noise monster; the computer.


I see the HVR-1600 has an extra RF jack, presumably so one can use different antennae for analog and DTV I assume. If one jack was left empty, say because all the user wanted was DTV alone, then that unused one potentially is another pathway for incoming RF noise, even for the other tuner that it serves. It should be capped off and not left empty. At the very least stick a well shielded cord, ideally RG6 quad-shielded but even just run-of-the-mill RG59 would do, string it away from the noise monster and don't leave any of its center core exposed at the other end (that would be an antenna too) snip it off.

Dual tuner cards are common for the European DVB-T OTA system (Hauppauge make both USB stick and PCI models) - so I'm surprised they're not as common for ATSC 8-VSB (though I see Hauppauge list a PCI HVR 1600 and a PCIe HVR 1800MC dual tuner model)? I've got both a Nova-T 500 PCI dual-tuner model and a Nova-TD-Stick USB model for DVB-T and have found them pretty good - though this means nothing in comparison to ATSC 8SVB/QAM models I would suggest.


Some of the newer DVB-T dual-tuner models sometimes have a useful trick (not sure if done in hardware or just by a clever driver) - in that you can use them in diversity mode to reduce the number of errors on a received multiplex. To do this you feed the two tuners from two different aerials/antennae (the Hauppauge models come with small whip aerials about 4 inches high with magnetic bases) sited a few feet apart (not the same antenna).


The PC then tunes both tuners to the same frequency, and then as the streams from the two antennae may well have errors at different times, by switching to the error free data dynamically, you end up with a digital stream with fewer (sometimes no) errors. I've used this in hotel rooms, and the results are amazing - from unwatchable to error-free in one case, when I was recording SVT HD in Sweden (18Mbs 720/50p H264 service sharing with one 576/50i MPEG2 service).


Of course in diversity mode you lose the ability to record a service on one mux whilst watching one on another.


(Wonder if this technique is as applicable to 8VSB ATSC?)

Put it this way, if each independent frame (or field) of video has an assigned number, then it could work. [Do they? I don't know.] The diversity section has to compare and analyze two identically numbered frames of video before it can decide which one to keep and which one to discard. If they aren't numbered it won't know which two to compare.


If you go to a HDTV store where you can view dozens along a wall all working in unison on the same signal in the same glance, you may notice that some of them seem out of sync with the others by just a few milliseconds. The thing to look for is when there is a camera cut/change to a new scene/angle. This proves that even when being fed the same wired signal different tuners' outputs aren't in perfect sync with each other. Signal path reflections off nearby mountains and such, common to OTA reception, would exacerbate this I'd think.



Quote:
Originally Posted by m. zillch /forum/post/14180407


Put it this way, if each independent frame (or field) of video has an assigned number, then it could work. [Do they? I don't know.] The diversity section has to compare and analyze two identically numbered frames of video before it can decide which one to keep and which one to discard. If they aren't numbered it won't know which two to compare.

Click to expand...

That isn't quite the way digital TV works. Rather than each field or frame being sent as discrete elements, the system sends packets of data as part of the transport stream, all of the same size (188 bytes?) that can contain audio, video, subtitles/captions, interactive content, EPG information etc. - and not just for a single service. Those packets are demodulated and errors detected in this process. AIUI each packet has an identifier to ensure that they are re-assembled in sequence.


If you have two tuners and two demodulators running on the same service, then you may detect an error on one packet on one tuner/demodulator, but get an error free version on the other tuner/demodulator, and thus use that one instead.


The error correction by diversity tuners is performed on the transport stream - upstream of any MPEG2 decoding. As such it doesn't just fix errors in the video, it does this in the audio, subtitles/captions, EPG, data etc. streams. Any buffering would be at the transport stream rather than decoded video level. This is particularly useful if you record multiple streams from the same multiplex (In the UK we have the SD services BBC One, BBC Two, BBC Three/CBBC and BBC News all multiplexed into one transport stream on National Multiplex 1 - and can record any or all of these using a single tuner)


It isn't done on a field-by-field or frame-by-frame process (after all the stream has audio and text data as well) - rather it is done on individual data packets AIUI.


After all MPEG2 and H264 don't send full fields or frames, they send a Group of Pictures, containing full I-frames (which are effectively a frame of video that can be reconstructed on its own - a bit like a JPEG image) but they only send these every 10-20 frames, in in between send B or P frames, which contain Predictive or Bi-directionally predictive motion-compensated difference between frames, not the whole frame, and different frames may have significantly different sizes, which is why MPEG2 encoders and decoders must buffer quite a number of frames to be effective. (Hence the delay between analogue and digital services in many cases)

Quote:
If you go to a HDTV store where you can view dozens along a wall all working in unison on the same signal in the same glance, you may notice that some of them seem out of sync with the others by just a few milliseconds. The thing to look for is when there is a camera cut/change to a new scene/angle. This proves that even when being fed the same wired signal different tuners' outputs aren't in perfect sync with each other. Signal path reflections off nearby mountains and such, common to OTA reception, would exacerbate this I'd think.

Click to expand...

That is just as likely to be caused by different amounts of digital processing delay in the display itself (some plasmas and LCDs delay by more than others after the MPEG2 decoder) I've seen some display add up to 7 fields of delay...


Where you DO see errors cause issues is often in A/V sync - where audio or video errors are not matched, meaning dropped audio / video frames cause audio or video to lag or lead. Usually rebooting or even just switching channels (which will flush buffers) can cure this.


It is unlikely that regular RF reflections would cause much delay - their duration is usually of the order of microseconds, not milliseconds.

OK I'll re-word it:


"Put it this way, if each independent packet has an assigned number, then it could work. [Do they? I don't know.] The diversity section has to compare and analyze two identically numbered packets before it can decide which one to keep and which one to discard. If they aren't numbered it won't know which two to compare."

Quote:
It is unlikely that regular RF reflections would cause much delay - their duration is usually of the order of microseconds, not milliseconds.

Click to expand...

When comparing two continuous streams of data, without a defined beginning or an end for reference, microseconds of timing errors count.


I'll agree the different delays in the TV stores' sets could just as easily be different brands of different de-interlacing chips or any number of other video processes, true, but multi-path (which caused ghosting in analog NTSC sets) is very much still a problem which can be addressed by putting a marker in the signal itself for the tuners to lock on to. This was added very late in the game for NTSC (in the 1990's IIRC) and only a few tuners came to market with such a GCR circuit (Ghost Canceling Reference). I never bought one because I never could get a straight answer as to who was and who wasn't transmitting the optional GCR signal. Then digital came along and I lost interest.



Quote:
Originally Posted by m. zillch /forum/post/14181483


OK I'll re-word it:


"Put it this way, if each independent packet has an assigned number, then it could work. [Do they? I don't know.] The diversity section has to compare and analyze two identically numbered packets before it can decide which one to keep and which one to discard. If they aren't numbered it won't know which two to compare."

Click to expand...

http://en.wikipedia.org/wiki/MPEG_transport_stream


Confirms that the first 4 bytes of the 188 byte transport stream packets include a 13 bit PID reference and a 4 bit continuity counter which presumably ensures that you know which packet is which?


(ATSC adds 20 bytes to each 188 byte packet for added Reed Solomon error correction)


There is also a PID (I think) carrying a PCR - Program Clock Reference - which provides the reference clock that other program streams refer to.


So the continuity counter provides an ID for each packet, and the PCR provides timing to rebuild the video and audio etc.

Quote:
When comparing two continuous streams of data, without a defined beginning or an end for reference, microseconds of timing errors count.

Click to expand...

But I think that is why there are both packet counter bits for each packet (for use at the data level) and a clock reference stream in the transport stream for use at the video/audio level.

Quote:
I'll agree the different delays in the TV stores' sets could just as easily be different brands of different de-interlacing chips or any number of other video processes, true, but multi-path (which caused ghosting in analog NTSC sets) is very much still a problem which can be addressed by putting a marker in the signal itself for the tuners to lock on to. This was added very late in the game for NTSC (in the 1990's IIRC) and only a few tuners came to market with such a GCR circuit (Ghost Canceling Reference). I never bought one because I never could get a straight answer as to who was and who wasn't transmitting the optional GCR signal. Then digital came along and I lost interest.

Click to expand...

Yep - but the effects of multipath on analogue and digital systems are very different. The reflected and delayed signal caused by multipath doesn't delay some bits - it just corrupts some.


(In DVB-T multipath can actually improve signals - because rather than a single carrier per RF channel, DVB-T uses COFDM where approx 2000 or 8000 individual carriers are used per RF channel, with the data for that RF channel spread across all the carriers. The carriers can have long guard intervals between symbols - which means if the multi-path is within the guard period it actually increases the received energy of the signal... This also allows single frequency networks to be used - allowing a receiver to receive signals from two transmitters on the same frequency - not really an option with analogue)