I'm planning to start a service which archives TV video content (at low quality). I'm exploring what hardware and video digitization equipment will be needed.

The TV channels in the geography I'm looking at are unencrypted and are provided through coaxial cable wires (similar to standard 'Cable TV' in the US).

There are about 200-250 channels, each of which is to be digitized and archived. I'm looking for a cost-effective, scalable solution which can be scaled up from an initial set of 5-10 channels to a max of 200. I'm not sure how TV tuner recommendations which work only for 3-4 channels at a time would scale up cost-effectively.

  • So these are all standard-definition channels using NTSC?
    – Flimzy
    Nov 28, 2011 at 20:15

2 Answers 2


I'm assuming you're talking about digital cable, since it's 2012.

The way this works is that your equipment tunes into a frequency over which a multiplex is broadcast. That's a digital stream that can be demultiplexed into a number of sub-streams. Each sub-stream is a TV channel.

There are standards for these things. The sensible thing to do (especially since they're unencrypted channels) is to use DVB receiver cards in a PC. I'd use Linux, but it's up to you.

You mention archiving these at "low quality". The problem here is that the channels are broadcast at a given quality, and the broadcaster has already done the hard work of digitising and compressing the content. If you want to lower the quality, you'd have to demultiplex, decompress, scale, and compress, and that's a lot of processing. Compression is especially expensive in terms of processing.

Since storage is cheap, and getting cheaper, I'd suggest storing the streams exactly as they arrive. This is what DVRs (e.g. Sky+, digital TiVo etc. do).

You have a choice between demultiplexing, and storing the streams you want individually -- or simply writing the whole multiplex to disk. Demultiplexing is cheap; but not as cheap as simply dumping every byte the tuner supplies you onto disk. Demultiplexing, of course, allows you to pick and choose which channels to discard.

I'm afraid my 2 minutes of Googling didn't reveal what the bit rate of a cable TV multiplex is - but it shouldn't be difficult for you to find out. Once you know that, it's simple sums to work out how quickly you'll fill disks, and what kind of sustained write speed you'll need.

There is free software available to drive DVB cards. One example is OpenDVB. It would be worth doing some reading around the MythTV community. Although they'll be focussed one or two channels at a time, they'll know their way around the technology, and the software is modular.

You'll need a DVB card for each multiplex you're recording, since each one can only tune to one frequency. Depending on what you discover about bit rates, it may be worth splitting the task between several computers - but in theory one computer can have multiple DVB cards.

As a vague guide on CPU power - reading from the tuner and demultiplexing take negligible amounts of CPU work. Any PC CPU of the last 10 years could do it without breaking a sweat. Encoding/compressing is really hard work. A top-end modern CPU has trouble compressing SD video in real time. To do it at all efficiently, you'd have to offload it to dedicated chips (which is how camcorders etc. manage it).


You would need some sort of set-top box to get the signal out of coax into a video format, right? What country/area is this in? Who is the provider of these channels? Do you have the permission/co-operation of the broadcasters?

The main issue you're going to run in to is how to view all 250 channels at once, and output them to something that can record. You can typically only receive one or two channels simultaneously down your wire for cable TV. What format are the channels you want to record being broadcast in? I assume DVB-C? You're going to need a way to tune into that signal and you can only tune into a single frequency at a time with a single receiver. You'll need 250 tuners, no matter what.

Assuming you can get 250 signals (some sort of super-splitter/amplifier?) and connect 250 set-top boxes (mount em all in racks), you will still need 250 tuners, one for each channel. You could buy 125 dual tuners (@ approx $80ea with a bulk discount = $10,000) and put 3 cards into a computer (42 cheap computers @ approx $150ea = $63,00) that run nothing but capturing & recording TV (then I presume copying the data to some massive SAN for archival?). Roughly $17,000 to record 250 channels is only $68/channel. Doesn't sound too far fetched to me.

Sounds like an interesting idea though. Would love to see what you end up doing.

  • You don't need 250 tuners. Just a tuner per multiplex. One frequency; several channels.
    – slim
    Jan 12, 2012 at 14:58

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.