Date: Fri, 7 Nov 1997 18:22:01 +0000 (GMT) From: Andrew Gordon <arg@arg1.demon.co.uk> To: Luigi Rizzo <luigi@labinfo.iet.unipi.it> Cc: multimedia@FreeBSD.ORG Subject: Re: Teletext decoding with the Hauppauge... Message-ID: <Pine.BSF.3.91.971107174531.15717A-100000@server.arg.sj.co.uk> In-Reply-To: <199711070732.IAA26304@labinfo.iet.unipi.it>
next in thread | previous in thread | raw e-mail | index | archive | help
On Fri, 7 Nov 1997, Luigi Rizzo wrote: > > I am currently working on a teletext system (using some rather better > > hardware that allows the complete page database to be maintained > > I don't think there is really a limitation on the hardware (see > below) Well, I looked at using the SAA5246, but the I2C is just too slow to grab all of the pages that way. I had also seen your 'brute force' approach using the BT848, but that seemed to me to need too much CPU for a practical solution (and in any case, the difference in cost between a 486 and a P200 easily pays for the better hardware). I also considered unsoldering the SAA5246 and replacing it with a SAA5249 - this would give rapid access to normal pages, but does nothing for rolling pages. The hardware I am now using is the Unitext2 ( http://www.pelican.com.au/ ) which simply gives you a 50Hz interrupt and you get to read out the raw teletext lines for that frame from a buffer on the card. The manufacturers have been helpful in supplying data on how to program the proprietary parts of it (tuning etc.) - the actual teletext decode is handled by an SAA5250. I think this will allow me to grab all of the pages with fairly low overhead - give me a couple more days to get the driver going and I will have some numbers. > > in RAM/on disk rather than waiting for individual pages). My plan for > > user interface is to make a CGI that serves web pages. > > Yes, this was something we hacked up a few years ago. Unfortunately > the code for teletext->gif conversion was a complete mess. Well, mine probably uses more CPU than it ideally should, but it seems to produce good images. > The SAA5246 can capture up to 8 pages simultaneously, Actually, only 4 pages if you want the Fastext links and page CRC. > and you can > set up filters to select specific pages or get the first one which > comes in. So, although a bit tricky, one approach could be to use > 2-3 such pages to capture any data which comes in, and the > remaining to capture specific pages. Well, you can't grab all of the pages via a wildcard entry - if there are 16 rows per frame transmitted, that means 16*40*50 = 32000 bytes/sec, and since a byte read on I2C takes about 20 clocks at 100KHz max, you can only read 5000 bytes/sec. If the broadcaster transmits interleaved magazines, you might just manage to read one magazine continuously, but if you have to cycle through all the magazines as well as cycling through all of the channels, your cycle time (ie. maximum age of the pages you are showing the user) is going to be quite long. This is especially a problem with rolling pages - a few of the pages here have about 40 sub-pages, so with a 20-second turnover time this means you have to stay on channel for more than 10 minutes to capture the whole lot. Even with the hardware that allows reception of all magazines in parallel, this is uncomfortably long to wait on each channel. I am thinking of having 2 cards in the machine: one would change channels rapidly, (say 2 minutes per channel) to keep most of the pages up-to-date, while the other would change channels only every 20 minutes, to make sure that all sub-pages of big rolling pages have been captured. With the five channels we have here, this would ensure that most pages would be less than 8 minutes old, with the largest rolling pages possibly being up to 50 minutes old. > > Most of what I am doing won't be directly applicable, but I can let > > you have a teletext page -> .gif file conversion utility if you want it. > > yes please... Sent in private mail.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSF.3.91.971107174531.15717A-100000>