You are correct, sir. Happy HiPix holder here.
Yay, HiPix. And maybe someday WinTV-HD.
Boo, ADTV. For archiving, anyway. (But I do admit I'm jealous of ADTV's real-time Pause. OTOH, I find that mostly I'm watching pre-recorded material these days!)
Aside related to real-time Pause and to multi-room viewing, not that you asked: I expect that when software HDTV players become practical in maybe a year (thanks to AMD and Intel for obeying Moore's Law), I expect I'll be able to have the HiPix recording to a network drive attached to one PC, and watch on another (fast) PC via the software player, running as little as one minute-file behind the recorder. (Gee, couldn't a 2nd HiPix on the 2nd PC do that today?!!! But an ADTV never will be able to allow either. Boo, ADTV.
Regarding error protection:
Before choosing the type of ECC, you need to understand channel error characteristics. If you get lots of scattered single-bit errors, block codes are indicated. If you get infrequent dropouts of multi-bit sequences, you want something more like a convolutional code. If an error in one track allows you to predict errors in approximately the same region of following tracks, you probably want to spread the data according to a pseudo-random spreading function, and treat the now-spread errors as multiple single-bit errors.
Because the DVSpoof code is completely malleable, it's possible to insert any type of error protection once the methods are decided. Every 4 MBytes might be the right places to put FEC bits, or maybe not.
So, I recommend we get enough reports of errors before we settle on a way to prevent it.
Btw, Andy says he also has some background in ECC.
Alternative backup methods:
Does $600 make sense for a backup system? Up to you. Note that the media cost for D8 tape is about $2/hour or <$0.25 per GByte uncompressed (2/3 that for compressed). Compare to DDS-3 or DDS-4 at $1 to $1.20/GByte or about $11-$13/hour uncompressed (1/2 that for compressed).
(I think you can find camcorders about $100 cheaper, but maybe the digital tape drives can be bought more cheaply, too.)
So, after, say 50 2-hour movies, you'll have spent about $75 to $100 more with DDS-3, or $375 to $400 more with DDS-4 (assuming you use every byte on every tape; camcorder tape fits about 1h13m per tape, btw).
And you get the camcorder functions for free. The Sony I have has this cool feature that it can record 640x480 stills either on tape or on internal Memory Stick (tm). My girlfriend's daughter went to a Senior Prom yesterday, and I was able to take video and stills, and review the results immediately. I got tremendous WAF points. You can't get those with a computer tape drive (depending on the W, of course).
You are, of course, free to do backups in any way you wish.
But the more different formats there are for archival copies, the more difficult it might be to trade material, if there's ever any material that can be traded, respecting, of course, any copyrights that might exist. Just something to keep in mind.
For this application, ZIP is not recommended. The problem is that ZIP requires perfect data integrity, else it bombs, and all data is lost even if only one bit is in error.
Also, ZIP compression introduces another two steps in the round trip to tape and back:
Skip steps 1 and 6 for current DVSpoof. Admittedly, the steps could be incorporated into the perfectly malleable code, but the time expended would still increase, even if the user-intervention-required steps were eliminated.
Aside: One way to think about the difference between ECC and Data Compression is that DC removes redundancy while ECC adds redundancy. You can combine the two sometimes, but you need to understand channel characteristics, and sometimes data characteristics. A special comment about double-density tape compression is that the loss of redundancy decreases the signal to noise ratio of the channel, thus, increasing the probability of an error by a factor of 2.
(The following comments fit in the "understand the data characteristics section.) ATSC streams are already fairly compressed thanks to MPEG2. I almost wonder if your 2:1 reduced transport stream was HDTV? Apparently the full channel bandwidth was not being used. Don't expect similar results with, say NBA games in HDTV where the compression algorithms are struggling to keep up with all the moving images.
One way to think about encryption, btw, is that it also removes redundancy. And when enough redundancy is removed, the effect of the tiniest error becomes catastrophic: the data cannot be recovered. Conversely, when there is sufficient redundancy, it is possible to retrieve the original data, even when some of the bits are not recovered.
At least that's the perspective that gives me some intuitive feel for these issues.
Obviously, you've come to the right place! http://www.avsforum.com/ubb/wink.gif