IEEE-1394 & Mini-DV dead??????

[Brian Ellis]

Hi!

I've just received some info from Panasonic Broadcast and Digital Division which states that "DV would be old news" by the time they could develop any new models.

They go on to say: "Given the upcoming availability of SDI cards for PC's I suspect you will see SDI replacing 1394 before too long. 1394 has too many limitations and it's far from the universal means of connecting two devices that most believe it is. Given the 3+ versions of 1394 now in use, the serious productions houses chose SDI or SDTI years ago.

Excuse my ignorance, but what is SDI? And does this mean that we are becoming victims of a techie advance before our existing equipment is half-out of its box?



------------------
Brian (the terrible)



[This message has been edited by Brian Ellis (edited 18 March 2000).]

[jeff b]

I wouldn't be too worried about it, since the low cost of IEEE-1394 and its current market penetration being well along, any new format/style of interface would take a couple of years to 'trickle down'. Plus, if it isn't as inexpensive as IEEE-1394, then what would be the point? After all, people will choose a more inexpensive solution that's already working over one that not only costs more to switch over to, but that isn't developed in the market's software yet at all. Just the lack of available software for alternative operating systems makes MS the money maker's choice, too. So, I don't see that SDI (whatever it may be) can compete to anywhere near the point of suddenly 'wiping out' the consumer/prosumer market share that IEEE-1394 has right now, and which will continue to increase in the future.

[Brian Ellis]

Jeff

I'm not so sure that the manufacturers are not capable of f***ing everyone up if they thought a buck or two could not be persuaded to come their way.

Anyway, since my original post, I've been doing a search and have come up with http://www.xs4all.nl/~brw/ds_products/sdti.html , which does explain what SDI is and what it is all about. And, if you go to the bottom of the article, see where it comes from: the same stable as RT-2000! Is there dirty work going on at the crossroads?

Well, let me see what I can interpret from this:
1. 1394 must have been buggered around à la Microsoft, if there are 3+ versions
2. 1394 is not dedicated and can hiccup very seriously if you try to Tx/Rx simultaneously data packets from different sources (especially if they are formatted differently), but SDI has sufficient bandwidth to allow this, as it has pre-emptive transmission to boot
3. SDI cards for PCs are becoming available and are low (?) cost.
4. This being so, are the video equipment makers not going to force the issue, first at pro level, then at prosumer level?
5. For us guys, the bottleneck is not going to be improved, the hard disk.
6. This seems to be two separate issues: DV being old news and 1394 being replaced by SDI.

------------------
Brian (the terrible)

[Pooh]

Interesting article Brian. You make reference to it being low costish. Do have any experience of the actual costs associated with these boards.

Pooh

[Brian Ellis]

Pooh

No! What may be low cost to a pro may be horrendously expensive for us. After all, their "low cost" cameras start at $10k



------------------
Brian (the terrible)

[Brian Ellis]

Hi, all!

Had a chat with a pro working for a TV company.

SDI cards are in the $2500 - 5000 ballpark and are designed to import TV without any compression. They require a SCSI Raid disk system (typically 4 x 50 Gb, 10,000 rpm, 55 Mb/s) for the video plus another SCSI disk for the audio on an NT 4 machine with a 450 MHz PIII or 2x 350 MHz PII processor and 128 Mb RAM (pref. 256 Mb). Typically, then, we are talking about $10k for this little bit of hardware.

The advantage? It is to be found mainly with NLE. If you edit a compressed format, e.g. DV or MPEG, you get a loss each time you add an effect, transition or what have you. This does not happen with an uncompressed format. The pros therefore prefer to compress, if needed right at the end of the finished article, but it implies that the source must be uncompressed, as well, hence DV going by the board. Another advantage is that SDI boards can be connected simultaneously to two sources (or four, if you put in two boards), so that a preliminary mixing can be done in real time, at the capture. NLE rendering is also in real time.

It would therefore seem that this is strictly for the pros at this time, there being no prosumer version likely to be in the offing. My interlocutor said that it is only in the past year that a few TV stations have started to use it but it does ease a number of problems for them.

------------------
Brian (the terrible)

[DrMordrid]

While SDTI may be useful for the high end cams there is plenty of potential for Intel to cause trouble for IEEE-1394 with the upcoming SATA (Serial ATA) drive interface.

Intel states it's for primarily for internal storage, but it could be easily adapted to external devices given it uses a thin 4 wire cable and offers 150-600 mB/sec transfer rates. It's also royalty free, which IEEE-1394 isn't.

These factors and Intel's open hostility for IEEE-1394 makes one wonder....

Of course if they DO go this far it would be awfully nice if a new DV compression were paired to it with 4:2:2 colorspace instead of the current 4:1:1 NTSC/4:2:0 PAL.

Dr. Mordrid




[This message has been edited by DrMordrid (edited 29 March 2000).]

[Brian Ellis]

Doc

An interesting post. Thanks for your input. I agree the transfer rates you quote sound interesting, but are they practical without, at least, using transmission line technology? I hae ma doots. In order to maintain even a semblance of a square wave pulse, it is necessary to transmit up to at least the fifth harmonic. To take your upper figure of 600 Mb/s, this is 4800 Mbit/s and the fifth harmonic will take this to 24 000 MHz which is somewhere between a microwave and infra-red frequency In reality, this would require a waveguide and I see little chance of plumbing in a camera or any other external device !!! With a carefully impedance-controlled wire system, at this frequency, I feel that wire lengths would be limited to about 2 or 3 cm, assuming non-reactive terminations. Anything longer and you would start to get enormous electromagnetic compatibility problems, as reflections would occur. Furthermore, it would necessarily be low impedance, probably in the 50 - 100 ohm range, so that relatively high powers would be required, exacerbating the radiation difficulties and it would use very expensive semiconductors. Quite frankly, I feel that there is a practical limit, with current technology, for a serial system to stop at about 50-60 Mb/s. Of course, this could be octupled for 8 bit parallel systems. A la rigueur, I could just foresee the possibility of the figures you mention if you had said Mbit/s and not Mb/s, but even that is a bit (no pun intended) pushed.

I'm surprised that IEEE-1394 is not in the public domain. It is very rare that the IEEE would publish a standard that was proprietary.

At this speed, we would not require any compression, at all

------------------
Brian (the terrible)