Disk recorders: past, present and future

By Oleg Bondarev

Review contents:
Disk recorders: past, present and future DV MasStor — a new page in NLE history Hard disk recorder DataVideo DV Bank DN-100 New surprises from Doremi Doremi Labs disk recorders and servers Disk recording devices of Ikegami Whipflash2 disk recorder Panasonic DVD recorder Sony disk recorders

The word combination “disk recorder” is no wonder today, but there was a time…

First of all, let’s make the term clear. DDR — digital disk recorder — a device for recording and playback of video- and audio file groups from the hard disk drive (sometimes from a digital medium like CD-R, MO, DVD-R etc.)

Just like many other hi-tech products, DDRs have no conventional classification. However, they can be divided, by some of their properties, the first and foremost of which is its area of use. Such a division is nothing but conditional, for many DDRs may be used in diverse areas, and still, some are extremely specialized.

DDRs became necessary as the “computer graphics” notion emerged. That’s the reason why graphics is the area where they are primarily used. Their basic function is to send a graphic file group from a hard disk onto a video medium (e. g., a videotape). Its secondary task is just the opposite: video data input from a tape onto a hard drive. The secondary character of this operation is proved by the fact that most of the earlier DDR’s either didn’t have a video input module or had it as auxiliary. Naturally, present day models are equipped with I/O modules. The Abecas Company pioneered in that field making its first DDR about 20 years ago. Its main The core problem consisted in that the hard drives of that day could not run a stream exceeding 1 MB/sec, considering that an uncompressed video stream equals 20 MB/s. Abecas embarked on the course of producing specialized large-size hard drives that could process the requested stream size. Bulky pilot models took 3…5 min to get started and allowed to play back only 36 sec of graphics at the most, making up for it by providing a superbly high quality. The Accom company later took a different way that proved to be righter over time: they used a host of hard disks working in parallel. To put it simple, imagine the first disk playing back the first picture, the second one the second and so on. RAID arrays now use this technology. This kind of approach made DDR’s more flexible and enabled the sending of large data quantities. The emergence of speedy and high-capacity disks made it possible for DDR’s to use a single standard-size disk.

One of computer disk recorders Diskus from Abekas

As soon as the “computer graphics” term came into being, the necessity arose to send the very graphics on to media used in further video production, in other words, on to video tapes. Early DDR’s cost a whole lot of money, and logically only the largest studios could afford them. Some ’do-it-yourself’ guys didn’t hesitate in using a video cam to take images right from the monitor. You can picture what they got in the end. Not only did it impair the resolution but also limited the capacity to just one picture, since there was no possible way for those low-capacity comps to play animation.

The invention of frame buffers brought about a way out. Those were devices having a certain memory (a maximum of 2 MB) that could convert the works of the ’graphics pioneers’ into a video signal. Depending on the memory size, they could run one single picture in up to 768×576 resolution and 16 million color graphic palette. The best of those devices provided a quality that is still considered a touchstone. The first best selling frame buffer was TARGA cards made by Truevision Co., that were massively replicated by others. It was them that established the TGA format, a long-recognized standard for computer graphic artists. But one picture is one, and what’s to be done with a group of pictures, i. e. animation, given that that particular stage saw a huge number of animation software packages. Another solution was found — a pic-by-pic image output (picture-by-picture dumping). This technique, though, required a recorder taping with down-to-picture precision, a special RS-233<->RS-422 converter and software making all this possible. It soon became clear that the recorder mechanism could not stand such a brutal handling (too much rewinding) and pretty soon breaks down. At that time, picture-by-picture recorders as well as their repair cost a huge sum of money. But since there was no other way to do it one had to put up with the fact.

However, when non-linear systems had just come into use, those in need of maximum quality did find ways to avoid using hard disks. For instance, one of the first Russians who bought a TARGA2000EISA card (probably few of us will remember such a PC bus), found an SDK for it. They made a DOS program allowing inputting uncompressed data getting around the hardware compression module, right into the 20 MB frame buffer of the card. Considering that in that period of time a 10:1 compression for 2 MB/sec stream would be qualified as a good result. As far as I’m concerned, good indeed. The size of input data was measured in seconds, but well begun half done. Another simple way that fellow Russians were quick to resort to was using a virtual disk. You might remember the Ramdrive utility — some system programmers still use it — it allows setting up a system disk that does not exist physically and is RAM-based. Unlike a hard one, this disk will not have strong stream restrictions. So if we open a virtual disk of sufficient size it can be used for recording and playing back stream video. But here come new limitations: the capacity of such a disk is limited by random-access memory that is not cheap today, and used to be a LUXURY.

But is it expensive anyway? Supposing, our purpose is to create a virtual disk that will store a video material equaling a 30-minute record on a Betacam SP standard tape. According to modern broadcasting quality requirements, a 25 Megabit/sec stream is pretty acceptable, that approximately amounts to 3MB/sec (MPEG-2 compression algorithm). That is to say, you’ll need 3 MB to record 1 sec. Therefore it will take just 5600 MB to make a half-an-hour record. To have an idea of the size of the disk, put together ten 512 MB RAM modules and you’ll get something about the size of a cigarette pack. With the current memory prices, that will add to $ 1000. In fact the mass production version of this device will consist of nothing but a couple of chips and will be considerably cheaper, have the size of a credit card, just about five times thicker. It will naturally require flash not random access memory.

Think of the advantages of such a medium. Random access is available without rewinding. The size of the recording/reading device for a tape like this is unlikely to exceed that of a common floppy drive, considering that most space will go to the quoted chips, digital-analog and analog-digital converters (though in all likelihood they will get obsolete after the digitizing process is complete) plus maybe a power supply unit. The absence of mobile mechanical parts will increase the reliability of modern-day broadcast recorders and their price will not go higher than that of a CD-ROM drive. Besides, a similar medium is birth-bound to be computer-compatible and it won’t take too much until every PC and MAC will have a built-in…let’s call it “RAM-drive” in the old fashion. Or imagine a laptop with a built-in non-linear real time system that doesn’t need hooking up with a recorder: all you have to do is insert your memory card into the side slot and go ahead. Dreams, you might say, but are they so far from real?

It’s been a long time since flash memory cards to be inserted into the laptop PCMCI-slot, started to exist, just like their little brothers, still more petite Compact Flash cards are now commonly used in pocket comps, digital photo cams and even mobile phones. There are pilot samples utilized as microcassettes for video cams, Sony Memory Stick is worth mentioning in this context. In spite of being costly and of limited capacity, they will neatly fit into the concept picture we’ve painted.

Regrettably, these steps are too tentative to influence the whole state of affairs. What’s going wrong then? The fault is not with technological limitations for sure, as they have long learned to make a reliable memory, without failures, so to speak. Moreover, several super high-capacity monocrystal memory units have reportedly been patented and shelved. Any detective or inspector will say, ’Look for the one who benefits from it’. Can you visualize the multitude of companies, small and giant, involved in producing precision mechanisms for tape recorders, hard disks and various drives? Plus those manufacturing videotapes and finally, computers! Giving up hard disks will eventually lead to a partial or even complete change in their architecture, and, in particular, require upgrading operating systems!!! I’m no economist, and yet am able to foresee what a vigorous world-market promotion of this tech may cause to happen — from financial collapse at the least to a world economic crisis, at the most. That might be the reason why we are still using Winchesters disks and tapes. But a ’velvet revolution’ has been under way and there’s not much time to wait.

DDR’s are broadly used in TV production, but their main niche is storing and airing the most frequent videos like screen themes, billboards, ads and commercials. DDR’s are truly unrivalled in this area. Many Ostankino TV Center employers remember a funny story when an assistant girl was in such a hurry to set the right screen theme that she managed to stick two (!!!) tapes into the recorder. DDRs feature the option of instant access to any video, which will not allow anything like that to happen. Besides, some of them can work by a playback list; in other words, they can play video clips in whatever order, regardless of their disk location.

Abekas 6000

In addition to this job, DDR’s are frequently used in transmitting sports events to replay crucial images in ’Slow motion’ mode. For this purpose, they may be equipped with additional Slow motion controllers.

DDR’s are also used as video sources in the tone-studio recording of film soundtracks. In this case they normally resort to expensive Betacam SP VCR’s. The peculiarity of this job (frequent rewinding) wears out the mechanical part of this hardware, and consequently, leads to significant mending expenses. Introducing inexpensive DDR’s featuring no complex mechanism allows to cut to a great extent the expenses and facilitates instant access to any part of the video. In this connection it’s appropriate to refer to a small German company named Steenbeck. One of its first models was a device called FED, designed precisely for recording soundtracks. Its basic feature was the possibility to utilize magnetic optical disks. The model in question did not get broad expansion for a number of reasons, the main of which, to my mind, was that lack of consumer’s demand for it. FED was released at a time when most people were not ready to ’go digital’.

Alluring is the idea to use DDR’s in video filming through linking it up right with a cam or mounting it in the form of an on-cam device. The chief advantage of such a solution is giving up tape media and tape-related quality losses occurring through analog-digital, digital-analog, and strange as it may seem, digital-digital conversion. It is common knowledge there is no single digital format so far. Solutions like that have already been found, the working algorithm being the following: after finishing the filming, you take out the cam hard disk and put it right into a non-linear mixing station bypassing the VCR or any kind of conversion. You can connect the cam directly to the non-linear mixing station as an option. Yet the expensiveness and fragility of the media (hard disks) confines the application of this solution to areas that admit of those factors. E. g., pavilion filming. In this case, though, high quality requirements excluding compression and high resolution (at least 2K per picture) make the size of data so huge that on-cam devices are discarded, that’s why fixed DDR’s are normally used there. It’s too early to speak about bringing DDR’s into mass operation in this area as yet, nevertheless, there do exist precedents: enough to quote “Star Wars: Episode II”, in which certain parts were filmed with the help of HD format DDR’s.

The TV journalism application of on-cam DDR’s are still curbed for almost the same reasons: high media costs, their fragility and a small number of on-cam-DDR-compatible mixing stations. But then again there is a mass of variants of DVD-R home application. Paint yourself a picture: you insert a blank DVD-R into the cam, and right after finishing the shooting, you can play in your home DVD-player. Well it’s actually not as simple as that, there are problems to conclusively settle, plus the ’format war’ hampering the whole business, however, the away-from-tape trend is obvious.

DDR’s second most important characteristic is the type of compression used for saving the disk space needed to store videos. This categorizing can be made as follows:

No compression DDR’s. These, as we said, are basically used when working with computer graphics.

M-JPEG algorithm compression DDR’s. Currently most widespread for being the ones made and sold the most, and, secondly, because MPEG-2-DDR’s are still pricey. M-JPEG algorithm utilizes solely in-picture compression of 8×8 pixel image blocks.

Disk recorder VR from ASC

MPEG-2 algorithm compression DDR’s. They are spreading intensely today, especially in TV broadcasting, for the reason of being able to process maximum material with given quality and disk capacity. This is made possible by the fact that in addition to in-picture compression, MPEG-2 applies inter-picture compression. However, due to this very peculiarity, a lost key picture may ruin a whole group of pictures, sometimes consisting of up to dozens of pics.

Non-standard compression algorithm DDR’s. Of limited extension, though still may be utilized, e. g., in Wavelet compression DDR’s.

The third important parameter is whether the DDR has audio/video input. SDI (SMPTE 259M) (sometimes with embedded digital audio) or YUV analog component interface are normally used for broadcasting. Tone studio work requires as little as having a composite video or Y/C, and digital audio AES/EBU I/O’s or their analog equivalent. Some DDR models support the SDTI interface that enables sending data at a 4x real time rate. Having one type of input or the other is not the end in itself; it only counts in terms of hooking up the DDR with the existing or would-be hardware. E. g., if all your hardware is analog, having a digital interface would be a waste of money. There are companies that complete their DDR’s with an HD SDI (SMPTE292) interface, those models being created especially to work High Definition hardware, HDCAM, for instance.

Relevant in a DDR are address-time code I/O’s as well as the option to run the disk recorder by means of an external controller or computer. These kinds of addition make DDR’s more versatile.

Now it’s time to get to know some real disk recorder modules made by different producers. Not all existing samples are placed here; they are almost all of the goodies that are on the Russian market…