Yes perhaps USB is faster but higher CPU overhead, not what you want for capturing. Plus as far as I know all cameras / VCRs only output true DV down firewire.
If you just mean connecting external drives, well some seem to do OK with USB 2, I always stick to firewire, again because of the CPU overhead thing.
It seems most external HD cases are now being made without firewire - just bought one with USB2 and SATA ports, no FW.
Fortunately USB2 seems to keep up with DV ok - I've been told that although it has a higher max speed than F/W, the latter has higher sustained speeds, but no problems so far ....
Firewire is for streaming audio and video - an asynchronous protocol. So dropped frames with a correct f/w path, should be non-existent - it is a stream.
USB is for transferring files. The file isn;t being sent correctly? then it gets resent. If an issue is encountered in the "file" transfer, then a "Packet" of info is resent. Audio and video streams are just that "streams". It is like opening up a tap of water. "OOps! I kinda missed that last drop of water (frame)!! I need to go back and get it! Stop the faucet!! I can't?" The stream has moved on . . .. the frames will NOT be resent! "Elvis HAS left the building!"
IMHO? - The present "advances" in USB are good for "other" work. I'm saddened by the demise and removal of a firewire option on external drives.
USB? Great for storage and "file" validity and consistency. You wanna make sure all is neat, and tucked up cosily in bed, with all its toes and fingers in place? Then USB. Streaming? I kinda think of the analogy above AND the way a rope gets paid out when you see an anchor descend to the bottom of the river. Firewire SHOULD keep the whole length of rope. USB - it'll question if the last bit was passed correctly. Look, if the USB is top-hole, top-shelf then one shouldn't get an issue. But this "one" = ME, will never - other than backup and so on - feel secure using it as a streaming device. No biggie - I would just need to re-capture, re-edit and maybe even re-connect and start working just what went wrong. Firewire? It kinda just works.
Firewire designs-OUT streaming issues. It is a real-time option.
USB is perfect for known-length files and Users who do need liberate that extra bit of real estate on their hard drive and for the purposes of security on backups.
Must admit I hadn't noticed the demise of 1394 on external drive boxes, most of mine don't even have USB. Certainly Apple seem to have killed off Firewire 800 which is sad, I've got one PC with 2x800 ports, love them, MUCH better connector too.
But eSATA is the way of the future for external drives, wickedly fast but kind of only point to point I think.
Then again GB ethernet is more than fast enough for DV and you can run that all around the place.
Well I walk around places like PC World and have been told by the urchins there that they will NOT be re-stocking f/w externals anymore. I can't speak for your experience, whether you noticed or not. My experience - they started to evaporate along with the f/w cabling - this for PC World - about 18 months back. Just did a web search on PC World: Freecom using USB 2 or Ethernet. Kinda wraps that up? And yes there are OTHER web suppliers doing f/w externals.
If I now wish to obtain f/w I have my sources. If you are a newbie to this stuff and are asking "Capturing Firewire or USB2?" then I reckon' PC World, or its equiv, is what is only on offer.
Maybe eSATA is for this chap, I dunno, but it wasn't in the question? Plus, I don't see them in PC World and the places I wander about. And it isn't what we had laid on the table right now. Yeah?
But what I attempted to do was to explain the pros and cons of F/w<>USB - as in the question? Not whether it is a Universal truth as to the "availability" of f/w. However, as an additional thought/observation I was finding it difficult to obtain these externals. Did I explain that well enough - Bob?
Here's a thought, do you think that the present, first point of contact for someone asking this question, could be PC World? What do you think?
Tell your son that it’s not about speed, it’s about throughput. Throughput is speed plus bandwidth. The problem with USB is that it SHARES the bandwidth across all of its ports! So a 4-port Firewire card has 400Mbps on EACH port. A 4-port USB Hub (like the kind found on your motherboard) has 480Mbps divided by 4 ports yielding only 120Mbps for each port (because it reserves bandwidth even if it isn’t used). I have read about newer USB cards that claim to not reserve and so provide full bandwidth as long as you’re not using the other ports but that is the exception and probably not what your motherboard is doing.
In addition, the USB protocol uses the CPU to transfer data. So your CPU speed has to be added into the equation and will be used during transfers. Firewire has no dependency on the CPU and keeps right on transferring data while the CPU is free to do other things.
In every way, Firewire 400 (IEEE1394) is superior to USB 2.0.
A few weeks ago I got serious with a backup program called Acronis TruImage, and am planning to use it in place of my old reliable Retrospect for imaging and archiving purposes. I was surprised by seeing on that product's message board that many users were suffering from corrupted backups, which is definitely not a good thing for backup software. All had been using USB2. Not a single case of Firewire corruption.
Then, too, keep in mind that 95% of users had USB2 to start with. Also, those little portable USB2 drives generally need a bit more power than a single USB port can deliver. Maybe in some instances the drive kinda works, but just barely, and the user doesn't realize it.
Thanks ~jr Now I can go kick his 17-year-old-know-it-all-self-assured little tush - they need that at times :-) Or I as his Dad need to not feel left behind as he has surpased me in many areas already ..... ( I hope spelling is one of them)
((Garo))
I liked your explanation Grazie. Yes, definitely Firewire between camera and computer.
For drives though? I've had tons of trouble with firewire drives and so have other people I know. Hate, hate, hate them. Afraid they'll fry my camera or deck while they go Phzzt. Had one firewire device fry a few years ago and take everything else that was firewire with it to the grave. It was an ADVC100 that fried.
Question's not too clear 'cause actual capture is usually from firewire so I presume the question is where to capture to?
When I've had to capture away from my workstation using my laptop,
setting the capture (via firewire) to save to an external hard disk via USB works fine & I've never had a problem. If I try to rendering the avi file that's on a USB connected hard drive I've run into BIG PROBLEMS, maybe because the IO is much too slow. Therefore, I transfer the captured file back to my workstation before working on it.
Example: tried to encode an avi I just left on the USB drive and it was still chunking away after 15 hours (& 50% done). I cancelled that, transferred to the hard disk and retried -- it took 6 hours to render.
Video Stream Capture to PC via F/W DV Capture via CANOPUS ACEDVio 4-pin
then OUT
ASUS MoBo > F/W to daisy chained Maxtor externals.
My point is that I have a clear distinction from CAPTURE (CANOPUS card) and my thence ONWARDS (ASUS MoBo)to the daisy chain of externals.
No USB. Just streamed thru' . I did try using the MoBo F/W for cap and chaining, but it was bonkers.
Rendering using my recipe works. Capturing using this setup works. I can go IN/OUT via VHS VCR . .or . .Camera( I don't! But have!) mostly Pannie Deck INs/OUTs . . all using the f/w connections. No USB.
DV IN/OUT via Canopus. From here via ASUS MoBo F/W to externals.
Point being is that I have learnt to "separate" the capture (card) process from the Mobo (card) process. Others might have had success, but here I've learnt NOT to expect everything of my MoBo.
This was the same for my laptop days. Dell 4-pin Digital Video f/w Video Solution IN/OUT. Then I added a MAXTOR dual PCMCIA F/W ClassII card for my firewire externals. This broke me free from the trouble I had using the single Dell 4-pin "Video Solution". I've carried this concept onwards to my MONSTA! pc. One card does capture leaving the MoBo free to process the "onwards" traffic to the external firewire drives
No USB in the chain - none at all!
Oh I can render out to my f/w externals in about the same time as my internal drives. Streams, streams, streams - the MoBo and all the other stuff aren;t dealing with files like USB does. Streams!
I guess that's why the USB rendering took so much time? Huh?
Hmm. This post and what has been written throughout has a valid background but the evidence that runs through it is almost flawed throughout.
Some facts:
1. Editable formats are readily available through the firewire host. DV isn't readily available via USB2 with just a few exceptions (e.g. Creative Labs Audigy 2 ZS Video Editor in DV passthrough mode - typ via CL-VidCap).
2. MPEG-2 and MPEG-4 are readily available via USB1.1/USB2. Whereas MPEG-2 is readily available on firewire only in the HDV family of resolutions in prosumer kit.
3. Gigabit ethernet is wholly more appropriate for the above applications given the point to point mode that it can operate, the simple cable infrastructure (ie DIY cables are easy) and the almost no-additional-drivers-reqiure approach that could be used. However no consumer or prosumer camcorder is using this. Some STBs provide ethernet access to media files. Gigabit ethernet can sink uncompressed SD and often mathematically lossless but compressed HD given the right southbridge or PCI back end implementation. Gigeth is the closest economic interface we have that could replace SDI and HD-SDI.
-----------------------
A. FW, USB and ethernet all have isochronous modes of operation. Where bandwidth is reserved.
B. USB driver and interrupt/memory interfacing is (to date) harder on the CPU than FW. This situation could change if we didn't want cheap all the time. Until then, despite the 480Mbps and 400Mbps separation between USB2 and FW400 respectively, the FW implementations we have available today are the sweeter choice. The protocol isn't at fault, and the USB hardware (to date) is only slightly inferior to FW - it is the drivers that really make the difference. I'll concede by comparing USB with ATA and FW with SCSI. FW usually has more of a quality approach in the implementation. Just as with SCSI U320 10+krpm drives, it isn't just the speed of the spindle or the better protocol that makes these drives better. They are made better for harsh environments and for folks who want reliability over price.
C. Despite the extra effort to create a better product where you see the firewire branding - FW has a history of poor electrical isolation. The spec is right, just that camcorders, hubs and other peripherals tend to be made in a way that when a power surge occurs, the port dies and needs repair by a technician. It usually works out cheaper to buy a new device. So unless we fight this in the courts, the manufacturers win. Jaded I know, but this problem doesn't appear anything like as prevalent in USB devices and is almost unheard of in ethernet setups (even with power-over-ethernet.
D. The implementation of USB, FW and ethernet is a single port or NIC per motherboard or PCI card. If you have multiple interfaces, you are on the same bus - with the exception of some ethernet cards with multiple controllers onboard. Please forget the notion that 4 port firewire cards have 4x 400Mbps port incarnations. They don't. The perception may be that you can get high duty cycle transfers on them, but this isn't a function of better plumbing.
E. External drives with USB2 and FW presentations are slower on the USB2 side primarily due to the poor technology in the bridge circuitry in the enclosure. Firewire had similar speed lag issues in the early days of 1st gen Oxford 9xx controllers. These were improved as the droves of people flooded from all market sectors to grab these handy external units. By the time USB2 was reaching popularity, the desire for speed had switched to the desire to support the largest capacity drive in the cheapest enclosure. A missed opportunity.
F. eSATA and Gigabit ethernet NAS/SAN solutions are likely to overshadow nearline hard-disk storage over time. Cardbus SATA controllers for laptops and motherboards with built-in ESATA capabilities are going to steer external drive buying folks back down the performance route and with the economy but firewire and USB are going to take a back seat. This is just as well as the patent holders have probably made their fortunes (USB is Intel I think, FW is Apple - or whoever is behind the patents - it is a bit like what differentiates WiFi-WLAN from Bluetooth-PANs).
G. USB2 can have DV 25Mbps running over it quite easily with no need to be worrying over dropped frames in either direction. Remember I explained that USB2 and FW are natively Isochronous, ethernet can have isochronous arbitration added to it (and it matters less if you only put the single peripheral onto the NIC, one at a time anyway). If you get dropped frames, change capture app or your operating system! As for USB2 being error corrected but FW not - forget that too. DV over firewire is aiming to be fire-and-forget (almost simplex) - but really, both protocols have the same realtime characteristics. USB is however more of a peripheral dumping ground through it's popularity as much as anything. FW-mice and FW-keyboards are unlikely to catch on. The use of USB-mice and USB-keyboards are perhaps a good justification for putting an additional USB2 controller into a PC that you use for realtime or fast peripherals. However this is being a bit picky.
H. The electrical spec of FW and the risk of damage could be enhanced/mitigated if folks protected their ports by adding the components that the manufacturers have missed out by short-cutting the recommendations. Resettable fuses would be better than signal diodes that simply blow. I've written about this before, but this is the single most ridiculous part of the firewire implementations we have in the market today. If our camcorders weren't also "western world jewellery" - I am quite sure that there would be a swell towards receiving satisfaction over when these ports fail when folks connect them up hot (ie plug and play). If the DC power between host and the devices doesn't have enough in common (DC reference levels wise) - then you are putting the end that receives current at some risk. Possibly enough to break the port.
I. Nothing is likely to change any time soon. Camcorders may start to increase their use of gigabit ethernet, Infiniband or HD-SDI if we demand more than long GOP MPEG-2 / MPEG-4 acquisition. However, in the meantime this type of kit is in the hand of those above mere-mortal status. Even though there is less semiconductor in these camcorders on the back end (by virtue of them needing to do less compression).
Does that help position why the answer to whether you should capture video remains firewire? What was said about USB2 was unfair and incorrect. I hope this post goes some way to correct the mis-information.
Lastly, did you know that AMD is falling behind Intel, but the chances are that within 12 months it will be ahead again. What I mean by that, is that given the market opportunity, a technology can thrive. Peripheral interfacing is simply not a thriving industry. Take it for what it is today but don't dis the underdog if he might bite you some day. ;-)
Riredale, I use TrueImage and have both firewire and USB2 externals. I have never been able to complete a backup to my firewire external. USB's work fine.
Grazie ! - What you and other contributers said was useful and meaningful to the consumer and forum but was less than accurate for me the engineer to pass by.
I don't think I've set out to change the message, but to set the record straight. USB2 is the poorer cousin to FW but it isn't like that for the right reasons. The design and 90% of the current implementation of USB2 is good. I'm sure I could find you an example where a FW peripheral performs badly and a USB2 peripheral performs well. FW is 95% good in the current implementation but that has nothing to do with the ability to capture AV through it.
I read the posts on this thread the other night and let it pass. Heck, it certainly does appear to be the way you guys describe it. However the reasons for why the FW hose gives better service isn't quite how folks were claiming.
Almost all hoses we connect to our computers are capable of being general purpose and with the right amount of care by the programmer - none of the problems we've saddled the USB2 interface with through urban myth would be in our vocabulary. Some interface types suit certain applications better than others and are the go-to interface of choice for certain market segments.
The assessment of how poor the USB2 interface is for capturing has stemmed from people's real experiences - some with some fairly early implementations of kit. USB2 is a much newer interface than FW400 and this has to be expected. Camcorders only have USB2 interfaces to help consumer sales and they don't push much data down the wire as these are compressed feeds that are aimed at using the camcorder for MSN video calls more than any form of professional editing purpose (again to date this is the case).
Windows XP SP1/SP2 really made a mess of FW800, and so explains what the real trouble here is. USB2 OS and driver support in Windows is flawed and will probably be fixed as and when the software gurus understand what the engineers have actually delivered for them. USB2 needs constant improvment - which the southbridges on our motherboard would be the first place to look for such an enhancement. Maybe VIA will do a decent USB2+FW400+FW800 card but I'd personally look to a different IC fabricator/integrator - thanks very much.
IMHO USB2 remains a safer interface to plug-and-play with and firewire remains the prefereable performance interface. In my book, both interfaces are overshadowed by the equally pervasive gigabit ethernet port (especially where that port lives on the southbridge with unfettered access to the northbridge (the CPU and memory). Gigabit ethernet is cheaper to implement, flexible, can share peripherals across multiple PCs concurrently, and with a bit of care it can also remain realtime for video applications. The licensing is cheaper and power-over-ethernet is safer and can sink more juice into the wire. Distances are also better if you don't daisy-chain device to device.
The protocol to perform a transfer between host and device in both USB and FW uses the same type of reliability mechanism. Both interfaces are also isochronous which means they are aware of time and bandwidth at all points. They can make the transfer asynchronously (at any time) but the conveyance is against agreed resource reservation criterion (I suppose a transfer that asks for 400Mbps or 480Mbps is not sharing bandwidth, but then such a transfer could be denied by the arbiter).
I suppose if this were the race track, or the forum for Fast Lane magazine, we'd be muddling over whether a supercharger was better than a turbocharger.....
One area that I didn't cover was that Vegas doesn't have a good history of capturing I-frame standard def video over anything other than DV, HDV or BlackMagic Decklink ports. The first two being firewire based bit-copy transfers.
The ability to capture from USB2 is really best left to oddball proprietary devices at this current time. There may still be some advantages in having these types of device in your arsenal. Especially if the USB2 device driver does have a capture mode and preview feed that Vegas can understand through it's WDM or VfW hooks.
The interface is just one of the layers in a communication.
I think the text that made me want to correct the messages placed here kicked off for me when it was claimed that firewire ports have the full quota of bandwidth per port. So a FW400 board with 6 ports would have 2.4Gbps of theoretical peak throughput in any direction of traffic flow. 4.8Gbps combining each direction to get an aggregate information flow figure- Not true! FW400 has 800Mbps aggregate channel bitrate - combining both directions. Fortunately this amount of bandwidth just about fits into what you can run on PCI32 at 33MHz. Although some southbridges have firewire presented on 266MByte/sec buses out of band of the PCI32/PCIe slots/buses. So as this was the main emphasis of my posting, I'm not really able to tell you where you have gone wrong specifically Grazie. From the impression made on you by what you have seen, you are not far wrong and the intent of your message is bang on.
Things progress, just as audio improvements were made by the industry with AC97 and then some more with Intel's HD Audio definition. Ethernet went through the CPU offload/acceleration hoop many years ago. TCP Checksums, DMA and vlan trunking became their improvers. Perhaps we'll see FW800 and USB2 having similar enhancements as the computer gets refined and made better with each generation.
I feel your pain with PC World ineptitude both in what they stock and who they use to sell their wares. I always thought the younger generation were the ones that understood how to programme the video recorder. Clearly they are busy recording for their parents better ends whilst their weaker minded mates are getting the minimum wage down at the local computer store. ;-)
Now, if only I could get hold of the £90 CreativeLabs Audigy2 ZS Video Editor with the downloadable proprietary release of VidCap - I'd then have more of a stick to wave. This unit supplies much of what you get with an ADS Pyro AVLink adaptor. Analogue capture to DV.AVI, DV.AVI to analogue (only SVIDEO and Composite mind you, no component), and DV to DV.AVI - all through a USB2 interface with (using VidCap) reasonably low CPU utilisation. You also have an onboard MPEG1 and MPEG2 encoder with image processing molarkie that you can use on the input hoses, or to decode in a hardware accelerated form for MPEG1/2 playback from the provided media player app. Interestingly - you get to convert DV.AVI to MPEG-2.VOB using the onboard hardware encoder (yes, that will be rubbish compared to MainConcept, but it will be quick when you need to call on it) by just running the sources through the VIdCap application. OK - it'll be a very disjointed workflow for Vegas to hook into. However it will suit my laptop very well (as I'll also get 7.1 audio support in there). All for 90 English notes. This is better than trying to get my camcorder repaired after I connected up to a PC with a powered belkin FW hub and lost both my Sony DCR and a Maxtor FW HardDisk (that continues to work but only via the USB2 interface - or IDE if I removed the wretched thing from the case).
So FW is nice unless you pop the port outside the warranty. Then I can tell you - you start to get the feeling that you've been ripped off when you think that this would have been a whole lot easier to stomach if your replacement camcorder was firstly DV based and secondly had a gigethernet hook to your PC. USB2 will be my modification to the setup, with a digital->analogue->digital step in the way, for the sake of economy.
In my research for what needed to be fixed on the camcorder (in the event that I fixed it myself or had a local repair guided by my foreknowlege) I discovered that the real world implementation on the circuits that drive Firewire has a propensity to miss out the port protection circuits. So where self resetting fuses were specified, the implementers have been putting in signal diodes. Blow them, no more firewire.
That is almost as depressing as finding you've got a dead pixel on your CCD/CMOS camera head or on your LCD monitor. :-)
I understood most of your tutorial. And as you say, I can only repeat the experience I've had. Once I waggle my tiny toes into the darkly deep waters of IT engineering, then it quickly turns into an accelerating descent to Davy Jones locker - and beyond! If that were possible.
What you said about not mixing acquisition with storage was right on the money. Don't cross the streamers!
You can pick and choose the interface spec between your PC and the extra/ext. storage - it is absolutely advisable to keep them separate where there might be a contention issue. However at the wire level, even that should work, but the drivers probably don't honour what they've been commanded to and slouch on the overall performance.
I mean, DV25 actually fits within i100/FW100. So there ought to be headroom. So the practical element is very important. Theorectical maximums and bandwidth reservation is a bit his and miss but it does get better and more accurate in it's form as the new generations of computers unfold on us.