IEEE 1394a (Firewire) faster than USB 2.0?

[getuhgrip]

I've not used any firewire devices yet, and I'm considering one of the Presonus or Tascam interfaces. I see they come in both flavors. Is one faster or better than the other?

Are there any variances of firewire ports that might not work with a particular device? This 1394a is apparently what my mobo has. It's got two different sized sockets listed as 1394a. What's up with that?

Thanks

 

[bball jones]

I was runnin a motu 896 on a mac with nary a stutter or burp. I'd patch an external fw drive to the motu and record direct to that external drive. Couldn't find a usb drive that I could record direct to.

 

[TimOBrien]

Yes firewire is faster than usb2.
The specs on usb2 are for BURST data transfers... useless to us as we are interested in SUSTAINED high-capacity data transfers and firewire walks all over usb in that department.
Firewire400 has the bandwidth to carry over 100 simultaneous cd-quality channels, while usb craps out after a handfull.
Firewire800 can do double that and they'll already approved specs for 1600 and 3200.
USB3 is just coming out but it will be years before we see any devices for it.

If your mobo has different sized firewire sockets, one may be 6-pin and the other 4-pin.
The only diffference is that 6-pin carrys dc power to the device (all the data is carried by the other 4pins.) I've used a 6-to-4pin cable on my Motu828mkII for years with zero problems.

 

[getuhgrip]

Thanks Tim. Firewire it is.

Does the 1394a reference have any specific significance? I mean, any firewire interface in going to be compatible with any version of firewire port, right?

I'm probly over-complicating this. I just don't want to run into any unseen heartbreaks the day I get this stuff.

 

[TelePaul]

Thanks Tim. Firewire it is.

Does the 1394a reference have any specific significance? I mean, any firewire interface in going to be compatible with any version of firewire port, right?

I'm probly over-complicating this. I just don't want to run into any unseen heartbreaks the day I get this stuff.

Generally, Texas Instrument are the preferred 1394 chipset.

 

[getuhgrip]

Generally, Texas Instrument are the preferred 1394 chipset.

Cool..I've got a TI calculator. I'll rip it open and grab the chip. Will it still work if I just tape it to the ....what am I spose to tape it to?

 

[drstawl]

Cool..I've got a TI calculator. I'll rip it open and grab the chip. Will it still work if I just tape it to the ....what am I spose to tape it to?

Tape?

You should know it needs to be glued into place.

 

[getuhgrip]

Yeah, glue is better!

I thought tele was bein' a smart ass! Ooops! Just checked mine in device manager....guess what? Texas Instruments!

 

[dgatwood]

The "a" was added later to distinguish it from 1394b, a.k.a. FireWire 800. Technically, I think 1394a is a subtle spec revision that also means that devices are expected to comply with the stricter inrush current limits from the 1384b spec and are expected to fall back correctly to S400 (and not S100) if they see an S800 device on the other end of the wire, but I'm not positive about that.

As for the two connectors, one is probably a 4-pin camera connector. As I've heard it, the folks designing the spec begrudgingly conceded to Sony's pressure to put in a 4-pin version of the connector for camcorders and other devices that couldn't provide bus power... which they and a few other laptop companies immediately started to abuse by putting the connector on laptops....

In short, it's just like the six-pin connector except that it only supports devices that can provide their own power (e.g. some hard drives, camcorders, etc.).

 

[RawDepth]

...Firewire400 has the bandwidth to carry over 100 simultaneous cd-quality channels, while usb craps out after a handfull...

That is not true. USB2 can carry over 450 CD quality channels. Let me explain...

The sustained data rate usually reaches around 320Mbps (or 320,000,000 bits per second) for bulk transfers in a best case scenario. If your interface converts audio at a sample rate of 44,100 and it takes 16 bits to describe the value of each sample, simply multiplying 44,100 x 16 gives a data stream of 705,600 bits per second (CD quality, streaming through the USB plug.) This does not even come close to the available data transfer rate of 320,000,000 bps for USB2.

You would have to record over 450 mono instrument tracks simultaneously in order to exceed the capability of the USB cable connection.

On the other hand...
The real drawback with USB is in the fact that all data I/O functions are handled by the computer's cpu. If the machine lacks memory or is burdened with other duties, you will likely experience lag if recording high track counts.

 

[jasonbmerrill]

usb = latency jitter

 

[dgatwood]

And USB controllers are seen b hardware manufacturers as likely to be used for unimportant devices, and the most common test is "does my mouse work", so the USB ports tend to share interrupts with things like the GPU or Wi-Fi, whose drivers hold off interrupts too long and cause the usable isochronous bandwidth to be close to zero. It's not inherent, just common. With FireWire, the most common test case is reliable DV transport (@25Mbps), so serious IRQ steering problems are relatively rare.

 

[TimOBrien]

Does the 1394a reference have any specific significance? I mean, any firewire interface in going to be compatible with any version of firewire port, right?

Saying 1394a is a way to not pay Apple a royalty for using the term Firewire.

Yes, for audio and video you ONLY want to use Texas Instruments firewire chipsets.
(You will see the term OHCI compliant.)
Belkin and SIIG make good reasonably-priced Firewire cards with TI chips.
Stay away from combo firewire-usb cards; they've been known to cause problems.

 

[TimOBrien]

That is not true. USB2 can carry over 450 CD quality channels. Let me explain...

The sustained data rate usually reaches around 320Mbps (or 320,000,000 bits per second) for bulk transfers in a best case scenario. If your interface converts audio at a sample rate of 44,100 and it takes 16 bits to describe the value of each sample, simply multiplying 44,100 x 16 gives a data stream of 705,600 bits per second (CD quality, streaming through the USB plug.) This does not even come close to the available data transfer rate of 320,000,000 bps for USB2.

You would have to record over 450 mono instrument tracks simultaneously in order to exceed the capability of the USB cable connection.

On the other hand...
The real drawback with USB is in the fact that all data I/O functions are handled by the computer's cpu. If the machine lacks memory or is burdened with other duties, you will likely experience lag if recording high track counts.

Again, USB = BURST. Firewire = SUSTAINED transfers.
You can quote mathematical specs but I'm talking REAL WORLD.

Show me a usb interface that has more than 16 i/os......

With my Motu828mkII I can hook 3 together for 60 guaranteed.

 

[steveneudaly]

If I'm not mistaken, USB bandwidth is split equally between all devices on the chain... meaning if your interface and mouse are running on the same hub, the mouse gets the same transfer rates as the interface.

Firewire gives bandwidth only to what requires it.

I think that's right.

 

[dgatwood]

Saying 1394a is a way to not pay Apple a royalty for using the term Firewire.

Yes, for audio and video you ONLY want to use Texas Instruments firewire chipsets.
(You will see the term OHCI compliant.)
Belkin and SIIG make good reasonably-priced Firewire cards with TI chips.
Stay away from combo firewire-usb cards; they've been known to cause problems.

The term FireWire is licensed royalty-free.

 

[dgatwood]

If I'm not mistaken, USB bandwidth is split equally between all devices on the chain... meaning if your interface and mouse are running on the same hub, the mouse gets the same transfer rates as the interface.

Firewire gives bandwidth only to what requires it.

I think that's right.

No. Some cheap USB hubs do slow down to 1.x speeds if a 1.x device is present on the hub, but that's because the hub is junk, not because it's the nature of USB.

 

[dantheman]

Hi,
I'm looking into USB/PCI/Firewire interfaces now, as I'm getting ready to buy a recording laptop. I'm probably going to get a Dell Studio 17, which has a 1394a port.
I'm trying to decide if I should go with a firewire interface but I've been hearding about this TI chipset issue and I'm confused.
Is the important chipset (with regards to Texas Inst. or not) the one in the cpu processor or the one in the interface?
Sorry, I'm pretty noob to this stuff and trying to take it all in.

Thanks.

 

[Bristol Posse]

Hi,
I'm looking into USB/PCI/Firewire interfaces now, as I'm getting ready to buy a recording laptop. I'm probably going to get a Dell Studio 17, which has a 1394a port.
I'm trying to decide if I should go with a firewire interface but I've been hearding about this TI chipset issue and I'm confused.
Is the important chipset (with regards to Texas Inst. or not) the one in the cpu processor or the one in the interface?
Sorry, I'm pretty noob to this stuff and trying to take it all in.

Thanks.

Laptops can be problematic. It's difficult to say who will be making the Firewire Chipset on a dell at any given time but the chances are good it is not Texas Instruments.
This may or may not be a problem. Most FW manufactureres test their equipment on Texas Instruments Firewire Chipsets and recomend a Texas Instruments chipset for use.
Another FW chipset may or may not be a problem. TI is preferred because they do not cut a lot of corners on their execution where as a lot of other Firewire Chipset (dice II, Ricoh, VIA etc) companies do to save money. For laptops FW is often an afterthought that is just stuck in as cheaply as possible and the onboard FW is often very badly set up sharing IRQs with all kinds of stuff that will cause it to not function well (Like Graphics processors)

If you are looking at recording less than 4-8 things at the same time then USB could be a better option on a laptop (although you may need to look at some IRQ re allocation and turning off hardware on the laptop to get clear recordings free of pops and clicks)

Firewire and USB are both ust protocols for streaming data. Firewire offers more sustained bandwidth and so is useful for times when you need to stream a lot of data at the same time.
Fire wire vs USB doesn't make any difference to the sound of the recording (so long as you have enough bandwidth to avoid pops and clicks).

once you have figured out what your bandwidth needs are pay more attention to the Quality of the featureset, Preamps and to a much lesser extent converters of the audio interface. they will actually impact the usefulness of the unit and quality of the recordings you get. The type of connection is irrelevant so long as it has sufficient bandwidth to cover your needs, by the time the signal hits the wire (USB, Firewire or PCI) it's already been converted into ones and zeroes

 

[dantheman]

Nice! Thanks, Bristol. That's a lot of really good info.
I've heard that laptops are tough but I think I can do it. I have to get a laptop anyway, and it will blow away my 6 year old desktop in power, so I figure I'll try. It's going to be the laptops primary function.

So it sounds like, at least in my situation, I should avoid firewire. From what I've read Dell uses Ricoh in their Studio 17s...their chat/sales people don't know who makes the chipsets. I know next to nothing about processors, but the Studio 17 seems powerful enough: i3-330M 2.13GHz (3M cache) / 4GB Shared Dual Channel DDR3 at 1066MHz.
I don't know anything about IRQ allocation, but I am going to strip this comp down to barebones, as far as software on there.

Now my question would be, will I run into any problems trying to record 1 or 2 tracks by myself 75% of the time and with friends probably using 4-8 tracks the rest of the time? Using the Studio 17, with those specs, and a USB interface?


Thanks!