USB 2.0 vs 3.0 audio interface

[RexRed]

I have noticed very few audio interfaces have been switching to USB 3.0. There has been ample time for them to do it yet there are very few out there.

I have to admit I am unknowledgeable about what is happening behind the scenes with my audio interface.

I just plug it in and it works. There is so much else to concern myself with.

But, is getting a 3.0 interface really worth it?

I do not record more than one or two tracks at a time at 24bit 48khz.

But I use a lot of effects and often have a lot of tracks playing at once.

Where exactly would the bottleneck occur?

I have been learning that audio interfaces differ quite a bit from one another.

Everything from preamp microphone/headphone impedance, frequency response, and noise floor etc.

Is USB 3.0 really necessary?

I have a new Steinberg UR22C USB 3.0 interface and I noticed I have to crank the microphone knob up quite a bit.

The UR22C is USB 3.0 but I am considering also getting a Motu M4x4 but that is USB 2.0.

I like that the Motu has an actual meter on the front and it has a lower noise level and I assume it will also have preamps that accommodate my microphone/headphones better.

Can someone give me the ins and outs on what the CPU does versus the interface and why and when 3.0 speeds would be used or necessary?

I use a ton of effects and I also master at the same time that I mix.

Thanks in advance for any help on this inquiry.

 

[msmcleod]

FWIW my RME Digiface USB does 32 input and 34 output channels simultaneously without any issues using USB 2.

[RexRed]

Thank you for your reply MsMcleod,  wow, so you are saying that USB 3.0 is really not so important. 

 I figure (correct me if I am wrong) the effects are all done by the CPU before they are turned into a wave that is played by the AI.

Also, I use livecast to broadcast video and audio over the internet in 4k and I hook to another streaming computer also. It gets complicated for sure. 

But it still seems if the AI was doing a couple stereo streams both ways in low latency it will still not reach the bandwidth limit of USB 2.0

I do not like that I have to crank the microphone gain so much on the UR22C, that bothers me.  The meters on the Motu would give me a clear representation of the signal, this is also attractive. I also have heard that the UR22C is very noisy compared to other AI's but still not perceptibly.

If I can eliminate the USB 3.0 factor, that gives me the freedom to go with the Motu.  

I am still going to keep the UR22C but I want to try the Motu and see if the preamps have a better impedance for my microphone that seems to be gain hungry.

My headphones are also gain hungry...

I don't like that I have to crank the headphone and microphone knobs up most of the way. 

 

Edited March 19, 2021 by RexRed

[OPunWide]

I signed up for an account just to comment on this. I didn't know the forum still existed. I'm about to resurrect an old Sonar setup. The old computer died due to age related illness. It was still running Windows 2000 (long story...). Essentially I'm starting over because current computers don't support the ISA interface on my old sound card.

And that's why I've been asking the same USB 2 vs 3 question.

I was really surprised by @msmcleod's answer; dubious actually. I quickly found an article that supports it. 
https://www.soundonsound.com/techniques/usb-firewire-thunderbolt-which-best-audio#:~:text=The%20USB%202%20specification%20states,a%20throughput%20closer%20to%20280Mbps.

The bottom line being: 
"At the theoretical maximum USB 2 bandwidth, you’d be able to record just over 40 tracks of 24-bit, 96kHz audio, while halving the sample rate to 48kHz would give you 80 tracks. Staying at 24-bit/48kHz, consider a more realistic real-world USB 2 bandwidth of 240Mbps (a slightly conservative figure, giving us plenty of overhead to allow for the connection limitations discussed earlier): you’d still have the ability to work with up to 40 channels of broadcast-quality audio simultaneously! Yet there are some companies who squeeze far higher channel counts from their USB 2 audio interfaces by building their own USB controllers. These tend to be among the more costly options, due to the extra work and design choices that go into developing and optimising this sort of solution. By way of example, RME’s MADIFace USB is a USB 2 bus-powered 128-channel digital audio interface. This is made possible by the use of the MADI protocol for handling the data transmission, which is far more efficient than the native audio-over-USB standard."

Maybe I won't wait for USB 3 interfaces after all!

[Hatstand]

If my understanding is correct USB3 busses supply more power than USB 2 but the transfer speed increase is not that mind blowing. This is probably why a lot of manufacturers haven't jumped. The difficult thing with USB 2 is that I believe it runs at the speed of the slowest peripheral attached to the bus, so if your audio interface is on the same bus as something that runs slowly it will be affected. Some people invest in a PCIe USB card specifically for the audio interface to run through. I have no idea whether this works or not though but I am sure there are articles out there that can confirm or deny.

[RexRed]

Great info OPunWide! Welcome to the forum!

That makes me think about something Hatstand.

How do you find out what is on your bus? I currently have all of my USB devices plugged into a PCIE card so my sound card is not reduced in speed.

But, wouldn't a mainboard's USB 3.1 or a USB 2.0 slot also be on different internal busses?

It does not seem likely that by using a USB 2.0 device it would slow down the USB 3.0 or 3.1  slot.

It seems each generation would have its own dedicated bus.

How do you find out which USB slots are on which bus?

And how does this all figure into lanes?

I have 44 lanes but I also have two PCIE 16x graphics cards. 

And how much of a limitation is the USB power?

Is it enough were they have to limit the amount of chips they can put into the USB 2.0 AI? 

Edited March 19, 2021 by RexRed

[jackson white]

https://www.rme-usa.com/rme-usb-technology.html

[rsinger]

10 hours ago, Hatstand said:

If my understanding is correct USB3 busses supply more power than USB 2 but the transfer speed increase is not that mind blowing. 

Not quite true. USB 2 is 480 Mbps and 3 is 4.8 Gbps which is an order of magnitude faster (and 3.1 is 10 Gbps), but transfer speed doesn't necessarily mean reduced latency. Part of it is that there are more data lines with USB 3 so some data is sent in parallel, but there is also a question of how the data is handled by the host. If the host system responded to data every time a byte was received, if enough data was sent, it would be possible to flood the system so all it did was respond to data and didn't have time to process it. I don't know what the protocol is for USB, but apparently the way the host handles incoming data isn't any faster for 3 vs 2. Since it isn't reducing latency it's not a selling point and vendors aren't flocking to it.

Edited March 19, 2021 by rsinger

[Hatstand]

3 hours ago, rsinger said:

but apparently the way the host handles incoming data isn't any faster for 3 vs 2

that is probably what I meant to say but I didn't express it very eloquently

 

[Will.]

USB 2.0 was released in the year 2000, which is also referred to as Hi-Speed USB. It is the most common version of the USB standard that we use every day. The maximum transfer speed of USB 2.0 device is up to 480 Mbps. Nowadays, USB 2.0 is being replaced by 3.0 in many high-end motherboards.

USB 3.0 was released in November 2008/2009, which is the third major version of the Universal Serial Bus standard for interfacing computers and electronic devices. USB 3.0 has gradually become the new standard for USB devices by providing a lot of improvements over USB 2.0.

The major one is the data transfer rate. The data transfer rate of USB 3.0 is up to 5 Gbit/s, which is about 10 times faster than the USB 2.0 standard.

So theoretically USB 3.0 is better than USB 2.0

[abacab]

Sure, but does it improve latency?

[Soundwise]

Here is some DTR info.

• USB 1.0 = 12 Mb/s ÷ 8 = 1.5 MB/s
• USB 2.0 = 480 Mb/s ÷ 8 = 60 MB/s
• USB 3.0 = 5,000 Mb/s ÷ 8 = 625 MB/s
• USB 3.1 = 10,000 Mb/s ÷ 8 = 1,250 MB/s
• USB 3.1 Gen1 = 5Gb/s (625MB/s)
• USB 3.1 Gen2 = 10Gb/s (1,250MB/s)
• Thunderbolt 1 = 10Gb/s (1,250MB/s)
• Thunderbolt 2 = 20Gb/s (2,500MB/s)
• Thunderbolt 3 = 40Gb/s (5,000MB/s)
• Samsung 970 PRO M.2 NVMe = 21.6Gb/s (2,700MB/s)
• Samsung 970 EVO M.2 NVMe = 20.0Gb/s (2,500MB/s)
• Samsung 960 PRO M.2 NVMe = 16.8Gb/s (2,100MB/s)
• SATA 3.0 = 4.8Gb/s (600MB/s)

 

 

[John Vere]

What was said about the bus power is the main difference. I bought a Motu M4 which stupid me, didn't listen to my own advice,  and it is bus powered ONLY. No option for a wall wart. Sure enough I was having issue right away. Motu was quick to respond with a long list of computer tweaks we already mostly knew. But the main advice was that I need to purchase a PCIe USB 3 card to power it up with. ?? Hey you guys didn't say that in your adds-- "This interface doesn't come with a power supply or even a jack for one so plan on spending $30 to make it work on most systems."  Even though my computer has 4 USB 3 jacks I need them for transfers etc. Apparently the M4 doesn't like to share so therefore a USB 3 PCie card.  

[Maestro]

Some USB Audio Interfaces have issues when plugged into a USB 3.x Port.  With PCs offering less and less USB 2.0 ports, I am avoiding USB 2.0 interfaces moving forward, because I'm lazy and like things to "just work."

[abacab]

"Some". I think that was an issue with some early models. I don't see any reason to avoid today.

I have no issues at all with a recent model Focusrite Scarlett (gen 3) USB 2.0 running on an Intel USB 3.1 Host Controller (it "just works").

[Bill Phillips]

On 3/20/2021 at 4:52 PM, abacab said:

Sure, but does it improve latency?

I don't think so. The USB 3.1 interface might. I'm wondering why there aren't any PCIe x1 or x2 interfaces which I suspect could drastically reduce latency. 

[abacab]

30 minutes ago, Bill Phillips said:

I'm wondering why there aren't any PCIe x1 or x2 interfaces which I suspect could drastically reduce latency. 

Based on these examples from RME and others, you can still get PCIe, but you better be willing to pay for it. https://www.sweetwater.com/c703--PCI_Audio_Interfaces

[DaGeek]

On 2/12/2021 at 9:51 AM, Jim Roseberry said:

AMD has finally gotten their ultra low latency performance together with the new Vermeer (5xxx) series.

I've got a 5950x based DAW (I'm typing on it right now).

 

The 10900kwill out-perform the 5950x when it comes to ultra low latency performance (example below).

In heavily multi-threaded scenarios (video rendering), the 5950x will significantly best the 10900k.

 

Lets say you want to run Helix Native (plugin version of the Line-6 Helix guitar processor)... at latency equal to or lower than the hardware version (which is 2ms).

When it comes to ultra low round-trip latency, the Presonus Quantum is as good as it gets.

Set Quantum to 96k using a 32-sample ASIO buffer size.  This results in 1ms total round-trip latency.

Load Helix Native and create a significant patch using two 2048-sample Cab IRs, delay, reverb, etc.

The 5950x is the first AMD CPU to be able to sustain this ultra low latency scenario completely glitch-free.

With the Threadripper 3970x, you'll experience glitches.

Needless to say, this is excellent performance.

With the 10900k, you can actually set the ASIO buffer size down to 16-samples (sub 1ms round-trip latency)... and it'll sustain the load glitch-free.

It's the first CPU we've tested that could actually do this...

 

The 10900k is a $500 CPU.

The 5950x is a $800 CPU.

 

Competition benefits all of us.

  

 

 

 

 

 

Based on what Jim said on another thread I am planning for Thunderbolt in the near future. Of course there are many other factors involved (most important having one of these beasts for a processor) but I doubt you could come anywhere close to these results with USB 2.0.

[Jim Roseberry]

4 hours ago, DaGeek said:

I doubt you could come anywhere close to these results with USB 2.0.

USB-2/USB-3 audio interfaces can't get much below ~4ms total round-trip latency.

Thunderbolt audio interfaces can get down below 1ms total round-trip latency.

Think of Thunderbolt as "external PCIe".

[Tezza]

On 3/19/2021 at 12:28 PM, RexRed said:

I don't like that I have to crank the headphone and microphone knobs up most of the way.

Wall wart driven USB devices generally have more power than the USB powered equivalent. Louder headphone outputs and louder more concise main monitor outs.

I don't know that USB 3.0 will change this much.

Audio device manufacturers won't tell you this because they want to sell all of their different interfaces.

For example, Steinberg will tell you that if you are using one of their USB powered device's, you should use low impedance headphones but if you are using a mains powered device then you can use up to 300ohm headphones. I noticed more overall omph moving from a usb powered device to a mains powered device. Apart from being louder, the bass was more solid and present.

Perhaps some USB devices deal with the power different or USB 3.0 might be a bit better, I don't know, but to me there is a difference in power output to monitors and headphones between mains powered and USB powered audio devices.

I don't know if this extends to pre-amp sensitivity for microphones that you plug in to the audio interface, it might, perhaps the phantom power supplied to condensers is better on mains powered devices, I don't know. I'm using the UR44 and I don't seem to have any problems with power output or microphone pre-amp sensitivity but I use phantom powered condensers, haven't tried  dynamic microphones.