Low latency audio interface primarily for softsynth playing

I have an ageing UA25 which works well but is USB 1 and with Windows 7 there is a known issue where a USB 1 device attached to a USB2 port causes periodic CPU spikes. Nonetheless things work fairly well on my core i7 laptop but I'm thinking about replacing it anyway.
 
I've heard good things about the RME Babyface but this is rather more than I need and is not cheap - can anyone recommend a unit which ideally is
(a) externally powered and NOT buss-powered - ground loops with USB powered audio interfaces have been the bane of my life.
(b) Has 2 analogue in and 2 analogue out with balanced I/O for both (more outs could be useful but not essential)
(c) Has really good low-latency performance for realtime playing of softsynths.
 
I looked at the Focusrite Scarlett models but there is some question over the latency. Not sure about Roland's newer units; their new large device, can't remember the name,got slated by Sound on Sound for the worst latency they'd ever measured, which doesn't exactly fill me with confidence. Any recommendations or do I just go for the RME?. I don't much like breakout cables, though, I'd prefer everything on the box.
 

I'm really raving on my recently acquired Motu Ultralite Mk III Hybrid. It's a little less expensive than the RME - has two pre's +  6 additional line-ins, 10 bal outs plus phones. Nice rugged metal case, not bus powered with USB, can do Firewire also. Built-in mixer & effects for tracking. Does ASIO, WaveRT..
 
For the amount of features it provides, it's really a good value at it's price point!

no digital in and out nixes the Forte for me.
 

i gotcha...
 
but i was seriously looking at the forte, because of all it's good qualities.
but alas, no digi IN/out

Goddard

musicroom
As I sorted through these I ruled out the ones you couldn't bypass the internal preamps.

Why is that?
 
Assuming you are referring to line inputs, do you actually have any direct knowledge or evidence of non-bypassable onboard preamps causing any problem?
 
Are you aware that even very high-end A-D converters employ preamps on their line inputs for buffering the input signals before they are applied to the converters?

 
I tend to think that less gain staging areas in my signal chain is a good thing. And yes, I have ran my outboard pre thru a decent onboard pre before and found the sound to be slightly affected in a negative way. I have never had the opportunity to do this through a high-end converter though. I'm not doubting you for specific interfaces. Of course each manf will tell us what we want to hear in regards to the onboard pre not affecting the outboard pre's signal to converter. For a test, I think I will run my outboard pre through my new impact twin's pres and report back. However, I'm very happy I have the option with this interface to bypass the pre and send signal directly to the converters. To me, it's a failing when manufacturers don't offer this routing option as a SOP. 

A/D's will have likely have buffer circuit on their line-in's for isolation, but I wouldn't call it a pre-amp.

Goddard
 
"Gain staging" is a popular term, sounds so professional.
 
 

 
Thanks

SuperG

musicroom

Goddard
 
"Gain staging" is a popular term, sounds so professional.
 
 

 
Thanks

I call it 'git your **** together!' I don't know how many times I've completely tossed a mix and started all over because I got antsy and started working before dialing in gain levels first. If you find yourself putting putting in gain through an effects plug-in, you need to sit down and think about it for a bit.

I'm quite certain he was throwing a p/a dig. He didn't like my preference and many others for wanting to bypass an interface's onboard pre. I am not the most technical musician songwriter in the big ole daw world. But a few simple principles help to keep my recordings from sounding too bad. One is to use as few gain stages in my signal chain as possible. Not saying it's a showstopper, but if I can avoid that, then I will. So yes, I do pay attention to the small things, gain staging being near the top of the list. 
 
Carry on
 
 

Ruling out any audio interfaces (or for that matter any other DAW input gear like preamp + ADC units) just because the mic preamps aren't bypassed when using the line inputs, due to concern about minimizing the number of  gain stages in the signal path might seem a wise choice but in reality isn't necessarily so, and applying such a criterion as a general rule rather than assessing each potential interface individually is rather foolish as different interfaces will, unsurprisingly, differ from each other in various aspects including their input circuitry.
 
Just as an example, consider the line inputs on the low-budget Behringer "Ultragain" ADA-8000 mic preamp + ADC/DAC unit, where the line inputs are on combi jacks with the XLR mic inputs with which they are wired in parallel. Thus, signals applied to the ADA-8000's TS/TRS line input jacks are applied directly to the mic preamp (offering 60dB of gain), just like the signals from the XLR mic inputs, and then to input buffer/filter circuits for the ADCs.
 
So, absolutely no bypassing of the mic preamps in the ADA-8000, only switching of which signal, from the XLR or TS/TRS inputs, gets applied to the mic preamp. Very basic and relatively cheap to implement.
 
And yet, surprisingly for it low cost, the audio performance of the ADA-8000's line inputs is rather well regarded, standing up quite well against much more expensive gear having discrete or mic pre-bypassing line inputs. And notably, they are capable of coping with rather hot input levels.
 
But please, don't take my word (or the word of anyone else unless convinced of their technical competence and experience and lack of any commercial agenda). Do your own research and evaluation.
 
And yes, I'm well aware that the ADA-8000 has other issues (power supply, ADAT sync), but those are not relevant to the audio quality of its non-bypassed line inputs.

Wanted to chime in a moment about round-trip latency.
When it comes to round-trip latency, not all units are the same (especially USB audio interfaces).
Yes, the latency from A/D and D/A converters is consistent... as well as the latency from ASIO input and Output buffers (at a given size).
But... the audio interface's driver uses a hidden safety-buffer.  This is the X-factor when it comes to audio interfaces.
The best performing drivers use a small hidden safety-buffer (RME, Presonus Audiobox VSL, MOTU).
Tascam USB units are notorious for using a large hidden safety-buffer.
Most drivers don't allow any user control over the hidden safety buffer... so you're SOL if the round-trip latency is high.
 
12-24ms total round-trip latency is too high to effectively play/record and monitor in realtime thru software EFX/processing.
Get that number down to 5ms... and it's a whole lot more practical.
 
Round-trip latency is only a factor if you plan to record/monitor in realtime thru software based EFX/processing.
ie:  Say you want to track a DI electric bass at 1am... and you want to hear the results (as you're playing) processed thru an AmpSim plugin like Ampeg SVX.
That's when you need low round-trip latency.
If you're not playing/recording and monitoring in realtime thru software based EFX/processing, round-trip latency is not an issue.
 
Playing soft-synths is dealing with one-way latency (playback)... and is very roughly half the round-trip latency.
 
If you're buying a new audio interface, IMO it makes sense to get a unit that provides low round-trip latency.
Even if you don't think you'll use it.  Six months from now... you may be in a different scenario.  If your audio interface has high round-trip latency, there's nothing you can do about it (short of doubling the sample-rate).

Jim Roseberry
If your audio interface has high round-trip latency, there's nothing you can do about it (short of doubling the sample-rate).

Although unfortunately, with some interfaces, doubling the sampling rate also results in doubling the buffer size (to compensate), in which event the overall latency will still remain basically the same (although the ADC and DAC conversion latencies may be decreased, they're typically quite small in comparison).
 
Also, it is important to keep in mind that as buffer size is decreased (to reduce latency) a greater load will be placed upon the system (which needs to work harder to service the buffers more frequently). So not only interface latency but also system loading may need to be considered.
 
Info on which audio interfaces are able to offer good low-latency performance when running softsynths may be found here (see DAWbench VI (CV and NCV) test results):
 
http://forum.dawbench.com/showthread.php?1548-Audio-Interface-Low-Latency-Performance-Data-Base&p=16069#post16069
 
http://dawbench.com/audio-int-lowlatency2.htm

First of all, thanks so much for all the people who replied here; Jim, of course, in particular, with his very detailed response.
 
I eventually decided to give the Scarlett 6i6 a try (and have posted brief comments in a separate post). For the price point it offered me four analogue ins and four outs, all balanced (afaik), two of the inputs will handle mic/inst, which is all I really need.
 
The Centrance latency test indicates a 6.4ms round trip latency at minimum buffer size. Reaper indicates a 3.3ms latency one way which is pretty consistent with that measurement and at that setting I've tried both three instances of Kontakt playing large sample sets and the 'ReaperTrek' project with around 18 MIDI tracks feeding into a single instance of Garritan Personal Orchestra (ARIA player) and can see no evidence of any dropouts or glitches with CPU sitting around a few percent (core i7 Acer laptop).
 
So at present given the cost differential between this and the RME products (and I did not much care for the Babyface breakout cable approach), I am very pleased with it. It is also - hooray - not bus powered. I've always thought this was a profoundly stupid thing to do because it just about guarantees ground loop problems due to the very spiky current draw down the USB cable causing a very low impedance noise source that can be very difficult to eliminate. I mostly got rid of the problem with the old UA25 by using some very heavy gauge copper cable (insulated cable with a very flexible copper braid conductor which is sold for wiring up car audio systems) and attaching this to the main mixer as a bonded ground connection back to the UA25, with a spade lug attached to one of the rear panel screws on the UA25, but it was still necessary to use ground-lifted cables between the mixer and the UA25 and with the Focusrite all these problems appear to have vanished - fingers crossed.
 
Incidentally I did look briefly at the larger 18i8, which does offer four mic preamps. I don't really need that but it is quite cost-competitive. However you only get two line outs on the back panel and I wasn't quite sure how the two front headphone outs worked in relation to the other outs - on the 6i6 they monitor line ins 1/2 and 3/4 respectively, so it seemed like you actually got less outs on the larger unit. And it is a good deal larger, physically; not an issue in the studio of course but I was thinking about potentially gigging with the laptop, where the smaller 6i6 would be easier to handle. In any case, I figured if I really needed more I/O then the (non-optical) SPDIF in and out on the 6i6 could easily be wired into another convertor to add another stereo in/out set.