[Jeremy Jongepier] I'd like to add that the 2i4 seems to be pretty much incapable of useable low-latency operation in some situations. Even with buffer sizes in the tens to hundreds of milliseconds it produces frequent clicks on the same system that runs a UCG-102 at 3x48/48k with no problems, a Thinkpad with an Intel chipset, 3.12.5-rt7. This kernel includes the patch[*] that was recently mentioned in an Ubuntu bug report by James Stone on this list, supposedly fixing lowish-latency operation of the 2i4. Interestingly, the clicks were never reported by jackd as xruns. Eager to move on, I haven't bothered to check if the clicks are caused by under- or overruns, or even both, let alone check which part of the setup was actually causing the poor performance. Tim [*] http://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=e…

[James Stone] The frequency of the clicks I experienced was related to the buffer size, and their sound was similar to the sound of xruns to my ears, so my first suspicion was xruns. Of course, with no further investigations made, it remains an unproven hypothesis. On the Thinkpad here there's no need to turn off the wireless to get low USB audio latency fwiw. I'm starting to feel uncomfortable playing the guitar through the box at latencies greater than around 4-5ms, so my upper limit for the buffer number and size is usually 2x64 or 3x48 at 44.1 or 48k. (Whether this is reasonable is another question, but I'm happy the criterion can be met by current hard- and software leaving ample CPU capacity for DSP.) Cheers, Tim

[James Stone] Tested the 2i4 again here with 3.0.89-rt118 and a variety of buffer/period setups, no luck: it's happily crackling away. For anyone interested in a rough impression, I did manage to record the output for live input this time though (on another box, through a mixing console, sorry), jackd parameters are in the filename: http://quitte.de/tmp/2i4-48k-1024x3-crackle.wav (clicks seem to align to 10ms intervals btw). jackd still reported no xruns, and the input stream shows no signs of clicking or crackling. Cheers, Tim

[James Stone] No such systems around here, so that's unlikely to happen. I might be able to try it one of these days on different linux USB host hardware, we'll see. Cheers, Tim

You might try emailing Alan Stern about this. Could be related to kernel USB handling.. James On 10 Jan 2014 16:30, "Kees van Veen" <kees.vanveen(a)gmail.com> wrote:

Jeremy Jongepier wrote: Thanks very much James, Jeremy for the pointers. I installed 3.2.0-33-lowlatency and have no crackling (nor usb_submit_urb's in the log) any more. I'll try 3.13rc6 mentioned in launchpad as well. Thanks, Kees

On Mon, Jan 13, 2014 at 8:54 AM, Tim Goetze <tim(a)quitte.de> wrote: Just as a matter of interest, does it do this with nothing attached to the input? Also, do you hear the crackles thru headphones when switched over to direct (0 latency) monitoring? James

[James Stone] Finally got around to do that and it's producing the same noises. At this point I think it must be assumed the hardware is faulty after all. :) Cheers, Tim

On Wed, Jan 8, 2014 at 8:40 AM, Jeremy Jongepier &lt;jeremy(a)autostatic.com&gt; wrote: I dunno - YMMV, but I just tried recording on the line in on the Scarlett 2i4 (from an mp3 player) and I couldn't detect any gain drift at all over 5 mins or so of re-recording the same snippet, and the noise when gain was set to just below clipping (going by the red warning LED on the device) was < -90dB (according to the spectrum analyser on audacity). Obviously this was line in rather than the mic pre. Happy to try again with some more accurate software if you point me in the right direction. Not sure I will get round to soldering any resistors onto leads for a while tho! James

On Fri, Jan 10, 2014 at 10:20 PM, Fons Adriaensen &lt;fons(a)linuxaudio.org&gt; wrote: Yep - the spectrum analyser showed a trace at around -93dB across all bands. Is this meaningful? I guess it may be higher for the mic preamps. With further testing, I think I can detect a similar change in gain effect from what Robin reported - I seem to be getting between -13.0 and -13.6 db at the lowest end of the spectrum for the snippet that is being played back. Seems a bit random where it comes though. I think they are analogue pots, so maybe this might account for this variance?? - anyway, I think it is too subtle to notice in normal use. James

On Fri, Jan 10, 2014 at 11:20 PM, Fons Adriaensen &lt;fons(a)linuxaudio.org&gt; wrote: It's reported as -81dB on line-in with jnoisemeter. The mic stage when turned right up (for my cheap SM57 copy) has background noise of -65 dB (with mic still on, and quite a bit of background noise, so not a great test), when the mic is switched off (but still connected) it is -95dB. In practice, it doesn't seem that noisy to my uneducated ears, but I guess others will have different opinions. The one thing I would say is that I am not that impressed with realtime performance overall. It is really only stable down to about 128/2. If I use Mixbus, I have to use 256/2 on jack. According to jack_iodelay this is around 27ms round trip. However, despite this, I haven't really noticed any audible delay that makes recording difficult. 128/3 seems better - around 15ms round trip according to jack_iodelay. With a realtime kernel - 3.6.4-bbq-rt-rt10, I can push jack right down to 64/2 - (approx 10ms round trip according to jack_iodelay) but this leads to xruns with minor stresses on the CPU. James

On 01/08/2014 02:41 PM, Fons Adriaensen wrote: I had calibrated -18dBFS on all of them to the same source (a 1KHz sine, which I believe to be 0 dBu). That's how I found out about the rubber controls effect in the first place. I also did two another measurements, setting the gain-dial of each devices at minimum and maximum. The 18i6 still came out worst.. but that is not too unexpected. The gain range of the 18i6 is about -10dB .. +30dB (per specs it should be -10..+36), while on the other two it's -20dB .. +20dB. I also have one of them each. So maybe I was unlucky to get a bad/good device. I also did not have access to pro-equiment to calibrate & measure them, but identical equipment was used to evaluate all. right. They don't need to be exact to the degree but at least give an indication. "About +5dB" is more useful that just a number, IMHO. The form-factor of the 18i8 is probably a bonus in this case. Skimming over http://wiki.linuxaudio.org/wiki/hardware_support the only comparable device would be the Alesis iO4 (unless you also consider more expensive equipment with more than 4 inputs). for environmental noise measurement, the devices do not need to be sample-sync, do they? You could probably use two stereo USB1 devices. best, robin

On 01/08/2014 04:43 PM, Fons Adriaensen wrote: I don't think I did. I used an unbalanced jack (out from a 4th soundcard) -> balanced XLR (pin 2,3) to calibrate. Then a XLR plug with a resistor between pin 2,3 for the noise measurement. Cheers! robin

On Wed, Jan 08, 2014 at 05:04:38PM +0100, Robin Gareus wrote: I just checked the calibration data for mu UA-5, which as far as the analog part is concerned, is similar to the UA-25 (apart from having separate line inputs which do not pass via the mic preamps). At max gain, the required input level for 0 dB digital is around -42 dBu, fairly normal for a mic input. To produce -18 dB digital with 0 dBu input you'd need 60 dB less gain. Maybe that's possible with the gain pots just above the minimum, but anything measured with that setting wouldn't be very representative. Ciao, -- FA A world of exhaustive, reliable metadata would be an utopia. It's also a pipe-dream, founded on self-delusion, nerd hubris and hysterically inflated market opportunities. (Cory Doctorow)