Not available as new stock, but pretty available as possibly NOS and more likely good used.
I think the more significant issue is the limited driver support. I use mine under WinXP and 7.1 Pro. Didn't fare so well with 8.1 Pro.
In retrospect the eMu USB cards of that era were tour-de-force efforts that still deliver good useful results if you recreate their required operational environments.
The object of the question is not clear, (no quoting) but if you are talking about the 0404 USB, yes it runs 24/192 and 24/96 full duplex.
Its a general rule that any device that runs 24/192 full duplex runs 24/96 duplex.
A card that did otherwise would be a like a car that runs 200 mph, but not 100 mph.
By full duplex I mean you can open two instances of Audition (for instance) and play a waveform in one and record in the other with no drop outs or glitches. When I wrote my first Linear Audio article I was using an Edirol UA-5 (>15yr. old) and it did this flawlessly at 24/48 under XP (USB 1.0). More than 1/2 the USB cards I have tried recently break in any number of ways, garbage or nothing in the record side, refusal to make the second connection, or freezing the app.
Also Arny I think in some cases we are talking about two different things, by performance I mean the 24bits not the THD or noise of the A/D. Failure in some cases is simply there are zeros in the last 8 bits without any warning.
I just ran this test, either Audition or CoolEdit works here. I recorded the noise in my office with the built-in sound card set at 24/48 (no USB involved) using Audacity set at 48K/32bit, it's pretty common to promote 24bit integers to floats for computations.
The picture is blown up to show a few individual samples at 16bit resolution, there are no samples that are not at exact 16bit values, note the data is saved at 32bit resolution (at least the app thinks so). The vertical scale is labeled in 16bit LSB's.
Attachments
How did you confirm that? Have you done a loop back frequency response to confirm this? Rightmark test? The highest I could find was 32bit/96KHz.
Last edited:
I've been acquiring and testing audio interfaces as a hobby for about 20 years. I once published my results on a web site called pcavtech.com.
Over the years I've used a lot of different software for testing: Spectra Lab, Cool Edit Pro, Audacity, Artia, VA, REW, Rightmark, etc.
My backup "if everything else fails" turns out to be the "Two instances of CEP" test that has been mentioned. Play 1 KHz or whatever in one instance and try to record in another.
I often have problems with dropped bits with CEP under win 7.1 pro and often use Audacity to circumvent that.
It was created by the conference software behind my back.
But, since you mentioned a link, a working one can be contrived by hand:
https://web.archive.org/web/20070303180526/http://www.pcavtech.com/
Yes, they aren't, Audacity gets the upper 16 bits and adds an LSB of 24bit dither. Here's some values normalized to 1V (1 16bit LSB = 3.0517578e-05 V), you will notice these are all divisible by 1 16 bit LSB with an occasional 24bit LSB remainder. The file will still essentially act like 16bits truncated unless dithered somewhere at 16bits, there is no way of assuring this is done.
BTW a good 1/3 to 1/2 of the folks trying to do this on various fora do it ears only and, surprise, can't tell.
Format time left right
; Sample Rate 96000
; Channels 2
0 -0.00054931641 0.00073242188
1.0416667e-005 -0.00061035156 0.00082397461
2.0833333e-005 -0.00067138672 0.00085449219
3.125e-005 -0.00061035156 0.00094604492
4.1666667e-005 -0.00045776367 0.00079345703
5.2083333e-005 -0.00024414063 0.00088500977
6.25e-005 -0.00030517578 0.00067138672
7.2916667e-005 -0.00012207031 0.0007019043
8.3333333e-005 -0.00039672852 0.00067138672
9.375e-005 -0.00021362305 0.00082397461
0.00010416667 -0.00042724609 0.00076293945
0.00011458333 -0.00024414063 0.00091552734
0.000125 -0.00054931641 0.00079345703
0.00013541667 -0.00039672852 0.00091552734
0.00014583333 -0.00057983398 0.00085449219
0.00015625 -0.00030517578 0.00079345703
0.00016666667 -0.00039672852 0.0007019043
0.00017708333 -0.00024414063 0.00054931641
0.0001875 -0.00030517578 0.0007019043
0.00019791667 -0.00036621094 0.00064086914
0.00020833333 -0.00036621094 0.00076293945
Last edited:
I used Audacity to make a short recording at (supposedly) 24 bits using the Scarlett 2i2. His data showed that it was 16 plus zero fill.
I want to repeat this with Goldwave, but made the mistake of uploading the newest version. As usual, I can now no longer find functions that I used to be able to. I managed to find the settings to get to 24 bit, but the sample rate settings for record are now beyond my grasp, and all I can get is 44k1 instead of 24/96. I am old, old, old.
edit: NM, I think I found it. Sheesh.
I want to repeat this with Goldwave, but made the mistake of uploading the newest version. As usual, I can now no longer find functions that I used to be able to.
I managed to find the settings to get to 24 bit, but the sample rate settings for record are now beyond my grasp, and all I can get is 44k1 instead of 24/96. I am old, old, old.edit: NM, I think I found it. Sheesh.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.