Can't get Asus Xonar SE detected on new hardware (e.g. Z590 / other Intel 5xx)? Try this.

[AnalogSteph]

My shiny new Asus TUF Gaming Z590-Plus WIFI (now that's a mouthful) mamaboard would not detect a Xonar SE soundcard by the very same manufacturer (oops). On top of that, the Intel I225-V onboard LAN would disappear when the card was inserted. At the same time, the card was proven to be working on a neolithic Q45 / socket 775 machine... but I didn't mean to be using it on my old system with its PS/2 keyboard/mouse power begging for retirement (an issue that popped up early this year, prompting the new build in a hurry - hence why I haven't been around here much lately).

Following two separate leads ultimately proved successful.
1. The ASM1142 USB 3.1 controller used on this card claims to support 64-bit DMA while it actually doesn't. That's clearly a bug. This particular chip was commonly used on boards from about Z97 to Z270 (2014-ish to 2018-ish or 4th gen to 8th gen Core), so not even that long ago. (Has someone ever made a combo USB 3.1 / soundcard? Seems like a missed opportunity, I mean, that chip must be extremely bored with USB 2.0 sound.)
2. For the "traditional" Xonars (i.e. the CMI8786/88 based ones) on Intel 500 series chipsets, MaxedTech recommends disabling VT-d (device virtualization).

So I ultimately disabled both VT-d and "Above 4G Decoding" (Platform Misc. settings, PCI) - et voilà, Windows 10 picks up the ASMedia USB controller and installs (generic) drivers like it's nothing. I got my LAN back, too. (I've got a bone to pick with that one as well - its driver does not seem to cooperate with the trusty Shrewsoft VPN Client or I presume rather its filter driver. Good thing I got the board version with WiFi. I225-V LAN does not have a great reputation in general, though I have what should be the fixed rev 3. A lowly Realtek card is on the way.)

On the downside, reported Uncore power increases by 3 W (~3.8 W to ~6.8 W) and mains input power increases by approx. 9 W with the card inserted. You can save about 2 W mains-side by disabling the card but I guess it does not play nice with ASPM as Uncore power remains elevated even so.

(Yes, I am one of these insane people that expect power saving features to actually work. Had idle power down right around 20-21 W mains-side, which is not too bad for a be quiet! Straight Power 11 550 W powering said board with an i7-11700, 2x 8 GB of Crucial DDR4-3200 @ 2933 with 1:2 memory controller gearing, a Samsung 970 EVO Plus 2 TB NVMe SSD plus my old 850 Pro for data transfer and an LG BH10LS38 BD writer both borrowed from the old rig, as I actually forgot to order a new optical drive aside from accidentally ordering the older 80Plus Gold version of the power supply instead of the Platinum. Probably about 1 W more on that alone. Oh, and the memory underclocking was not entirely voluntary, but I had never seen the 7-Zip benchmark spit out a "Decoding Error" before. Might still try 3200 with some manual timings another time, but 2933 is fine. I am using Throttlestop to rein in turbo clocks anyway. Needless to say, I was aiming for a system that is both less power-hungry than my old one and very, very quiet, and this one delivers.)

BTW, the onboard audio of this board seems fairly alright even if it's "just" an ALC1200 lurking under the shield. Line input noise floor at input gain 0 dB (default is +6 dB) comes out to -98ish dB(A) in Audacity (24-bit, 48 kHz, A-weighting applied as EQ). A lot of 1/f noise but generally quite clean other than that, so the advertised shielding and dedicated PCB layers must be doing something. The Xonar SE (ALC1220 inside) on an open input reads about -104 dB(A) (line) / -101 dB(A) (mic) instead. (Note: Level meter calibrated to full-scale square wave, a full-scale sine is at -3 dB, so dynamic range would be 95 / 101 / 98 dBFS(A), respectively.)

Boy, the RMAA results of the Xonar SE are eccentric. No-input noise level is -106.4 dB(A) but dynamic range at -60 dBFS comes out to 112.2 dB(A) - uh, what? (Yes, 1/f noise level is visibly lower with the -60 dBFS signal.) Crosstalk is -64.2 dB flat across the band (repeatably), a layout problem perhaps? At least frequency response seems to be dead flat, while the ALC1200 exhibits about +/-0.025 dB worth of periodic filter ripple from presumably the ADC (we have seen worse). Loopback DR and noise level for the ALC1200 (-0.9 dB out, in -1.2/1.3 dB) come out to 105 dB(A) each, while channel separation tops out at >90 dB as expected. I have yet to measure absolute levels but am expecting around 2 Vrms for the Xonar SE and and 1 Vrms for the ALC1200. Loopback distortion is clearly better for the better chip, less than 0.002% rather than almost 0.006%.

The ALC1200 performance level reminds me of the ALC898-powered SB Audigy FX (which however accepted 2 Vrms in/out). Even the quirk of dynamic range dropping in 44.1 kHz is still there, even if it's not quite as drastic at 98.7 dB(A) (the FX managed to drop from 103.7 to 92.1 dB(A), this being pretty much entirely on the A/D side). In any case I can't seem to fault Asus' implementation of the thing. They clearly did a better job with this one than with the ALC1220 on the Xonar SE.

I just ordered the Xonar SE for curiosity's sake, btw, I've been meaning to check one out for years. Right now I am using the lowly onboard audio instead of my Xonar D1. Eventually I guess I'll just get a decent little USB audio interface (something a little less toasty than the Focusrite Forte I still have floating around, which only work reliably with its external power supply). USB is the main reason why I went Intel anyway. Have had zero issues with audio hiccups so far in general, and DPC latency remains well in the green even with WiFi activity.

 

[AnalogSteph]

Update: The so-so crosstalk is definitely on the input side of the Xonar SE, connecting its output to the ALC1200 input yields exemplary results in this regard. I still don't see how you would yield meh crosstalk in a line-level application, unless the input jack ground is barely even soldered or something. I may have to consult the multimeter. The mysterious excessive 1/f noise in the noise level test also seems to be entirely on the SE input.

With this measurement I was also able to subtract frequency responses in RMAA (cross-connected vs. ALC1200 loopback), showing a smooth curve with no (visible) sign of periodic filter ripple. So that must be on the ALC1200 ADC as suspected. (ALC898 spec was +/-0.03 dB ADC, +/-0.0005 dB DAC. So the ALC1220 seems to be more along the lines of the ALC899 with +/-0.0005 dB in both directions, as befits a TOTL part. The irony: That's probably better than a number of budget DACs and audio interfaces.)

I'll definitely have to check absolute levels... the level difference between both inputs only seems to be about 3 dB, and I still have to set my mixer's output lower than for the D1, so I suspect Xonar SE max input may be about 1.4 Vrms only. At the same time, I was basically maxing out the output in loopback, and that's supposed to reach 2.0 Vrms on the ALC1220, so... *shrug* Hmm, if I switch to speaker out in the Xonar control panel (which seems to lose ~6 dB over headphone out) I'm getting at most -2.6 dB in on the ALC1200... Will have to check how much that is.
EDIT: Duh, I forgot that even the older ALC898 is like 1.45 V in / 1.1 V out. So you would expect a difference of ~3 dB between an input like that and one that can take 2 Vrms.

One oddity I found in the cross-connected setup was that one of the two devices seems to momentarily drop out when the ALC1200 input is even slightly beyond clipping. Speaking of issues, trying to measure with ASIO in and out was a bust. Apparently RMAA can't use two different ASIO drivers at the same time? ASIO4All, here I come...

OK, that works, but now that I've got the right devices selected, the weird 1/f noise in noise level testing is back. Some sort of weird auto power saving feature in the ALC1220 perhaps? Hmm, this seems to go away if I switch to headphone output mode. Guess I'll have to test in loopback like that again.

Oh great, ASIO4All doesn't pick up the Xonar input if that is switched to line-in mode. That seems to be a Xonar driver bug, actually, as neither does the native ASIO driver. Realtek build 9087 is only offering me a mic in in line-in mode, too. Jack retasking ain't easy, it seems.

So then I used plain ol' MME again and got this:

The noise level discrepancy still is there but substantially reduced.

Note: You must by all means keep an eye on Xonar Audio Center and make sure that the right inputs and outputs are selected. Due to jack retasking, rear front L&R can be used in both headphone and speaker mode, and likewise the input can be operated as both a mic-in and line-in. Also, switching between them seems to confuse the living hell out of RMAA, requiring restarting that.

Here I dropped output levels by 20 dB, switched to mic in and increased gain as required for near identical input level:

The "evil" THD is basically all H2 at -87ish dB, 3rd is at -110 dB and that's about it.

And here's HP out --> line-in again at -1.6, -3.1, -6.2 and -9.9 dBFS:

IMD (swept) increases quite substantially towards the higher frequencies, though a large part seems to be H2. This is not a Cosmos ADC we are talking about here after all.

So is the Xonar SE a worthy Xonar D1/DX replacement? Kinda. Input distortion levels are not a match for the older cards, but the low-Z, wide dynamic range, DC-coupled headphone output will be more versatile in practical use. I'm pretty sure that the ADC section is limiting dynamic range first but may break out my Atom Amp+ to verify.
EDIT: And there you go:

Gain delta between high and low on the Amp+ is 12.8/12.9 dB, so since low is supposedly unity (yup: insertion loss approx. 0.1-0.2 dB, close enough), we can conclude that output dynamic range would have to be at least 121.5 dB(A). (If I use this value to estimate ADC input noise floor more accurately and use the result of that again to estimate Amp+ output noise level, I'm even getting something close to 123.5 dB(A).)

I can't help but think that there may be some sort of trickery going on with this chip (or just plain ol' jitter). Noise level at different signal levels seems to be quite variable, dropping about 10 dB between THD (-3 dB) and DR (-60 dB) tests. I won't be too surprised if it turns out that analog gain is being used to shift the dynamic range of more modest, less power-hungry converters on the fly.

Atom Amp+ test, 192k:

That's a bunch of noise shaping, not particularly high order though (2nd-3rd tops). It's still there if I turn down the Atom Amp+, so must be all ADC.

Jeez, RMAA is so flaky with this card, it just terminates left and right. It also seems you need to go into output device properties and tick "Disable all effects" to get hi-res out. Not sure whether you can tell Xonar Audio Center not to revert that almost immediately. Shared mode beyond 48 kHz with this card could be annoying.

Hey, once I disable all the effects I can get my 108 dB of dynamic range in loopback even in 44.1 kHz! Or wait, I don't think I tested this on the Xonar SE yet. So the ALC1220 ADC may be way more impervious to jitter than the ALC1200 job. Good to know! (Well, either that, or the CM6620 generates substantially cleaner clocks. I may have to turn off spread spectrum and try the onboard audio again.)

Update: Did the same Amp+ test with the ALC1200 (input level -1.4 dB), got 96.9 dB(A) of dynamic range on the channel with less ground loop noise. So amp output noise comes out as -97.6 dBFS(A) (when RMS-unsumming the -105 dB found in loopback previously as a first estimate for ADC noise), for an output DR estimate of 110.4 dB(A). That's pretty much in line with chip specs. RMS-unsumming -110.4 dB from -105 dB thus yields a 106.5 dB(A) input dynamic range estimate.
Iterating again with this value gives -97.4 dBFS(A) post-amp or 110.2 dB(A) out, and an input DR estimate of 106.6 dB(A). That's rather better than the 102 dB chip spec, much like what we found on the Xonar SE (ALC1220 spec 120 dB out, 110 dB in, where it looks like more like 123 dB(A) out and 113 dB(A) in).
(I'll have to think about why this iteration works the way it does, and possibly even make a calculator. We are obviously assuming an ideal noiseless amplifier, but I think the Amp+ should be a pretty decent approximation at this performance level. -110.2 dB(A) down from 1.1 V is 3.4 µV, and I'd assume Amp+ EIN at high gain to be sub-µV level.)

 

[AnalogSteph]

Update:
Onboard ALC1200 output measurements with 100 Hz, amplitude 0.707 (-3 dB) sine -
Rear L/R out: 1.024 / 1.043 Vrms
Front headphone out: 1.268 / 1.283 Vrms 0.939 / 0.928 Vrms (see update)

So 0 dBFS level comes out to approx:
Rear L/R: 1.45 Vrms (and you can slightly overdrive the line-in with this, so 1.4ish on that?)
Front headphone out: 1.80 Vrms 1.32 Vrms (see update)
(and yes, I have verified that both are free from obvious distortion up to -0.1 dBFS, at which point the Windows audio limiter would kick in - rear out for amplitude 0.98855 is 1.430/1.455 Vrms, front headphone out is 1.31 / 1.295 Vrms.)

That's not bad at all. (Xonar SE calculated 0 dBFS input level then comes out to 1.95 V, pretty much what it should be.) I did think the headphone out had a fair bit of oomph when connecting my HD580s, though I doubt output impedance is any lower than the typical 75 ohms. It's dead quiet with these cans at least. Plugging in my IEMs (KZ ZSX) reveals audible, but not unacceptable hiss, and they get rather more thin and nasally compared to the Atom Amp+ as you would expect from the impedance response. The fact that there seems to be some effect enabled by default that does something along the same lines won't help.

Rear L/R THD at 100% output seems to be 0.004-0.0045% according to RMAA, dominant H2 (~88 dB below fundamental). At least it seems to be benign and flat with no surprises across the audio spectrum. Hardly chart-topping but almost certainly completely inaudible.

Some more loopback measurements shown here.

I pulled out the Xonar SE again and wall power still is up from 20-21 W idle to 25-26 W with just a Realtek RTL8125B NIC in (it would have been 28-29ish with the SE). That kinda sucks. Looks like for max power efficiency, you almost have to go USB.

Update: The "dts Custom" effects are enabled by default for headphone out. Go to Realtek Audio Console and click the "dts Custom" logo to open a window where you can turn them off. This meant I promptly had to redo the output measurements as well.

Oh great, the onboard Realtek has the same exact quirk of the Xonar SE - you can't get more than 48 kHz out in shared mode unless you tick "Disable all effects".

The ALC1200 input shows a bit of ultrasonic noise going on, looks like 4th order noise shaping, still lower order than a CS5361/CS4272 or similar.

My current estimates for dynamic range relative to respective 0 dBFS levels:
rear speaker out 111.8 dB(A)
rear line in 107.8 dB(A)

(Same for Xonar SE:
rear headphone out 123.5 dB(A)
rear line in 115.7 dB(A))

 

[bennetng]

@AnalogSteph
Attached some RMAA tests on my new but low end Alder Lake board with ALC897. All of them are loopback results. No more excessive passband ripple on the ADC side in any sample rate. No more increased 44.1k ADC noise floor but 44.1k has some high frequency roll off, haven't check if it's on the ADC or DAC part, or both. My Windows 10 is on MBR partition and it doesn't work with the intel GPU, with a GPU card installed there is some noise when linking the X-Fi card and ALC987 together, so not very fair to post the ALC897 results in this way at this moment.

My 9 years old X-Fi PCIe card works fine as well. The motherboard is the cheapest one I can find with slot 1 and 4 being PCIe so that I can have both GPU card and soundcard installed without putting them too closely.

The main issues is USB3, as mentioned here:

USB compatibility issue

I submitted a request at https://event.asrock.com/tsd.asp a week ago, only got an automated notification...

forum.asrock.com

 

[AnalogSteph]

97 dB(A) in loopback is some of the best results I've seen for this chip. Zero degradation in 44.1 kHz either, which is unusual... as is the rolloff up top that you already noticed. Overall very acceptable results for lowly onboard audio though.

You can try using the mic-in with say 20 dB worth of analog input gain for a better estimate of output dynamic range.

Have you noticed a default input volume setting of +6 dB on this chip as well? It was like that with the Asus ALC1200 drivers at least.

No more excessive passband ripple on the ADC side in any sample rate.

Unless, of course, the filters are designed to be complementary, though I don't think this is that easy.

My Windows 10 is on MBR partition and it doesn't work with the intel GPU,

Yeah, I noticed during testing that my UHD750 graphics don't seem to have a conventional video BIOS either, just UEFI. It wasn't a major issue for me as I was setting up the system from scratch anyway, but it is a bit of a nuisance. I can still boot UEFI with a BIOS-only GT610 though.

with a GPU card installed there is some noise when linking the X-Fi card and ALC987 together, so not very fair to post the ALC897 results in this way at this moment.

You will always have ground loop issues in such a scenario, it's quite unavoidable. It's still fine for checking frequency response.

In the past I have used my Behringer HD400 for galvanic isolation, that should get you some decent dynamic range readings at least. Otherwise you need a floating source like a DAP or laptop.

My 9 years old X-Fi PCIe card works fine as well.

Interesting, which model exactly?

I have been contemplating giving an Audigy FX v2 a shot after learning that it exists. It's basically a reboot of the old FX (with ALC898), now with an ALC1220.

I wonder whether it might be possible to hack a Xonar SE for use with an external USB2 header, ideally without creating any insidious ground loops...

The motherboard is the cheapest one I can find with slot 1 and 4 being PCIe so that I can have both GPU card and soundcard installed without putting them too closely.

If proximity is an issue, the soundcard's board layout is no good and sports internal ground loops. The enormous currents circulating on a GPU will be exposing those quite ruthlessly. Unfortunately even external loopback may prove too much in such a situation.

The main issues is USB3, as mentioned here:

That's not good. If that's an isolated problem, I would suspect that the chipset may have a soldering issue, losing contact on some balls when the USB portion heats up. That would require reflow or maybe even reballing, and if that's beyond your abilities, returning the board and getting a replacement. Unfortunate but no cheap yet complex consumer-level product ever has a 0% failure rate. If it turns out to affect all boards of this type, someone may have been too skimpy with decoupling caps.

 

[bennetng]

Have you noticed a default input volume setting of +6 dB on this chip as well? It was like that with the Asus ALC1200 drivers at least.

I am only using the Microsoft "High Definition Audio Device" because I don't really use the Realtek regularly and therefore don't want to install the Realtek driver. With everything up and running I blocked automatic update and such. Couldn't remember the default gain value as I often make adjustments anyway. But yes, I tried both line in and mic in and optimized the input and output gain settings, so what you see are the best results I can get in a loopback.

Interesting, which model exactly?

X-Fi Titanium HD

If proximity is an issue, the soundcard's board layout is no good and sports internal ground loops. The enormous currents circulating on a GPU will be exposing those quite ruthlessly. Unfortunately even external loopback may prove too much in such a situation.

My concern is ventilation as all standard GPU cards occupy 2 slots, also pretty dangerous as the GPU fan blades are only millimeters away from the back of the soundcard if the usable PCIe slots are 1 and 3 or 2 and 4. I am using an mATX case and mATX board so 1 and 4 are the best distance I can get.

That's not good. If that's an isolated problem, I would suspect that the chipset may have a soldering issue, losing contact on some balls when the USB portion heats up. That would require reflow or maybe even reballing, and if that's beyond your abilities, returning the board and getting a replacement. Unfortunate but no cheap yet complex consumer-level product ever has a 0% failure rate. If it turns out to affect all boards of this type, someone may have been too skimpy with decoupling caps.

Many years ago I had a Gigabyte LGA775 board, and not a particularly low end model, failed two times within warranty period, and long before that I had an expensive SCSI Plextor CD writer died briefly after warranty. Also a rather expensive Roland audio interface purchased in 2001 with poorer line-in measurement results than a Soundblaster live purchased in 1999. So I gave up the idea of buying expensive stuff long time ago.

 

[AnalogSteph]

But yes, I tried both line in and mic in and optimized the input and output gain settings, so what you see are the best results I can get in a loopback.

I would still suggest doing Mic in with +20 dB of input preamp... you can pretty much add 20 dB to the dynamic range value this gets you, for a better estimate of output dynamic range. (Note, determine actual gain steps while setting levels. I found it was more like +18.something dB at "20 dB" here.) In a chip of relatively modest converter dynamic range like this, analog input noise should be a fair bit lower than converter noise.

I am using an mATX case and mATX board so 1 and 4 are the best distance I can get.

Understood. I went with a mid-sized case (Fractal Define R5) and a fullsize ATX board... my old computer's BTX board wasn't exactly empty either.

Many years ago I had a Gigabyte LGA775 board, and not a particularly low end model, failed two times within warranty period, and long before that I had an expensive SCSI Plextor CD writer died briefly after warranty. Also a rather expensive Roland audio interface purchased in 2001 with poorer line-in measurement results than a Soundblaster live purchased in 1999. So I gave up the idea of buying expensive stuff long time ago.

I kind of get where you're coming from... you can get unlucky even with expensive products, and their construction quality isn't necessarily any better. (RME and jumpy encoders, anyone?) On YT I recently saw a post from someone who had been battling several issues with his MSI Z490 board since 2020, including stability problems and deaf WiFi. In the US that might have been a case for lemon law, though that's mostly a thing in the rather more expensive automotive world.

Come to think of it, my new case arrived with a slight blemish on the front and a non-functional power LED, though I don't much care about either. (I've seen modding instructions for the power button assembly, so if in doubt I might just tear the thing apart and put in another LED with color of my choosing, not a big fan of blue anyway.) I've had no complaints with the actual board (though you're pretty much out on your own if you insist on using OEM-grade RAM at stock 1.2 V, my Crucial sticks required slight underclocking at 2933 CL21). I wasn't exactly thrilled with the connectors on my be quiet! Straight Power modular 550 W PSU but other than that that's been fine, too.

The problem is, a sample size of one doesn't tell you an awful lot. That Gigabyte board may well have had some type of actual construction issues (any indication of bad caps by any chance?), but with the Plextor you may have just gotten unlucky.

Now as for that old Roland interface, for years it was quite common for input and output reference levels to be substantially mismatched, so loopback dynamic range could be quite meh, and if the line-in was designed to handle +20 to +24 dBu, results at consumer line level (+8 dBu) may not be all that great either. Back in 2001, mid-grade ADCs were rated at 100, 102, 104, at most 105 dB(A) of dynamic range, and we all know that actual complete interfaces often fall somewhat short of converter specs. If you're at 97 dB(A) and losing 12 dB due to level mismatch, suddently that's just 85 dB(A). That's well within reach of consumer hardware of the day - a lowly Terratec DMX XFire 1024 card using a CS4294 (2000-ish?) would post 87 dB(A) in loopback, for example. (These Crystal things were some of the few really decent AC97 codecs.) The lowly Sigmatel codec on the old Live! should still have had plenty of filter passband ripple and distortion though, that's just how things mostly were in those days.

The 2000s were a wild time for PC sound equipment. The same money that would get you an AC97-based job like said DMX XFire 1024 at the start of the decade would get you an Asus Xonar D1 (~111 dB(A) in loopback, CS4398 + CS5361) at the end. You don't see an extra 24 dB of dynamic range every day. Mind you, another decade earlier you might have had a SB Pro with glorious 8-bit audio (22.05 kHz stereo or 44.1 kHz mono), so...

 

[bennetng]

Roland results on iXBT:

Тест программы RightMark Audio Analyzer: Roland SC-D70 Wave

Not very bad with the results above but mine with the same test condition (loopback, to rule out ground loop issues) was worse as THD was approaching 0.1%, and actually audible in ABX test when listening to a tone with appropriate frequencies. With a battery powered separate source (a feature rich Discman with headphone out, line out and optical out) I found it was on the ADC. Also another RMAA test (no longer online) on a Japanese site showed similar results as mine as well, so probably some QC issues or even different components/revisions. Another interesting thing is by doing these tests, I found that my Discman has obviously poorer results with skip protection enabled, likely because it used a lossy format to store the buffer samples.

IIRC bit depth was not a limiting factor on the Roland, especially considering the original SB Live only supports 16-bit. The Roland is also using unbalanced RCA with lower than 2Vrms. The Live's Sigmatel ADC is noisier but with flatter frequency response and lower distortion, at least on 48kHz. 44.1k performance on the Live is of course, a mess, which was common sense among DIYers at that time. I also had an AWE32 variant and bought 8MB SIMMs to play with SoundFont, that card was indeed quite noisy when compared to typical Discmans and MD Walkmans at that time.

 

[bennetng]

OK, tested the ALC897 line-in using X-Fi's output, with the GPU card installed. Also loopback results of the X-Fi to have a reference frequency response. Looks like the 44.1k droop is on the ADC. No change in ripple so far. The X-Fi uses TI PCM1794A and PCM4220.

X-Fi's loopback results without GPU card are here:

https://www.audiosciencereview.com/forum/index.php?attachments/x-fi-titanium-hd-zip.29512/

 

[HorBromin]

My shiny new Asus TUF Gaming Z590-Plus WIFI (now that's a mouthful) mamaboard would not detect a Xonar SE soundcard by the very same manufacturer (oops). On top of that, the Intel I225-V onboard LAN would disappear when the card was inserted. At the same time, the card was proven to be working on a neolithic Q45 / socket 775 machine... but I didn't mean to be using it on my old system with its PS/2 keyboard/mouse power begging for retirement (an issue that popped up early this year, prompting the new build in a hurry - hence why I haven't been around here much lately).

Following two separate leads ultimately proved successful.
1. The ASM1142 USB 3.1 controller used on this card claims to support 64-bit DMA while it actually doesn't. That's clearly a bug. This particular chip was commonly used on boards from about Z97 to Z270 (2014-ish to 2018-ish or 4th gen to 8th gen Core), so not even that long ago. (Has someone ever made a combo USB 3.1 / soundcard? Seems like a missed opportunity, I mean, that chip must be extremely bored with USB 2.0 sound.)
2. For the "traditional" Xonars (i.e. the CMI8786/88 based ones) on Intel 500 series chipsets, MaxedTech recommends disabling VT-d (device virtualization).

So I ultimately disabled both VT-d and "Above 4G Decoding" (Platform Misc. settings, PCI) - et voilà, Windows 10 picks up the ASMedia USB controller and installs (generic) drivers like it's nothing. I got my LAN back, too. (I've got a bone to pick with that one as well - its driver does not seem to cooperate with the trusty Shrewsoft VPN Client or I presume rather its filter driver. Good thing I got the board version with WiFi. I225-V LAN does not have a great reputation in general, though I have what should be the fixed rev 3. A lowly Realtek card is on the way.)

On the downside, reported Uncore power increases by 3 W (~3.8 W to ~6.8 W) and mains input power increases by approx. 9 W with the card inserted. You can save about 2 W mains-side by disabling the card but I guess it does not play nice with ASPM as Uncore power remains elevated even so.

(Yes, I am one of these insane people that expect power saving features to actually work. Had idle power down right around 20-21 W mains-side, which is not too bad for a be quiet! Straight Power 11 550 W powering said board with an i7-11700, 2x 8 GB of Crucial DDR4-3200 @ 2933 with 1:2 memory controller gearing, a Samsung 970 EVO Plus 2 TB NVMe SSD plus my old 850 Pro for data transfer and an LG BH10LS38 BD writer both borrowed from the old rig, as I actually forgot to order a new optical drive aside from accidentally ordering the older 80Plus Gold version of the power supply instead of the Platinum. Probably about 1 W more on that alone. Oh, and the memory underclocking was not entirely voluntary, but I had never seen the 7-Zip benchmark spit out a "Decoding Error" before. Might still try 3200 with some manual timings another time, but 2933 is fine. I am using Throttlestop to rein in turbo clocks anyway. Needless to say, I was aiming for a system that is both less power-hungry than my old one and very, very quiet, and this one delivers.)

BTW, the onboard audio of this board seems fairly alright even if it's "just" an ALC1200 lurking under the shield. Line input noise floor at input gain 0 dB (default is +6 dB) comes out to -98ish dB(A) in Audacity (24-bit, 48 kHz, A-weighting applied as EQ). A lot of 1/f noise but generally quite clean other than that, so the advertised shielding and dedicated PCB layers must be doing something. The Xonar SE (ALC1220 inside) on an open input reads about -104 dB(A) (line) / -101 dB(A) (mic) instead. (Note: Level meter calibrated to full-scale square wave, a full-scale sine is at -3 dB, so dynamic range would be 95 / 101 / 98 dBFS(A), respectively.)

Boy, the RMAA results of the Xonar SE are eccentric. No-input noise level is -106.4 dB(A) but dynamic range at -60 dBFS comes out to 112.2 dB(A) - uh, what? (Yes, 1/f noise level is visibly lower with the -60 dBFS signal.) Crosstalk is -64.2 dB flat across the band (repeatably), a layout problem perhaps? At least frequency response seems to be dead flat, while the ALC1200 exhibits about +/-0.025 dB worth of periodic filter ripple from presumably the ADC (we have seen worse). Loopback DR and noise level for the ALC1200 (-0.9 dB out, in -1.2/1.3 dB) come out to 105 dB(A) each, while channel separation tops out at >90 dB as expected. I have yet to measure absolute levels but am expecting around 2 Vrms for the Xonar SE and and 1 Vrms for the ALC1200. Loopback distortion is clearly better for the better chip, less than 0.002% rather than almost 0.006%.

The ALC1200 performance level reminds me of the ALC898-powered SB Audigy FX (which however accepted 2 Vrms in/out). Even the quirk of dynamic range dropping in 44.1 kHz is still there, even if it's not quite as drastic at 98.7 dB(A) (the FX managed to drop from 103.7 to 92.1 dB(A), this being pretty much entirely on the A/D side). In any case I can't seem to fault Asus' implementation of the thing. They clearly did a better job with this one than with the ALC1220 on the Xonar SE.

I just ordered the Xonar SE for curiosity's sake, btw, I've been meaning to check one out for years. Right now I am using the lowly onboard audio instead of my Xonar D1. Eventually I guess I'll just get a decent little USB audio interface (something a little less toasty than the Focusrite Forte I still have floating around, which only work reliably with its external power supply). USB is the main reason why I went Intel anyway. Have had zero issues with audio hiccups so far in general, and DPC latency remains well in the green even with WiFi activity.

Hi!
I have the same problem, I have the same motherboard and sound card, and Windows does not immediately detect it.In general, what happens is that If the PC is turned on cold (it was de-energized for several hours), then after booting the system, Windows does not see the card. However, if you wait about 30 minutes, the card starts to be detected in the system. But recently, the sound card began to be detected after only a couple of hours, which is no longer normal, now it has stopped being detected at all. I tried to set the settings you specified in the BIOS, but it didn't help. Maybe you have suggestions, what else can try

 

[AnalogSteph]

Sounds like PCIe ASPM was left on for the card for some reason; normally the BIOS automatically turns it off. Which BIOS version are you on? Have you tried reseating the card? If push comes to shove, you may have to try it in some other system (the older the better, ideally some Core 2 Duo or thereabouts) - maybe the card is actually defective (bad electrolytic in the digital section). If it comes back to life when warmed up with a hairdryer, that would be very sus.