Any recommendations for a motherboard with good audio performance at 192KHz?

[Modd Audio]

I tested a rog strix b550 gaming, which ran at 192KHz but it had a very bad clock jitter issue.   Sorry I can not link to my results since it is in youtube vide format. Providing a link to my free to use test software is also against policy here as the source is shared with other projects and not open source.   

I still do need a MB with clean audio, preferably one that uses zero phase halfband filters.

 

[Psittac]

Why motherboard audio and not external?

[8tg]

17 minutes ago, Psittac said:

Why motherboard audio and not external?

Basically this

 

Even some pretty basic dongle dac kinda deals will do you better than any motherboard audio.

The board itself is simply noisy and dependent on too much else going on around the audio hardware. I’d recommend an external solution or if you must go internal, a solid sound card. I use an HT omega eclaro in a secondary system which is great for that 192khz/24bit stuff. Though on my main pc I run an external Fiio KA3 which is inexpensive and supports basically anything up to DSD512

[ImWilly]

For that needs, if you only wants to have 'cleaner audio' and 'must be plugged directly from motherboard audio jack', the nearest possibly to upgrade B550 to X570 and look for at least "ALC1220" or "ALC4080"

Usually the 'best' know motherboard for 'high capacity audio' is MSI PRO X670-P

 

But, if you want a better than stock motherboard audio jack plug, I would recommend to get inexpensive external audio interface, such as Focusrite or  Steinberg, or PCIe soundcard on your mobo. Because no matter how good the capacitors or shielding on high grade audio capacitors, motherboard produces a really noisy signals especially in current PCB design, there are significant lots of electrical nodes (even harder to be seen by naked eyes) compared to two decades ago

A separated audio interface / PCIe soundcard literally puts a distance of DAC (Digital to Audio) process from noisy motherboard 

 

Additional tech tips, adding UPS or iFi GND loop breaker may help achieving cleaner sound.

[Psittac]

4 hours ago, Modd Audio said:

I tested a rog strix b550 gaming, which ran at 192KHz but it had a very bad clock jitter issue.   Sorry I can not link to my results since it is in youtube vide format. Providing a link to my free to use test software is also against policy here as the source is shared with other projects and not open source.   

I still do need a MB with clean audio, preferably one that uses zero phase halfband filters.

 

I think the OP is more focused on the science of it all than the actual results.  I think I get where he's coming from, you pick a topic and want to know about it inside and out.  Back 20 years ago no one knew the Athlon XP-M system better than I did, multiple motherboards and two cpu's, I got everything out of them that I could and that was even on water.  There were faster or better systems available at the time but I took my budget memory and cpu and tuned every number on the memory and knew everything that system could do.  I could tune for FSB and low ghz or vice versa.  I ran benchmarks literally all day and night.  I actually witnessed first hand silicone degradation. I wish I could find my old forum posts because I outdid everyone on that platform.  Even found the top contender and posted three screen shot's.  One with stock setting's and a stock windows background, one matching his result's and an angel embracing a demon, then one beating the pants off him with a hot bikini babe for a screen shot.

 

My point being is he is looking to be a min-maxer.

 

Correct me if I'm wrong.

 

Oh I never had the holy grail of motherboards, DFI lan-party.  I still wonder what I could have pulled out of those Mushkin second tier samsung die's with that board.  I only had the DFI Infinity.  The Abit AN-7 yielded the highest clock speeds while the Asus something or other (A7N8X-E Deluxe had to google it) was the most reliable.  I didn't have a lot of money at the time but did have a leaking cpu cooler that would drip on my 3rd tier gpu and blow something in the mb, so newegg unknowingly gave me MANY motherboards before finally cutting me off.  Today I wonder if Paul handled my RMA's and I wonder if I went toe to toe with Linus (he had the same platform around the same time).  Computers were so slow back then with a single core, that you could literally SEE improvements in typical use, not just benchmarks.

[Spuriae]

I remember that video. What is the physical mechanism for clock jitter manifesting as high frequency noise in a noise floor test? For instance, this motherboard review shows a similar high frequency noise bump but also decent jitter performance; noise shaping, IMD, aliasing, poor grounding, and plain old interference all seem like more likely culprits.

 

All Realtek motherboard outputs should use basically the same filters, since the digital filter is baked-in to the Realtek codec.

 

That aside, the most audible issue with motherboards audio (at least for headphone users) in my experience is the high output impedance, and that's a nearly universal issue no matter which motherboard you pick.

[VoidX]

2 hours ago, Spuriae said:

What is the physical mechanism for clock jitter manifesting as high frequency noise in a noise floor test?

Think of jitter as a wavelength and convert it to frequency. You want to use 192 kHz to prevent this, but this is usually already done on the DAC level. Most of them have native frequencies of at least 96 kHz, doesn't matter where you set them on the driver level. This is used to spread the noise over a larger bandwidth (white noise affects each band equally, across all present bands) so the top of it can easily be filtered out.

[Modd Audio]

I want to use MB audio to reduce latency.   I normally write stuff for embedded systems but I ported everything to a PC, now I am trying to tune a pc to be a bit more realtime.  I can get a 5mS latency using usb audio but I am trying to do a bit better.  Jitter becomes an issue because it can destabilize certain filters.  

Thanks for the link to the Gigabyte Z390 Aorus Motherboard review, it uses the same ALC1220 codec and shows the same issue.  I was unsure if it was just my board. They did not diagnose it as jitter, but to be fair, I had to ask a phd at a fruit based company for his opinion on this sort of plot.  

I think I am going to try a sound blasterX AE pcie card for this project.

 

Regarding: What is the physical mechanism for clock jitter manifesting as high frequency noise in a noise floor test?
It is my understanding that jitter is a way to shape noise.  Thermal (white) noise in this case. Those tiny difference in timing push more energy into the highs because they are in a sense above nyquist so they alias.  You also have to consider how a codec works, a CIC filter into a few half-bands.  It runs at much higher rates at lower bit depths on the front end.   Then to complicate things further you have to consider that moving from the s-domain (resistors/caps) to the z-domain (discrete time samples) is not linear and causes its own issues with high frequencies.  

[VoidX]

51 minutes ago, Modd Audio said:

I want to use MB audio to reduce latency.

MB audio is typically using a PCI lane, so latency is the same as with any internal sound card. If you go for latency, use ASIO. You can have an interrupt every 16 samples, that's the lowest, if I remember correctly. That's still 0.33 ms at 48 kHz. It requires vendor support, which you normally don't get for MB audio.

[SimplyChunk]

3 hours ago, VoidX said:

MB audio is typically using a PCI lane, so latency is the same as with any internal sound card. If you go for latency, use ASIO. You can have an interrupt every 16 samples, that's the lowest, if I remember correctly. That's still 0.33 ms at 48 kHz. It requires vendor support, which you normally don't get for MB audio.

If you look hard enough you can find ASIO drivers for Realtek onboard.  it's just usually not added on basic install.  You have to dig in the driver package and find it and then install it manually