Incredibly bad ALC1220-VB Headphone out Noise and Distortion

[vrships]

I just learned the headphone source I used most for a year is so noisy and distorted. It's hard to believe based on reviews of similar motherboards and or even previous and more inferior on-board DAC/amps.

This is the front headphone-out from a realtek ALC1220-VB from Gigabyte X570s aorus master, which actually directs the signal to SABRE ES9118 to process and has a dedicated headphone amp for front port. Meaning it has everything to outperform an ALC1220 base version. Gigabyte boasts quite much about it on the webpage:
https://www.gigabyte.com/Motherboard/X570S-AORUS-MASTER-rev-10#kf

I just ran a ton of RMAA loop tests and was astouned by the bad numbers. I did quite a lot RMAA about 10 years ago when I start touching audiophile stuff so I guess I still remember how it should be done. The front headphone port looped to back line-in (or mic), regardless of numerous different configurations I tried, only has SNR and THD+N around 70db. (bit depth and sample rate doesn't help. Doing 32bit, 192kHz usually is worse than 16bit, 48kHz)

ALC1220 VB + ES9118, headphone-out to line-in
Frequency response (from 40 Hz to 15 kHz), dB	+0.01, -0.01
Noise level, dB (A)	-73.4
Dynamic range, dB (A)	72.7
THD + Noise, dB (A)	-66.7
IIMD + Noise, %	0.052
THD, %	0.00732
Stereo crosstalk, dB	63.0

Just to see how bad this is, ALC1220 (older version than VB) setup on Asus and AsRock and even older Gigabyte X390 could easily do 95db+ to 100db:

ASRock X399M Taichi Review

ASRock showcases their engineering know-how once again, this time by taking AMD's Threadripper CPUs and stuffing it all into a mATX form factor. 32 threads and mATX, along with a Taichi treatment? Yes, PLEASE!

www.techpowerup.com

https://www.reddit.com/r/hardware/comments/b8p9l8

RMAA measurements of Gigabyte Aorus Pro Wifi Z390 (Realtek ALC1220)

Hi all, I just built a new PC with a Gigabyte motherboard of the model Aorus Pro wifi z390. I have been debating whether it would be worth it getting an external DAC so I tried measuring it with Rightmark Audio Analyzer (RMAA). For those who don't know this program, it is basically an audio...

www.audiosciencereview.com

So i thought maybe these results are done at the line-out or speaker-out ports, so they do not suffer the additional distortion from the headphone amp. So I tried mine with speaker out and line in next to each other at the rear panel: (still bit depth and sample rate doesn't really matter)

ALC1220 VB + ES9118, speaker out to line-in
Frequency response (40Hz to 15kHz)	+0.04, -0.10
Noise Level, dB	-90.1
Dynamic range, dB(A)	90.3
THD, %	0.00273
THD + Noise, dB(A)	-82.2
IMD + Noise, %	0.012
Stereo crosstalk, dB	-85.7

Much better than headphone out, but still far worse than other ALC1220.

One reason I can think of is that the front jack is impacted by the groud loops in the chassis because it's away from the rear panel. Is that supposed to be this bad? And my results are quite repeatable despite different load in the PC. The only huge difference is when GPU is heavily loaded (~350W), the SNR and everything will drop by another 50dB!（And strangely, I couldn't hear a bit of this -20dBFS noise in the headphone) But If I load the CPU up to 200W, it doesn't hurt the front jack numbers.

I have a standalone USB DAC/amp just to confirm the method is not systmetically wrong. Connecting the headphone out from USB/DAC to the same ALC1220 line-in, I get:

Centrance DACmini CX, headphone out to ALC1220VB line-in
Frequency response (40Hz to 15kHz)	-0.03,-0.13
Noise Level, dB	-97.9
Dynamic range, dB(A)	97.9
THD, %	0.00206
THD + Noise, dB(A)	-87.2
IMD + Noise, %	0.00516
Stereo crosstalk, dB	-79.1

This looks normal (And it still barely beats all the other ALC1220).


If these results are true, I really should avoid using the on-board headphone out because it's trash Audio. But my USB DAC is glitchy and function is very limited, and I don't wanna get into another round of HiFi shopping I decided to quit years ago. Can someone that has this setup with ALC1220 and front headphone jack confirm me if my concerns are real? Is there anything I am missing or have done so wrong to cause these numbers?

 

[twsecrest]

With the wires connecting the motherboard, to the front panel headphone jack.
The wires can pick up electrical noise that is generated inside the computer case.

 

[_thelaughingman]

In most cases that I have seen, implementation of onboard Dac’ in motherboards has been subpar and near disastrous for good audio quality. I honestly gave up on using onboard audio chipsets because of poor audio quality and have gone the route of USB dacs since early 2000s. FWIW, I’d stay away and just use the DAC/amp you have on USB. You can invest in a $200 Dac/amp stack such as JDS labs Atom series and get great results with no noise penalty.

 

[vrships]

In most cases that I have seen, implementation of onboard Dac’ in motherboards has been subpar and near disastrous for good audio quality. I honestly gave up on using onboard audio chipsets because of poor audio quality and have gone the route of USB dacs since early 2000s. FWIW, I’d stay away and just use the DAC/amp you have on USB. You can invest in a $200 Dac/amp stack such as JDS labs Atom series and get great results with no noise penalty.

In 2000s the on-board audio was extremely poor. The AC97 standard was a joke compared to the current HD Audio standard. I know how the manufacturers implement the circuits and components matter quite much to the final audio performance. But from my memory, all objective loop-back tests since ALC892 age, from any quality motherboard manufacutuer, reliably get us SNR>90dB and THD+N <0.01%. But my ALC1220 which is supposed to be still best of the line are performing 30dB worse.

If I could get mine on board working normally as it should, this solution absolutely wins in its functionality and conciseness. And nowadays they have enough power to drive 300ohm headphones to deafening volumes, and Realteck ALC since 3 years ago has supported DSD direct output. I really hope this could work out.

 

[vrships]

With the wires connecting the motherboard, to the front panel headphone jack.
The wires can pick up electrical noise that is generated inside the computer case.

As I understand it, the signal lines and ground wire are picking up the same amount of noise, it would cancel out for headphone output so it will not deteriorate the sound quality. This will be an issue for loop-back tests since the line-in ground has different noise levels that conflicts the headphone out ground. This is why ground loop issue results in very bad measurements if not taken care of. But I still feel getting 70dB SNR repeatedy is way too bad, even with all the added noise, for what the onboard audio can handle.

 

[DVDdoug]

In the analog days 70dB was GREAT!!!

The current ACX audiobook spec is 60dB or better, and they will actually reject your work is "over processed" and the noise floor is too low.... I don't know what that limit is but "pure digital silence" will be rejected.

I've had motherboard audio with audible noise when the hard-drive was running. I never measured it but it was worse than -70dB.

I was doing some "experiments" with an oscilloscope on another computer once and the output was completely unfiltered. I was shocked! You could see the digital stair-steps! I wasn't using this computer for "high fidelity" listening... I just had some average computer speakers connected but I had never noticed anything wrong... Then after thinking about it.... The harmonics would be ultrasonic so I couldn't hear them, plus the waveform might be filtered by the amplifier (built-into the speakers) and the speaker drivers would provide mechanical filtering.

 

[BeerBear]

Can someone that has this setup with ALC1220 and front headphone jack confirm me if my concerns are real? Is there anything I am missing or have done so wrong to cause these numbers?

It's quite normal for the front audio output to perform worse. The cable running inside to the front port has more chance to pick up some interference. While sometimes you can also get the opposite. See this video, for example.


Also, the way you tested this is not ideal. First, because the line input on the motherboard can itself be a bottleneck in terms of audio performance.
And second, because there can be some ground loop issue adding to the noise. Ideally, you would record the signal into an electronically separate battery powered device. You could use a USB soundcard connected to a smartphone, for example.

 

[AnalogSteph]

Using the front mic in to record the front headphone out should yield more sensible results. (Headphone outputs should also be tested using dummy loads or real ones.)

Also, if things really were that bad, you should be able to hear it with (sufficiently sensitive) headphones / in-ears.

Unfortunately, front panel wiring may very much vary. Some front panels will connect audio ground to chassis, yielding a nice ground loop. Others that don't may have substantial amounts of shared ground return resistance (1 wire in the HDA cable for mic + both headphone channels, and gauge may vary), so in low-impedance headphones L+R will be somewhat attenuated while L-R will not, effectively giving a slight stereo widening effect most noticeable by reduced bass.

There's even more variables... for example, in these days of powdercoated cases, you can't necessarily count on the chassis providing good electrical bonding between mo/bo and PSU, and it's quite well-known that the PCIe x16 connector may be carrying substantial ground return currents (in excess of its rated 75 W) when used with GPUs using aux power, i.e. that's the lower-resistance ground return path compared to the PCIe power cable. Poor ground contact on the 24-pin ATX connection can cause bizarre symptoms like NVMe SSDs dropping out - imagine mo/bo ground is elevated by 1 V... receiving 11 V instead of 12 V is not going to bother the CPU voltage regulator much, but if the SSD gets 2.3 V instead of 3.3 V, that's way out of spec.

 

[vrships]

It's quite normal for the front audio output to perform worse. The cable running inside to the front port has more chance to pick up some interference. While sometimes you can also get the opposite. See this video, for example.


Also, the way you tested this is not ideal. First, because the line input on the motherboard can itself be a bottleneck in terms of audio performance.
And second, because there can be some ground loop issue adding to the noise. Ideally, you would record the signal into an electronically separate battery powered device. You could use a USB soundcard connected to a smartphone, for example.

Thanks for sharing the video. This actually proves that ALC1220 on X370 Taichi indeed are noticeably better than ALC897 on Pro Z690.

The THD, noise and Dynamic range results of ALC1220 are pretty decent even compared to standalone units. And their power capability on 600ohm headphones is also acceptable. Output impedance and max power are the two things heavily limited by onboard solutions because they just don't have enough space for components, other than that, the intrinsic quality of the DAC and amplifier are more than satisfactory for ALC1220.

And again the front/rear difference is much smaller for these units than mine and my ALC1220 front is still almost worse than all onboard units here. Very concerning.

The line input of on board is a limiting factor but it only adds to the noise/distortion, etc by very small amount if 1) you line out quality is ~90dB 2) your line in quality is better than 100dB. Because the line-in doesn't go through amplifier circuits and ADC is less susceptible to unpredictable noise/errors, it is easy to achieve the spec'ed 110dB performance for most boards.

A theoretical 90dB performance output, adding all the bad contributions from a 100dB line in , will results in a number of only 89.5dB. That wouldn't be a problem to differentiate less than top-tier DAC+amps.

I'll try to isolate the ground of the line in using some external device to see if the measurement can improve.

 

[vrships]

Using the front mic in to record the front headphone out should yield more sensible results. (Headphone outputs should also be tested using dummy loads or real ones.)

Also, if things really were that bad, you should be able to hear it with (sufficiently sensitive) headphones / in-ears.

Unfortunately, front panel wiring may very much vary. Some front panels will connect audio ground to chassis, yielding a nice ground loop. Others that don't may have substantial amounts of shared ground return resistance (1 wire in the HDA cable for mic + both headphone channels, and gauge may vary), so in low-impedance headphones L+R will be somewhat attenuated while L-R will not, effectively giving a slight stereo widening effect most noticeable by reduced bass.

There's even more variables... for example, in these days of powdercoated cases, you can't necessarily count on the chassis providing good electrical bonding between mo/bo and PSU, and it's quite well-known that the PCIe x16 connector may be carrying substantial ground return currents (in excess of its rated 75 W) when used with GPUs using aux power, i.e. that's the lower-resistance ground return path compared to the PCIe power cable. Poor ground contact on the 24-pin ATX connection can cause bizarre symptoms like NVMe SSDs dropping out - imagine mo/bo ground is elevated by 1 V... receiving 11 V instead of 12 V is not going to bother the CPU voltage regulator much, but if the SSD gets 2.3 V instead of 3.3 V, that's way out of spec.

That's a lot more details than I am aware of. The front panel header of the chasis could also contribute a lot to noise and interference and mine has mic and headphone in one 3.5mm jack, which I am not sure if it worsens the problem.

I checked again when doing RMAA, fully load the GPU will create a -70~-80dB noise floor from 10~3kHz from HP-out to the line-in, which I can absolutely not hear at all otherwise using the headphone. But creating a pink noise from software, which reads <-90dB floor to the line-in RMAA, is clearly noticeable from the headphone. So that means the noise in loopback to the line-in is probably not real if the output is delivered to the headphone? Not sure if I configured everything correctly here so the conclusion makes sense?

 

[unpluggged]

it would cancel out for headphone output

Why should it cancel out?

 

[lashto]

In the analog days 70dB was GREAT!!!

I don't have any complaints about tube amp at 50-60dB. Haven't heard many from others either.
And LPs are also around 50-60dB but they sem to have a hige (and growing) fan club

OTOH, haven't heard of any fans of DACs below 90dB (tube output DACs count as analog).

 

[shevalier]

Some front panels will connect audio ground to chassis USB, yielding a nice ground loop.

Article: Fix front panel audio EMI noise

I was having a problem with my front panel audio port for a long time until yesterday I found a solution. I am sure many of you have that problem.. So here I start. SYMPTOMS While listening either through headphones or speakers through the front panel audio, you can hear a whining noise...

www.techpowerup.com

or

PC HD audio front panel rewiring + EMI filtering - Page 1

PC HD audio front panel rewiring + EMI filtering - Page 1

www.eevblog.com

Need to cut the front panel ground polygon.

 

[SebaMK82]

HI! check this https://www.audiosciencereview.com/...te-z390-aorus-motherboard-audio-review.13083/
Its same chipst sound es9118 / 1220 VB.
Works like a DAC around $100...
I have a z390 Aorus Master and works fine for me.

 

[drankorgel]

Hello, i'm not super familiar with all things science and audio but i stumbled uppon this thread, looking for the best port to connect my new headphones to on also a ALC1220-VB.

You say "front headphone-out". Do you mean the bridged front out on your computer case by this? Like as opposed to sticking it directly on the motherboard on the back of the computer? Because after reading this thread and other threads, I couldn't find a clear answer and decided to just read the motherboard manual which states:

When I connect my headphones to this port (so to be clear: the port on the back of the pc) the headphone amp option also comes available in the Realtek software. Which doesn't happen when connected to the front port.

So i don't know if i'm misinterpreting your thread or anything but maybe it wasn't meant to be connected to the front panel in the first place?