miero
Senior Member
- Joined
- Aug 1, 2018
- Messages
- 351
- Likes
- 380
Why many DACs shares a distortion "bump" between -40dBFS and -20dBFS? What a specialty is at these levels?
-
Welcome to ASR. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
Do You Need an External DAC/Headphone Amplifier?
- Thread starter amirm
- Start date
Onboard codecs are neither USB nor SPDIF, they are PCI by nature. If you read the links in my previous reply, I linked to Archimago's review of ALC892:
https://www.audiosciencereview.com/...gh-of-a-sample-rate-is-enough.4037/post-95240
Look at his J-Test results with CPU and GPU stress test, there are no differences. Also, if you see my twin tones test, the tiny spurs showed up in Archimago's 44k J-Tests and "Digital Filter Composite" are actually resampling artifacts, they are not jitter.
He also tested the Lynx L22 soundcard with CPU/GPU stress test. Same result, slightly noisier with some power supply related spurs, but not higher jitter.
http://archimago.blogspot.com/2017/04/retro-measure-2002-lynx-l22-pci-audio.html
Sythrix
Senior Member
- Joined
- Mar 8, 2018
- Messages
- 331
- Likes
- 264
Do you have an actual desktop Amir? I am interested in how current motherboard built-in sound for PCs is measuring, especially considering many MBs are advertising a level of sound and clarity on par with an exterior or dedicated solution. They even go so far as to separate the main traces on the MB, creating a visible path to show you the audio components aren't directly interacting with the other MB components (don't know if I believe them).
Anyways, with these solutions generally sitting directly above the power supply in most systems, along with the potential for noise inside of a computer system (fans, electrical noise, etc), it would be interesting to see some actual measurements. Frankly I don't believe the manufacturers and their claims.
Too bad it's such a hassle to test stuff like that, considering you would need an entire PC, with all relevant components on and running, to get an accurate portrayal of how it measures.
I currently have an Asus STX II sound card. It would be nice to know whether or not it had been worth the investment, especially considering it is also an interior device.
Anyways, with these solutions generally sitting directly above the power supply in most systems, along with the potential for noise inside of a computer system (fans, electrical noise, etc), it would be interesting to see some actual measurements. Frankly I don't believe the manufacturers and their claims.
Too bad it's such a hassle to test stuff like that, considering you would need an entire PC, with all relevant components on and running, to get an accurate portrayal of how it measures.
I currently have an Asus STX II sound card. It would be nice to know whether or not it had been worth the investment, especially considering it is also an interior device.
έχω δίκιο
Active Member
- Joined
- Apr 20, 2018
- Messages
- 249
- Likes
- 282
I meant no insult. I thought that you were horsing around with me when you wrote "Do different trackpads matter for sound quality, how do you measure that?"
I was making an analogy, not saying that trackpads affect sound. An external trackpad is worthwhile if the built-in trackpad is of poor quality, much as an external DAC/headphone amp is worthwhile if the built-in audio hardware is of poor quality.
έχω δίκιο
Active Member
- Joined
- Apr 20, 2018
- Messages
- 249
- Likes
- 282
As you state in your message, the only way to know is to perform measurements on your particular system. You would need to measure the performance of the Asus STX II card in your system, and then you would need to remove it and measure the performance of the motherboard sound system.
It might be interesting to see how Amir's desktop performed, especially if it showed admirable performance, but I don't see how one could extrapolate from that with regards to some other system.
OP
- Thread Starter
- #86
I do but it has been randomly crashing. So I finally bit the bullet and examined the crash dump and instantly found the offending driver. Uninstalled that and she is up. But the software on it is pretty old and I am upgrading it all. Should be able to post at least some data later today.
Asus mentioned RMAA performance. To get an apple to apple comparison, here is how the RME ADI2-Pro (non FS, old firmware) performed in RMAA:
http://prosound.ixbt.com/interfaces/rme/adi2pro/+24.shtml
http://prosound.ixbt.com/interfaces/rme-adi2pro.shtml
Here is a discussion about "AES17 dynamic range" vs "SINAD".
https://www.audiosciencereview.com/...ts-of-schiit-yggdrasil-v2-dac.3607/post-88533
I've really enjoyed my Asus Xonar Essence ST PCI card.
Sythrix
Senior Member
- Joined
- Mar 8, 2018
- Messages
- 331
- Likes
- 264
True, but I could send my card to Amir if he was interested and we could measure it that way, coupled with whatever his motherboard measured at. Then I could get an idea of its general performance. I don't fancy the idea of sending my whole system, but we would get some data.
OP
- Thread Starter
- #90
As promised, here are the measurements of my desktop Gigabyte Gaming B8 motherboard powered by an i7 CPU:
It makes lofty claims of audiophile sound:
I will be testing its headphone performance later.
First, let's start with our usual dashboard at full volume (default was 65%):
We get decent output compared to my laptop but still shy of 2 volt I like to see.
SINAD is a few dB worse than my laptop! As I always do, I knocked down the level gradually and at -6 dBFS, the performance is much improved:
That's 14 dB higher SINAD! Alas, now our output voltage has fallen way down to just 0.67 volts RMS.
Frequency response is ruler flat to 17 kHz:
At 22 khz, we have -3.4 dB droop. I like to see flat response to 20 kHz.
This is how it does on jitter and noise:
Lots of spikes visible although levels at -120 dB are most likely not audible.
Let's look at intermodulation distortion (IMD) versus level. We use IMD test because it can detect non-linearity without having to have wide capture bandwidth to gather higher frequency harmonics as THD requires. In this case, I am using the SMPTE dual tone which is pretty sensitive to low frequency distortion:
She starts doing well, matching dedicated external DACs but then goes nuts right at -35 dBFS but eventually recovers at -15 dB. In that region it is much worse than my laptop (in pink).
Let's analyze that by looking at the spectrum, first at -40 dB where response is well behaved:
On the left you see our composite waveform. it is a 60 Hz tone that is being modulated by the 7 kHz tone (shown as solid band because they are too close together). The spectrum on the right shows them nicely separated (the two peaks).
Here is what we get at -30 dB where things should be getting better (due to better SNR), not worse:
Focusing on the spectrum, we see a rise in our base noise floor (below 60 Hz tone) accompanies by harmonics of our 60 Hz which now spray throughout the spectrum.
I changed the frequency to 70 Hz but the same outcome remained so it has nothing to do with power supply.
There is definitely a design mistake here which is also visible in some desktop DACs but is much worse in this instance.
Let's look at linearity:
The higher output voltage helps it do a lot better than my laptop. Strangely though there is some positive offset as low as -43 dBFS or so. Eyeballing this, looks like we have good linearity up to CD's 96 dB dynamic range.
Speaking of dynamic range, I decided to start showing this from here on since manufacturers always advertise it (as SNR):
Even though it misses its 120 dB spec, it is still good at 107 dB.
Summary
We see more design issues here than on my HP Laptop. Peak performance is better but with that comes some anomalies such as mid-level rise in intermodulation distortion, and jitter.
Clearly we need to be measuring these PC subsystems to find the real jewels out there. I am open to building a case-less PC if people can loan their motherboards for testing.
It makes lofty claims of audiophile sound:
I will be testing its headphone performance later.
First, let's start with our usual dashboard at full volume (default was 65%):
We get decent output compared to my laptop but still shy of 2 volt I like to see.
SINAD is a few dB worse than my laptop! As I always do, I knocked down the level gradually and at -6 dBFS, the performance is much improved:
That's 14 dB higher SINAD! Alas, now our output voltage has fallen way down to just 0.67 volts RMS.
Frequency response is ruler flat to 17 kHz:
At 22 khz, we have -3.4 dB droop. I like to see flat response to 20 kHz.
This is how it does on jitter and noise:
Lots of spikes visible although levels at -120 dB are most likely not audible.
Let's look at intermodulation distortion (IMD) versus level. We use IMD test because it can detect non-linearity without having to have wide capture bandwidth to gather higher frequency harmonics as THD requires. In this case, I am using the SMPTE dual tone which is pretty sensitive to low frequency distortion:
She starts doing well, matching dedicated external DACs but then goes nuts right at -35 dBFS but eventually recovers at -15 dB. In that region it is much worse than my laptop (in pink).
Let's analyze that by looking at the spectrum, first at -40 dB where response is well behaved:
On the left you see our composite waveform. it is a 60 Hz tone that is being modulated by the 7 kHz tone (shown as solid band because they are too close together). The spectrum on the right shows them nicely separated (the two peaks).
Here is what we get at -30 dB where things should be getting better (due to better SNR), not worse:
Focusing on the spectrum, we see a rise in our base noise floor (below 60 Hz tone) accompanies by harmonics of our 60 Hz which now spray throughout the spectrum.
I changed the frequency to 70 Hz but the same outcome remained so it has nothing to do with power supply.
There is definitely a design mistake here which is also visible in some desktop DACs but is much worse in this instance.
Let's look at linearity:
The higher output voltage helps it do a lot better than my laptop. Strangely though there is some positive offset as low as -43 dBFS or so. Eyeballing this, looks like we have good linearity up to CD's 96 dB dynamic range.
Speaking of dynamic range, I decided to start showing this from here on since manufacturers always advertise it (as SNR):
Even though it misses its 120 dB spec, it is still good at 107 dB.
Summary
We see more design issues here than on my HP Laptop. Peak performance is better but with that comes some anomalies such as mid-level rise in intermodulation distortion, and jitter.
Clearly we need to be measuring these PC subsystems to find the real jewels out there. I am open to building a case-less PC if people can loan their motherboards for testing.
Zilfallion
Member
- Joined
- Feb 16, 2018
- Messages
- 82
- Likes
- 65
Outside the IMD, that's about what I was expecting in output power and THD+N, a little better in some areas, but that IMD is really weird and pretty poor. Still relatively decent besides that. I can see why you might not need an external DAC/Amp if you didn't have hard to drive headphones, though it's far from perfect. Gives me a decent idea of what my motherboard does since I doubt they vastly changed their implementation between motherboards from B250 to AB350.
OP
- Thread Starter
- #93
Wow, fascinating. Given the reference to RMAA, seems like they did indeed measure it.
Which motherboard has that?
Thank you. Could you measure the output impedance, too?
Is it really worth going to all that trouble?
I think one of the best reasons for buying external DACs and amps is that it makes for one less thing you need to worry about when buying motherboards/tablets/PCs
OP
- Thread Starter
- #98
I will do that when I test it against load.
I agree, but for multi channel movie sound it may be an option. For example right now I go HDMI out to an AVR, but I suppose if I discovered a motherboard that had some great DAC I might consider analog out.
Seems like lots of work for no gain however.
Sythrix
Senior Member
- Joined
- Mar 8, 2018
- Messages
- 331
- Likes
- 264
While I applaud you for considering this, I think there are two issues.
1. An open-case testing scenario is not typical usage. I believe the results would be suspect without the metal enclosure, fans and other components in a tightly confined space, as you would see in a real desktop build.
2. Not many if any people are going to want to tear down their system to send just the motherboard and CPU along. It would be a lot of work and of course leave the system useless. Unless you catch someone who is in the middle of building a new system, I don't believe you'll get many takers. Honestly it might work out better just to see if anyone could lend the whole system.
Regardless, thanks for measuring yours! Gives us an idea of what built-in desktop sound can measure like.
Similar threads
