• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

Impulse response of two DACs (Topping D10 & FX-Audio DAC X7)

daftcombo

Addicted to Fun and Learning
Patreon Donor
Joined
Feb 5, 2019
Messages
766
Likes
309
#1
Hi,

I have at home a Topping D10 and a FX-Audio DAC X7. The later sounds harsh and messy in the vocals and I wanted to know why.

I measured the RCA ouput of each DAC with my Focusrite Scarlett 2i4 2nd Gen and REW. Here are the .mdat files:
https://www.sendspace.com/filegroup/mChxt2IZ+Kd9ZkKNdWrOpQ

The main difference is for the impulse response: almost no oscillation before the peak for the DAC X7.

There is also a bigger phase shift for that unit.

Do you think those measurements could account for what I hear?

Cheers.
 

Attachments

solderdude

Major Contributor
Joined
Jul 21, 2018
Messages
1,717
Likes
2,665
Location
The Neverlands
#3
Hi,

I have at home a Topping D10 and a FX-Audio DAC X7. The latter sounds harsh and messy in the vocals and I wanted to know why.

Do you think those measurements could account for what I hear?

Cheers.
Vocal frequency range is between 100Hz and 8kHz. No reconstruction filter rings in that part of the frequency range. It only rings on illegal signals and signals very close to the Nyquist frequency (which music doesn't have a lot of) so it is highly unlikely that would account for what you hear.
 

daftcombo

Addicted to Fun and Learning
Patreon Donor
Joined
Feb 5, 2019
Messages
766
Likes
309
#4
Vocal frequency range is between 100Hz and 8kHz. No reconstruction filter rings in that part of the frequency range. It only rings on illegal signals and signals very close to the Nyquist frequency (which music doesn't have a lot of) so it is highly unlikely that would account for what you hear.
Hi,
From the distortion, phase & other graphes, I cannot spot any obvious difference which could account for the perceived difference, which is obvious and could be spot on a ABX, I'm confident here.
 

daftcombo

Addicted to Fun and Learning
Patreon Donor
Joined
Feb 5, 2019
Messages
766
Likes
309
#6
It looks like the Topping is using a linear phase filter (causes pre-ringing) and the FX-Audio is using minimum phase (no pre-ringing).
Do you know other examples of DACs using each of those filters?
 

pwjazz

Active Member
Joined
Sep 21, 2018
Messages
215
Likes
224
#7
Do you know other examples of DACs using each of those filters?
I don't know that it's usually published in the specs. Some DACs allow selecting the filter. For example, I have a Hifiberry DAC+ which lets me pick from a variety of filters, including linear and minimum phase.
 

solderdude

Major Contributor
Joined
Jul 21, 2018
Messages
1,717
Likes
2,665
Location
The Neverlands
#8
I don't know if HQPlayer has a trial period but if you hear a 'hardness' in the mids it may be worth a try.
Use in upsampling mode.
You can ask @Miska who is a member here.

Maybe, just maybe he can help you here...
At least it looks like this could be a case where the software could be potentially of help here (for technical reasons)
 
Last edited:

m8o

Member
Joined
Jan 28, 2019
Messages
40
Likes
14
#9
Vocal frequency range is between 100Hz and 8kHz. No reconstruction filter rings in that part of the frequency range. It only rings on illegal signals and signals very close to the Nyquist frequency (which music doesn't have a lot of) so it is highly unlikely that would account for what you hear.
But isnt it also the case that ringing can interact with other legitimate or illegitimate signals (building up even more illegitimate signals) and produce IM distortion in the lower audible frequencies?
 

solderdude

Major Contributor
Joined
Jul 21, 2018
Messages
1,717
Likes
2,665
Location
The Neverlands
#10
Where would the illegitimate signals come from ?
They don't exist in recordings. Only in certain (illegal) test signals that are intended for specific tests only.
The little info there is nearing Nyquist is so low in amplitude that if it somehow became 'mirrored' into the mid frequencies it would be masked by those frequencies as they are much higher in level.

The only way of gathering proof for this is to record the analog out of both DAC's (in at least 192/24) and use the software from pwjazz to see how much it differs from the original file. The fun part is you can use real music that shows the differences clearly.
 

Blumlein 88

Major Contributor
Joined
Feb 23, 2016
Messages
5,983
Likes
5,218
#11
Yes HQ player has a trial period. Bottom of this page. Read the Quickstart guide thoroughly before trying to use it.

https://www.signalyst.com/consumer.html

To daftcombo, you could make your recordings with the Focusrite (maybe at 96 khz) and then see if you can ABX them in the Foobar ABX. Confidence in ABX things can sometimes turn out to be misplaced. And don't feel bad it happens to everyone. Whether that is the case here I don't know.
 

bennetng

Addicted to Fun and Learning
Joined
Nov 15, 2017
Messages
557
Likes
413
#12
Hi,

I have at home a Topping D10 and a FX-Audio DAC X7. The later sounds harsh and messy in the vocals and I wanted to know why.

I measured the RCA ouput of each DAC with my Focusrite Scarlett 2i4 2nd Gen and REW. Here are the .mdat files:
https://www.sendspace.com/filegroup/mChxt2IZ+Kd9ZkKNdWrOpQ

The main difference is for the impulse response: almost no oscillation before the peak for the DAC X7.

There is also a bigger phase shift for that unit.

Do you think those measurements could account for what I hear?

Cheers.
The problem is, if you measure impulse response, then you can see potential differences in impulse response. If you measure SINAD then you can see potential differences in SINAD, if you measure IMD, jitter... etc etc.

So, let's say you can really hear a difference after doing an ABX test, are you sure the differences must be due to impulse response? If you record the output with another ADC you will also add another layer of distortion, adding even more uncertainties. Impulse response is one of the things that can be easily visualized in measurements even with some very limited hardware like the Realtek ADC on every PC mainboard. If you are seeing similar performance in other aspects you are probably just seeing the bottleneck of your ADC.

If you really want to isolate impulse response alone, you should not compare two DACs, but compare the impulse response.

Use a software resampler with phase and other settings which will affect impulse response:
https://hydrogenaud.io/index.php/topic,67373.0.html
https://www.audiosciencereview.com/forum/index.php?threads/delta-sigma-vs-“multibit”-what’s-the-big-deal.5359/post-146220

Use foobar's file converter with the resampler plugin to upsample some files with different settings. You may also need to adjust the audio level during conversion to avoid filter induced clipping, to make sure the "harshness" is not caused by clipping.
https://www.audiosciencereview.com/...oesnt-adhere-to-scarlet-book.5901/post-132540

Then ABX the converted files.
 
Last edited:

m8o

Member
Joined
Jan 28, 2019
Messages
40
Likes
14
#13
Where would the illegitimate signals come from ?
Weren't you just talking about ringing? The ringing is an illegitimate signal [I should have used the word 'tone' instead]. And what I was saying further is, is it not that though those can be in an inaudible frequency, thru IMD can create other illegitimate signals [tones] in the audible frequencies. Yes, low in amplitude; difficult to hear/detect as you point out. But isnt this what people call "the digital sheen", and what we do detect in the effect on musical playback when we change (or compare) a dac's filter?
 

solderdude

Major Contributor
Joined
Jul 21, 2018
Messages
1,717
Likes
2,665
Location
The Neverlands
#14
I am quite certain that what we detect when changing filters (44.1/16) is roll-off in the treble which can be several dB and is quite audible.
When folks complain about the digital sheen they are listening to a DAC that has no roll-off.

Of course, there can be the odd recording and amplifier or transducer that may cause some other effects by interference but those that complain about the sheen or glare say it is always there, not in the odd recording.
This leaves something that is measurable which is roll-off within the audible range.

Funny part is that the same folks also complain about this with 96/24 and 192/24 recordings where there is no roll-off in the audible range and filters ring well outside of the audible range.

It appears to be a mix of sighted vs unsighted + placebo and in some cases apples and oranges that are compared. Throw them on one pile and ....

The OP assumed that the difference in filtering would account for the perceived sound differences.
I wasn't the one saying it could be ringing. I argued that vocals don't have harmonics that are so high up that they could be causing ringing and make voices sound poor.
The filter ringing signals are merely test signals that don't (can't even) exist in music recordings. Just load music into an editor, remove the lower frequencies and zoom in on the 'fast transients' and you can see there are no spikes and sudden jumps unless the files are corrupt.
 

andreasmaaan

Major Contributor
Patreon Donor
Joined
Jun 19, 2018
Messages
3,517
Likes
2,702
#15
Hi,
From the distortion, phase & other graphes, I cannot spot any obvious difference which could account for the perceived difference, which is obvious and could be spot on a ABX, I'm confident here.
Would love you to prove that ;)
 

andreasmaaan

Major Contributor
Patreon Donor
Joined
Jun 19, 2018
Messages
3,517
Likes
2,702
#16
If the FX X7 is a similar design to FX X6, perhaps the explanation lies here?

THD+N vs Frequency Topping D30 Exasound E32 FX-Audio DAC-X6.png


Or even better, are there measurements of the X7 anywhere that anyone knows of?
 
Last edited:

daftcombo

Addicted to Fun and Learning
Patreon Donor
Joined
Feb 5, 2019
Messages
766
Likes
309
#17

andreasmaaan

Major Contributor
Patreon Donor
Joined
Jun 19, 2018
Messages
3,517
Likes
2,702
#18
I know of only one "hi-fi" review, in french:
https://www.qobuz.com/fr-fr/info/Hi-Fi/Bancs-d-essai/FX-Audio-DAC-X7-restitution-claire181524
and there are no measurements. It suggests phase shifts but don't really explain how they came to that conclusion.
IMHO, subjective reviews of DACs are almost useless. Most DACs sound the same or very similar. Something needs to be really amiss (perhaps intentionally) for a DAC to sound halfway near as distinctive as most subjective reviews suggest they do ;)

Did someone check my own measurement? i linked them in .mdat format in my first post.
Had missed that but just having a look now. How did you take the measurements?

The Topping's harmonic distortion measurement looks very different from the measurements Amir took on the AP Analyzer... I wonder if these measurements aren't dominated by whatever setup you used to take them?
 

daftcombo

Addicted to Fun and Learning
Patreon Donor
Joined
Feb 5, 2019
Messages
766
Likes
309
#19
I made a 20 Hz o 22 kHz sweep in REW, having the RCA out of each unit at a time straight in the input of the Scarlett 2i4 2nd Gen. Both REW and the soundcards are configurated to work in ASIO.
Maybe it's not a perfect method from an absolute point of view, since the Scarlett is not perfect neither (though I calibrated REW for it) but I thought it could help comparing both the D10 to the X7 since the method is the same for both and since we already have a difference in the shape of the impulse.

Indeed there is a very strange and bad harmonic distortion for my D10! I don't know why.
X7 looks cleaner but going nastier in the treble, a bit like the X6.
 
Last edited:

andreasmaaan

Major Contributor
Patreon Donor
Joined
Jun 19, 2018
Messages
3,517
Likes
2,702
#20
I made a 20 Hz o 20 kHz sweep in REW, having the RCA out of each unit at a time straight in the input of the Scarlett 2i4 2nd Gen.
Maybe it's not a perfect method from an absolute point of view, but I thought it could help comparing both the D10 to the X7 since the method is the same for both and since we already have a difference in the shape of the impulse.

Indeed there is a very strange and bad harmonic distortion for my D10! I don't know why.
X7 looks cleaner but going nastier in the treble, a bit like the X6.
Given the measurement setup, it's safe to assume that the devices don't measure worse than that. So the X7 seems to have a cleaner top end than the X6.

Not sure I'd trust the measurements to be accurate otherwise, however. Both devices show distortion in the range of 0.3 - 1%. That seems completely implausible to me for a DAC, let alone the D10 which measured more in the range of 0.0003 - 0.001% on Amir's AP.

Hard to say what it is you're hearing without more accurate measurements IMHO. The next thing I'd try would be to see if I could consistently pick the difference blind.
 

Similar threads

Top Bottom