• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. 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!

Source computer's HUGE effect on fidelity

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
20,524
Likes
37,057
We know (from reading other audiophile accounts on other forums) that the feeding a DAC is an inherently risky, complex, and difficult task. Everything related to the endeavor can cause tremendously audible differences in the resulting sound. USB connections especially have seemed like one big mistake to pursue. Starting with the computer as source one may need a linear rather than switching supply for the beastly machine. One must choose the playback software carefully or disaster could result. One then needs a USB cable. Should one be so foolish as to simply connect a simple cable then no hope is left for pleasing musical results. At a minimum you need a linear supply for the USB 5 volt lines, a USB filtering reclocking device (one is well advised to spend 300% of the cost of the DAC on the USB connection). Then one should feed the DAC itself with nothing if not a linear PS with audiophile approval.

I thought I would provide some data on just how large the effect of poorly done connections and a lack of care in use of a PC source for music via USB really is. Please, don't try this at home. The variability in sound can be maddening.

At the core of my examination the components that did not change. I used a Focusrite Forte for the DAC. Attached to a wall wart PS. Horrifyingly connected via the stock 1 meter clear shielded USB cable without any intervening protection. It fed via balanced connection an Antelope Audio Zen Tour ADC. Also carelessly connected via the stock 2 meter shielded USB cable. Powered by the wall wart power supply. The Forte was operating at 48 khz and the Zen Tour was recording at 192 khz.

Source computers:

Circa 2010 Lenovo T410 laptop running Windows 10 Creators Edition with spinning 5400 rpm HD and 4 gig of ram (powered by a wall wart).

2014 Apple Macbook Pro with SSD and 8 gig memory powered by the battery.

In the green graphs the Macbook was connected to the Forte and the Zen Tour was feeding the Lenovo. In the blue graphs computers were reversed. And with predictable consequences I might add.

Now for the results.

First I'll show the 300 hz and below FFT while a 1 khz tone is being played looking for corruption by the power supplies bleeding through.

300hz and below.png


Next I'll show the silent noise floor. The little pip at 31 khz is an idle tone in the Forte DAC.

silence.png


Now the 1 khz high level tone.

1khz.png


-4 db of white noise. Please steel yourself before viewing the result.








white noise.png


Now I think you can handle the 18+19khz IMD tone result.

18+19khz tone.png


As a bonus, we'll look at a portion of a twin tone sweep at near max level. Two tones placed 1 khz apart and swept across the band. Near 19 khz in this image.

twin tone sweep.png


Hopefully, these sobering results will prevent other audiophiles from thinking one can grab any old PC lying about and chuck the nearest handy USB cable into the USB jack on a DAC and expect music to result.
 
Last edited:
OP
Blumlein 88

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
20,524
Likes
37,057
I saved the 12 khz tone for its own post. As one knows much of the horrid effects are ethereal jitter effects. The 12 khz tone is not a precise enough test, but I thought I would show it.

A note: the difference in clock speed between the DAC and ADC clocks was 317.9 parts per billion. After swapping around the computers the difference increased to 320.1 ppb. An increase of 2.2 ppb.

EDIT:to correct decimal mistakes.

12khz tone.png
 
Last edited:
OP
Blumlein 88

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
20,524
Likes
37,057
Shocking , profound results.. so if I’m reading it right, if I play led Zeppelin through my computer and feed a DAC via USB it will probably end up being the spice girls and all music thusly ruined??....??....?

If you are lucky. It could be worse, so take proper precautions.
 
OP
Blumlein 88

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
20,524
Likes
37,057
Do you mean -40 there instead of -4?

No it is - 4db for the whole signal in its 24 khz bandwidth. Just happens with binning by the FFT the levels drop to near -40 db per frequency bin.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,393
Location
Seattle Area
Ah yes. Momentary brain fart. :)

I will summarize your testing: different computers produced identical results on a decent DAC regardless of test signal.
 
OP
Blumlein 88

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
20,524
Likes
37,057
Ah yes. Momentary brain fart. :)

I will summarize your testing: different computers produced identical results on a decent DAC regardless of test signal.

Well that would be the short uninteresting version.

I actually planned a "Can you trust your specifications?" thread after testing the new to me recording interface. I noticed I had done identical tests with reversed computers and posted this instead.
 

FrantzM

Major Contributor
Forum Donor
Joined
Mar 12, 2016
Messages
4,337
Likes
7,731
I saved the 12 khz tone for its own post. As one knows much of the horrid effects are ethereal jitter effects. The 12 khz tone is not a precise enough test, but I thought I would show it.

A note: the difference in clock speed between the DAC and ADC clocks was 317.9 parts per billion. After swapping around the computers the difference increased to 320.1 ppb. An increase of 2.2 ppb.

EDIT:to correct decimal mistakes.

View attachment 9499

Hi
You seem to forget however that the brain , not the ears, is the processing device. Once you know that there is a staggering 2.2 parts per BILLIONS difference, then you would understand why the bass was more organic, the midrange more fleshed out and the general width of the soundstage expanded beyond our galaxy. Results would have been even better if you had used the $2500 USB cable from Desterberg Audio...
All that tells me your system is not resolving enough
o_O
:p
 

Brad

Active Member
Joined
Nov 8, 2016
Messages
114
Likes
35
No it is - 4db for the whole signal in its 24 khz bandwidth. Just happens with binning by the FFT the levels drop to near -40 db per frequency bin.

Can you explain this a bit further?
Is it just a normalisation issue?
 

Sal1950

Grand Contributor
The Chicago Crusher
Forum Donor
Joined
Mar 1, 2016
Messages
14,073
Likes
16,608
Location
Central Fl
All that tells me your system is not resolving enough
Or his ears aren't educated enough to hear the very obvious differences that the measurements didn't reveal.
Our knowledge and measurement devices are not yet sophisticated enough to reveal the changes.
Any first day reader of the Audiophool Media has learned these basic truths. :eek:

@Thomas savage Love the new avatar. :p
 

Thomas savage

Grand Contributor
The Watchman
Forum Donor
Joined
Feb 24, 2016
Messages
10,260
Likes
16,298
Location
uk, taunton
Or his ears aren't educated enough to hear the very obvious differences that the measurements didn't reveal.
Our knowledge and measurement devices are not yet sophisticated enough to reveal the changes.
Any first day reader of the Audiophool Media has learned these basic truths. :eek:

@Thomas savage Love the new avatar. :p
..It’s my other side ;)
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,393
Location
Seattle Area
Can you explain this a bit further?
Is it just a normalisation issue?
I plan to do a full article on this. For now, simple explanation is that the noise is spread across many buckets and the display only shows the individual buckets. To know the true noise floor, you need to add them all up which can easily be 10s of dbs.
 
OP
Blumlein 88

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
20,524
Likes
37,057
Can you explain this a bit further?
Is it just a normalisation issue?

To expand on what Amir said already.

White noise has equal signal at all frequencies. Let us suppose I have max signal level across a 24 khz band. The sum total of all frequencies add together to equal the signal strength.

Now if I filter this into two bands 0-12 khz and 12 khz to 24 khz each portion has half the total signal level. The total signal level for each band drops by - 3db.

If I instead broke it into 4 bands 0-6 khz, 6-12 khz, 12-18 khz and 18-24 khz, then each portion has one quarter the total signal level and each portion will be at - 6db.

An FFT does something like this. It divides the total bandwidth being measured into many small bands of frequencies which are called bins of the FFT. So the signal level in each band drops to a lower level. In the case of my graph it was several thousand bins and enough to drop the signal from - 4db for the total 24 khz bandwidth to around -40 db for each bin of the FFT. Here is the same exact signal plotted with an FFT having much fewer bins. Notice the graph is courser because of fewer bins to work with and the signal level is higher on the graph because each bin being wider in frequency now contains more signal level.

white noise 4kfft.png
 

pozz

Слава Україні
Forum Donor
Editor
Joined
May 21, 2019
Messages
4,036
Likes
6,827

JeffS7444

Major Contributor
Forum Donor
Joined
Jul 21, 2019
Messages
2,347
Likes
3,508
I used to cringe when reading about audiophile PCs, typically loaded with fastest CPU, max RAM, some sort of Pro flavor of Windows, sometimes stripped down to only the most essential services. I felt like spreading the rumor that nothing less than Windows Datacenter would suffice for the highest levels of fidelity, but decided that my attempt at humor could become part of accepted audiophile wisdom, and I’m not that evil yet (and more importantly, hadn’t figured a way to monetize it)
 

Jimbob54

Grand Contributor
Forum Donor
Joined
Oct 25, 2019
Messages
11,066
Likes
14,697
I used to cringe when reading about audiophile PCs, typically loaded with fastest CPU, max RAM, some sort of Pro flavor of Windows, sometimes stripped down to only the most essential services. I felt like spreading the rumor that nothing less than Windows Datacenter would suffice for the highest levels of fidelity, but decided that my attempt at humor could become part of accepted audiophile wisdom, and I’m not that evil yet (and more importantly, hadn’t figured a way to monetize it)
Unless one uses Roon (and I assume similar software) to manage larger libraries and/ or large streaming service playlists. Then you actually do need need a decent bit of RAM/ processor to get the thing moving. But thats not related to audio quality on playback- just usability.
 

majingotan

Major Contributor
Forum Donor
Joined
Feb 13, 2018
Messages
1,511
Likes
1,781
Location
Laguna, Philippines
I used to cringe when reading about audiophile PCs, typically loaded with fastest CPU, max RAM, some sort of Pro flavor of Windows, sometimes stripped down to only the most essential services. I felt like spreading the rumor that nothing less than Windows Datacenter would suffice for the highest levels of fidelity, but decided that my attempt at humor could become part of accepted audiophile wisdom, and I’m not that evil yet (and more importantly, hadn’t figured a way to monetize it)

I'd advise against high powered computers as I can't stand the fan noise while playing music. Even at 800 MHz music playback these chips still needed fan operation which is ridiculous. I have a fanless Surface Pro with all of the services on normal startup and never worry about my computer fan noise ruining my listening session
 

bequietjk

Senior Member
Joined
Jan 4, 2019
Messages
462
Likes
474
I wonder if there will be a positive benefit using a Matrix Element H VS native USB port.
 

digicidal

Major Contributor
Joined
Jul 6, 2019
Messages
1,981
Likes
4,838
Location
Sin City, NV
I'd advise against high powered computers as I can't stand the fan noise while playing music. Even at 800 MHz music playback these chips still needed fan operation which is ridiculous. I have a fanless Surface Pro with all of the services on normal startup and never worry about my computer fan noise ruining my listening session

There are many fanless options (I use this at home for roon and these at work for silence for my users). No need to use anything with fans (or any moving parts for that matter) in an audio PC. Sure most of them take a little bit of technical knowledge and mechanical assembly skills... but much less than assembling an amplifier kit or speaker kit requires. Plus there are many retail options as well.

But despite having a fanless setup which is decently powerful as my Roon core (3.5GHz 4C CPU, 16GB RAM, 512GB SSD) with so many inexpensive Roon capable output devices (my oppo player and BT speaker, Marantz prepro's and Denon AVR all work very well) I could have installed it on a 1U rackmount server with "wind tunnel fans" like I use at work - and I still wouldn't hear it downstairs on the other side of the house. :)
 
Top Bottom