Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Dedicated soundcards - are they worth it? (Read 13509 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Dedicated soundcards - are they worth it?

I'm currently using a onboard Realtek ALC899 for all kinds of playback: Movies, music and games.
It is configured at 48kHz and 24bit and hooked up to a Sennheiser HD518 and a Teufel C 200 B (via line, not USB).

As for my ears, I can make out a difference in a blind listening test between LAME v4 and lossless if the conditions and the song* are optimal. Anything above that is beyond me and v4 is already pretty difficult to make a out a difference (= no impact on casual listening).

Now, could I (or you) possibly hear a difference when I upgrade to a dedicated soundcard by Asus or Creative? I'm talking about hearing a difference, not measuring with RMAA and the likes.

I can't really take those "audiophile reviews" serious. They talk and praise cards for their clarity and additional soundstage and whatnot, sounding as if it were obvious to everyone. Sound more like bollocks to me...
Spec-wise the ALC899 already seems to be pretty good after all.

So is an investment of about 70-90€ worth the difference - if there is even a difference? And in case the answer is yes: can you recommend a card in the pricerange (needs to be PCIe)?

* Used Rue Des Cascades by Yann Tiersen from the Rue Des Cascades Album CD (44.1kHz and 16bit)

Dedicated soundcards - are they worth it?

Reply #1 –
Rmaa your device and see if its any good. If its poor you might hear a difference. If its already fine you won't.

Dedicated soundcards - are they worth it?

Reply #2 –
Unless you can notice some particular problem with your source compared to other sources, I wouldn't think it's worth the bother. I have heard some onboard soundcards that are just complete crap, with very audible static, popping, crackling, etc, but if none of that is present it's probably fine.

Dedicated soundcards - are they worth it?

Reply #3 –
RMAA Results Realtek ALC899 (using regular cheap 2m cable)

doesnt seem to be so bad.

As for popping and the likes: Didnt hear anything like that since I got my new motherboard.

Dedicated soundcards - are they worth it?

Reply #4 –
It mostly depends on the mainboard. I've had a board with an ALC899 that had horrible interference noise from pretty much any component in the PC. Replaced it by another brand with same components and it's perfectly quiet.
And that was the only difference I heard between the onboard and a dedicated card when all driver-side enhancements are off. Otherwise those enhancements might color the sound.

Dedicated soundcards - are they worth it?

Reply #5 –
Quote from: Ceremony64 on
RMAA Results Realtek ALC899 (using regular cheap 2m cable)  doesnt seem to be so bad.  As for popping and the likes: Didnt hear anything like that since I got my new motherboard.


24/48 ... is 48kHz native SR for this codec? ... how's the more common mode 16/44.1?

Dedicated soundcards - are they worth it?

Reply #6 –
Realteks are getting better and better.

What you need to look at is the output power it can deliver and it output impedance.

Dedicated soundcards - are they worth it?

Reply #7 –
Quote from: ChronoSphere on
It mostly depends on the mainboard. I've had a board with an ALC899 that had horrible interference noise from pretty much any component in the PC. Replaced it by another brand with same components and it's perfectly quiet.
And that was the only difference I heard between the onboard and a dedicated card when all driver-side enhancements are off. Otherwise those enhancements might color the sound.

My mobo is an ASRock Z77 Professional-M. Guess its doing a good job in that aspect!

as for 16/44.1:
http://storage.superconnected.org/random/R...ekALC899v2.html

P.S. id never use those "enhancements". might as well buy beats if I wanted those >_<

Dedicated soundcards - are they worth it?

Reply #8 –
Quote from: Ceremony64 on
Quote from: ChronoSphere on
It mostly depends on the mainboard. I've had a board with an ALC899 that had horrible interference noise from pretty much any component in the PC. Replaced it by another brand with same components and it's perfectly quiet.
And that was the only difference I heard between the onboard and a dedicated card when all driver-side enhancements are off. Otherwise those enhancements might color the sound.

My mobo is an ASRock Z77 Professional-M. Guess its doing a good job in that aspect!

as for 16/44.1:
http://storage.superconnected.org/random/R...ekALC899v2.html

P.S. id never use those "enhancements". might as well buy beats if I wanted those >_<


I don't like that channel imbalance in the FR graph. Did you play with L/R ?

Dedicated soundcards - are they worth it?

Reply #9 –
Quote from: Aldem on
Quote from: Ceremony64 on
Quote from: ChronoSphere on
It mostly depends on the mainboard. I've had a board with an ALC899 that had horrible interference noise from pretty much any component in the PC. Replaced it by another brand with same components and it's perfectly quiet.
And that was the only difference I heard between the onboard and a dedicated card when all driver-side enhancements are off. Otherwise those enhancements might color the sound.

My mobo is an ASRock Z77 Professional-M. Guess its doing a good job in that aspect!

as for 16/44.1:
http://storage.superconnected.org/random/R...ekALC899v2.html

P.S. id never use those "enhancements". might as well buy beats if I wanted those >_<


I don't like that channel imbalance in the FR graph. Did you play with L/R ?
no i did not. Though it might be the cable to blame here. It is merely 0.5db difference. and even if its not the cable: as its constant, it can be corrected.

Also, on tuesday, I'll try different input/output ports, as those are freely assignable. I wonder if they make a difference!

Dedicated soundcards - are they worth it?

Reply #10 –
The mismatch could also be the A/D since hes testing in loop back.  I get the impression that the line in on a lot of onboard systems is an afterthought.

Output impedance is a good point though with some headphones.  Then again, plenty of highend sound cards also have terrible headphone out.

Dedicated soundcards - are they worth it?

Reply #11 –
Quote from: Ceremony64 on
RMAA Results Realtek ALC899 (using regular cheap 2m cable)

doesn't seem to be so bad.


I agree with the others who make cautious comments about the +0.5 - 0.5 dB channel level mismatch. This can affect soundstaging.

The channels on even the cheapest onboard audio interface should match quite well - the Realtek ALC887  chip on this Asus M5A97  motherboard I'm typing on has channel balance within 0.1 dB per an audio rightmark test.

The cables are cheap and I had to insert them several times before the channels matched at all, which is probably a critical reflection on the connectors on the cable or more likely the connectors on the back of the PC.

The performance of this audio interface is almost identically the same whether in 16/44 or 24/96 mode, which tells me that it is probably a 16/44 or 16/48 physical device front and back ended with software resamplers in the device driver.

Quote
As for popping and the likes: Didn't hear anything like that since I got my new motherboard.


Good. Most of such things are due to data loss which in turn is due to issues with driver buffering or system board internal data path contention.

You can double check for this problem by using the same cable loop as you used for the Rightmark testing, using Audacity's multitrack feature to re-record a 100 Hz tone (which you can also make with Audacity).

Zoom in on the re-recorded wave closely enough to see the individual sine waves and visually scan 10-30 seconds of re-recorded data. If there are clicks and pops they will be clearly visible as occasion, usually random brief instances of waveform distortion.

Dedicated soundcards - are they worth it?

Reply #12 –
Quote from: saratoga on
The mismatch could also be the A/D since hes testing in loop back.  I get the impression that the line in on a lot of onboard systems is an afterthought.


I can confirm that a high percentage of onboard audio interfaces have significantly asymmetrical design and performance.  C-Media chips show in my tests this way more often than Realtek.

The degraded performance for the record size generally takes the form of reduced dynamic range and higher distortion. The playback side might have 85 dB dynamic range while the record side may be more like 75 dB. You need to run your Rightmark tests with a second audio interface that has known high performance to verify this.

However, I've never seen this aymmetry  bad enough to affect record levels.

Quote
Output impedance is a good point though with some headphones.  Then again, plenty of highend sound cards also have terrible headphone out.


Agreed. As a class of devices portable digital players seem to be more likely to provide a really low source impedance.  Name an AVR or even pro audio interface whose source impedance can compare with the excellent $29.95 Sansa Clip?  I can't.

Dedicated soundcards - are they worth it?

Reply #13 –
Today I assigned line out to all but one connection of the ALC899 chip and tested them one by one (loop back to the same line in though!). The differences were minimal (44.1@16bit), with a tiny improvement on FR (+0.01, -0.04) and closer level change (0.0 dB / 0.4 dB). Guess thats my output port from now on

I also used a shorter(50cm vs 2m) and slightly higher quality cable which didnt improve measurements on the previous I/O ports. Guess I need 500€ placebo cables for that :3

As for the sine wave test: Couldnt get to it yet. Maybe tomorrow!

Dedicated soundcards - are they worth it?

Reply #16 –
No AVR, no pro audio interface, but.....NwAvGuy measured an output impedance of 0.13Ohm for the affordable Fiio E7.

Dedicated soundcards - are they worth it?

Reply #17 –
Quote from: skamp on
Quote from: Arnold B. Krueger on
Name an AVR or even pro audio interface whose source impedance can compare with the excellent $29.95 Sansa Clip?  I can't.


Audiotrak DR. DAC 3 is rated at 0?, though I don't own a unit to verify that.


That appears to be an exceptional claim. If they said less than 1 ohm or perhaps even less than 0.1 ohm then it might be more believable. 

Of course any  comparison between a $340+ DR DAC and a $29.95 portable digital player has more than a few amusing properties of its own...

I believe that the *Magic* headphone amp in the Sansa products is just a minor subfunction of a < $10 chip whose main purpose in life is being a stand-alone computer with nontrivial processing power and interfaces.

Dedicated soundcards - are they worth it?

Reply #18 –
Quote from: Arnold B. Krueger on
Of course any  comparison between a $340+ DR DAC and a $29.95 portable digital player has more than a few amusing properties of its own...


Yup. I didn't originally buy it for this, but it turns out that chaining my €30 Topping NX1 to whatever is a cheap and easy fix. With its incredible battery life, I even use it without a USB tether for power and recharge it once a month, or less.

Quote from: Arnold B. Krueger on
I believe that the *Magic* headphone amp in the Sansa products is just a minor subfunction of a < $10 chip whose main purpose in life is being a stand-alone computer with nontrivial processing power and interfaces.


The "magic" in the NX1 is also a single (albeit dedicated) wunder-chip: MAX9722

Dedicated soundcards - are they worth it?

Reply #19 –
Quote from: skamp on
Quote from: Arnold B. Krueger on
Of course any  comparison between a $340+ DR DAC and a $29.95 portable digital player has more than a few amusing properties of its own...


Yup. I didn't originally buy it for this, but it turns out that chaining my €30 Topping NX1 to whatever is a cheap and easy fix. With its incredible battery life, I even use it without a USB tether for power and recharge it once a month, or less.

Quote from: Arnold B. Krueger on
I believe that the *Magic* headphone amp in the Sansa products is just a minor subfunction of a < $10 chip whose main purpose in life is being a stand-alone computer with nontrivial processing power and interfaces.


The "magic" in the NX1 is also a single (albeit dedicated) wunder-chip: MAX9722


And its secret is that it is both a negative power supply generator and an op amp.  This differs from the Sansa approach which seems to be to create an active elevated ground reference for the headphone outputs.
 

Dedicated soundcards - are they worth it?

Reply #20 –
In case if anyone's interested, I've made some measurements with the two onboard Realtek codecs I have within my reach, ALC861 and ALC888. Both are rather old, the first board is from ca. 2005, and the other one is from 2008.

ALC888 (as well as all newer Realtek codecs, I presume) has built-in headphone buffer amplifier on each output, which should be automatically engaged when the output function is set to "Headphone" in the driver control panel.

Here are the output voltages measured with various resistive loads in "Line out" mode and in "Headphone" mode:

Code: [Select]
ALC888 on Gigabyte P35-DS3L
1 kHz sine wave, Line out (back panel green)

Load (Ohm)  Output voltage (VRMS)  Calculated output impedance
            Left     Right         Left     Right

unloaded    0.924    0.916         -        -
2000        0.852    0.844         169.0    170.6
1000        0.788    0.781         172.6    172.9
270         0.564    0.560         172.3    171.6
200         0.491    0.489         176.4    174.6
100         0.338    0.337         173.4    171.8
62          0.242    0.241         174.7    173.6
51          0.206    0.206         177.8    175.8
39          0.171    0.170         171.7    171.1
24          0.113    0.112         172.2    172.3
10          0.050    0.050         174.8    173.2

Output impedance (least-squares fit): 174.1 Ohm, 173.0 Ohm


1 kHz sine wave, Headphone out (back panel green)

Load (Ohm)  Output voltage (VRMS)  Calculated output impedance
            Left     Right         Left     Right

unloaded    0.935    0.926         -        -
2000        0.933    0.926         4.3      0.0
1000        0.929    0.925         6.5      1.1
270         0.912    0.918         6.8      2.3 
200         0.900    0.914         7.8      2.6
100         0.863    0.895         8.3      3.5
62          0.815    0.864         9.1      4.4
51          0.790    0.843         9.4      5.0
39          0.752    0.810         9.5      5.6
24          0.668    0.710         9.6      7.3
10          0.416    0.439         12.5     11.1


I could not achieve 1 Volt RMS with ALC888 even with all volume controls set to maximum, although the datasheet claims that the "typical" full-scale unloaded output should be around 1.4 VRMS (with ALC861 it is indeed about 1.4 VRMS, or 2 Volts peak; I had to adjust the master volume down to get 1 VRMS). I don't know whether it's a hardware or a software issue (I only tested one of the ALC888 outputs). Also note the channel imbalance with the amplified output.


And here are the measurement results for ALC861:

Code: [Select]
ALC861 on Biostar 945P-A7B
1 kHz sine wave, Line out (back panel green)

Load (Ohm)  Output voltage (VRMS)  Calculated output impedance
            Left     Right         Left     Right

unloaded    1.039    1.030         -        -
2000        0.960    0.952         164.6    163.9
1000        0.890    0.883         167.4    166.5
270         0.644    0.639         165.6    165.2
200         0.563    0.560         169.1    167.9
100         0.391    0.389         165.7    164.8
62          0.278    0.277         169.7    168.5
51          0.236    0.236         173.5    171.6
39          0.194    0.194         169.9    168.1
24          0.126    0.126         173.9    172.2
10          0.054    0.055         182.4    177.3

Output impedance (least-squares fit): 168.6 Ohm, 167.4 Ohm


1 kHz sine wave, Line out (back panel blue)

Load (Ohm)  Output voltage (VRMS)  Calculated output impedance
            Left     Right         Left     Right

unloaded    1.042    1.036         -        -
2000        0.955    0.948         182.2    185.6
1000        0.877    0.872         188.1    188.1
270         0.608    0.605         192.7    192.3
200         0.524    0.521         197.7    197.7
100         0.350    0.350         197.7    196.0
62          0.245    0.245         201.7    200.2
51          0.208    0.208         204.5    203.0
39          0.171    0.170         198.6    198.7
24          0.111    0.111         201.3    200.0
10          0.048    0.047         207.1    210.4

Output impedance (least-squares fit): 197.0 Ohm, 196.4 Ohm


I have tested two outputs (green and blue connectors). The function for green connector can not be changed, it's always "Line out". The blue connector can be switched to "Headphone", but this does not affect the output voltage at all. Again, I'm not sure if it's the hardware or a software issue. Looking at ALC861 datasheet it seems that three out of four outputs should be amplified.

Dedicated soundcards - are they worth it?

Reply #21 –
Quote from: alexeysp on
I could not achieve 1 Volt RMS with ALC888 even with all volume controls set to maximum, although the datasheet claims that the "typical" full-scale unloaded output should be around 1.4 VRMS (with ALC861 it is indeed about 1.4 VRMS, or 2 Volts peak; I had to adjust the master volume down to get 1 VRMS). I don't know whether it's a hardware or a software issue (I only tested one of the ALC888 outputs). Also note the channel imbalance with the amplified output.


DACs don't generate their own voltages, they're more like OpAmps.  You provide them with a voltage and they allow you to output a signal within that voltage range minus some loss through a transistor.  If the specsheet says 1.4 Vrms maximum, it means that if you provide the maximum rated supply voltage and operate at max output, you will get that level.  If you use a lower supply voltage, you will get proportionally less voltage output.

1.4 Vrms is 3.960 Vpp.  Account for 2x junctions at ~0.5v drop and you have a 5 volt supply.  Assume a 3.3 v supply, and you get 0.8 Vrms, very close to what you observe.  Most likely your motherboard runs the chip off the 3.3v supply.  This is much more common as most digital logic has moved to lower voltages than 5v.

Dedicated soundcards - are they worth it?

Reply #22 –
Quote from: saratoga on
Assume a 3.3 v supply, and you get 0.8 Vrms, very close to what you observe.  Most likely your motherboard runs the chip off the 3.3v supply.  This is much more common as most digital logic has moved to lower voltages than 5v.


Like most mixed signal chips these codecs have separate digital and analog power supply pins. The logic is 3.3 V only, but the analog section can be supplied by anything from 3.3 to 5 V. I just assumed it should be 5 V for both boards, but you're probably right, the Gigabyte board likely uses 3.3 V for both digital and analog rails. I'm just too lazy to open up the case and check it

Dedicated soundcards - are they worth it?

Reply #23 –
I would not be surprised to see maximum output remain 3 dB below what it could be when not using WASAPI in exclusive mode or similar; at least that's what I saw on this Vista machine. 0.935 V + 3 dB would be about 1.3 V.

The ALC898 on my SB Audigy FX has about 0.2 dB of output channel imbalance, too, so maybe there's some manufacturing issues with Realtek's higher-end chips at times?

So it's a clever little amplifier IC that makes the NX1 such good bang/buck. Certainly somewhat more high-tech than ye old BTech BT928 a decade ago...

Dedicated soundcards - are they worth it?

Reply #24 –
Quote from: stephan_g on
I would not be surprised to see maximum output remain 3 dB below what it could be when not using WASAPI in exclusive mode or similar; at least that's what I saw on this Vista machine. 0.935 V + 3 dB would be about 1.3 V.


How would reasonably well written software compromise the peak undistorted output of an audio interface?