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 13516 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Dedicated soundcards - are they worth it?

Reply #25 –
max (output level) = min (analog clipping level, max digital level @ max gain)

I was assuming that the results are limited by the latter factor here; analog PGAs seem to have gone a bit out of fashion lately. Of course as long as analog gain can be increased to make up for lower digital levels (like my Clip+ can), you'd be right. Excess gain reduces dynamic range bragging rights though, so 0 dBFS at full volume tends to be placed somewhere near analog clipping level - at least that has usually been the case in my limited experience with PC consumer audio (I do know one exception, which is the FiiO E10 in high gain mode).

alexeysp never said anything about maximum undistorted output, just maximum output full stop. Which I would be inclined to be taking literally - output with the volume cranked up all the way. If in doubt, some distortion analysis of test sines may give hints regarding how close to analog clipping we are; the 3rd harmonic tends to show a marked increase when approaching clipping, while the same would not be expected if we're just hitting maximum digital level.

Dedicated soundcards - are they worth it?

Reply #26 –
Quote from: stephan_g on
I was assuming that the results are limited by the latter factor here; analog PGAs seem to have gone a bit out of fashion lately.


FWIW I don't think I've ever seen a PC part without one.  Its expected by the Intel HD audio spec, so most likely any motherboard will have one.

Sometimes parts designed for stand alone electronics don't have them though (e.g. some amplifiers/receivers), mostly because you wouldn't use one in those systems anyway.

Dedicated soundcards - are they worth it?

Reply #27 –
Quote from: stephan_g on
alexeysp never said anything about maximum undistorted output, just maximum output full stop.


This test was primarily to evaluate the output impedance. I was aiming for 1 VRMS unloaded as a reference level, and I can tell that for ALC861 it was seemingly distortion-free (confirmed with the oscilloscope reading; the visible distortion appears with the 51 Ohm and lower loads). I can't state the same for ALC888 though, as I didn't perform any additional checks for distortion with this codec, and the datasheet says that the volume control goes to +42 dB, so it's possible that the output is clipped at maximum volume settings. I'll try to do some RMAA tests to complement the previous measurements. Will post the results later.

 

Dedicated soundcards - are they worth it?

Reply #28 –
I've finished RMAA testing with ALC888 and ALC861. Seven loopback tests were done for ALC888: line out and headphone out to line in, unloaded, 1000 Ohm loaded, and 100 Ohm loaded, plus one additional test of the headphone out with the actual headphones (ATH-A700, 65 Ohm). Three tests were done for ALC861 (line out unloaded, with 1000 Ohm load and with 100 Ohm load). The html reports are posted in the uploads forum.

Built-in headphone amplifier in ALC888 improves on the crosstalk and distortion with low impedance loads, and does not seem to add any significant noise (the noise figures obtained in the tests are likely due to the ADC noise floor). However, the low frequency roll-off with low impedance loads is quite atrocious, for either line out or headphone out. Supposedly this is due to the small output coupling capacitance. The roll-off in ALC861 tests is much less noticeable.

Regarding the voltage levels, I had to set both master volume and line input level controls below 50% to avoid clipping (the input level control is very sensitive in both codecs). This resulted in output voltage of ~0.33 VRMS for ALC888 and ~0.7 VRMS for ALC861. With the loaded line out tests I had to adjust the master volume (and the line input gain in case of 100 Ohm load) to increase the signal to approximately match the initial level. When testing the ALC888 headphone out the volume control settings were identical for all tests. The actual measured voltages are as follows:

Code: [Select]
ALC888 on Gigabyte P35-DS3L
Output voltages (VRMS)
line out          headphone out
left    right     left    right
unloaded          unloaded
0.329   0.325     0.333   0.330
                  
1 kOhm            1 kOhm
0.330   0.326     0.331   0.328
                  
100 Ohm           100 Ohm
0.241   0.236     0.313   0.311


Code: [Select]
ALC861 on Biostar 945P-A7B
Output voltages (VRMS)
line out
left    right
unloaded
0.715   0.708

1 kOhm loaded
0.777   0.768

100 Ohm loaded
0.420   0.416

Dedicated soundcards - are they worth it?

Reply #29 –
Quote from: alexeysp on
Code: [Select]
ALC888 on Gigabyte P35-DS3L
Output voltages (VRMS)
line out          headphone out
left    right     left    right
unloaded          unloaded
0.329   0.325     0.333   0.330
                  
1 kOhm            1 kOhm
0.330   0.326     0.331   0.328
                  
100 Ohm           100 Ohm
0.241   0.236     0.313   0.311


Code: [Select]
ALC861 on Biostar 945P-A7B
Output voltages (VRMS)
line out
left    right
unloaded
0.715   0.708

1 kOhm loaded
0.777   0.768

100 Ohm loaded
0.420   0.416


My little source impedance calculator analyzes the above as:

Code: [Select]
ALC888 on Gigabyte P35-DS3L            
V(open)    V(load)    R(load)    Source Impedance
0.333    0.313    100.00    6.39
0.330    0.311    100.00    6.11
            
ALC861 on Biostar 945P-A7B            
V(open)    V(load)    R(load)    Source Impedance
0.715    0.420    100.00    70.24
0.708    0.416    100.00    70.19

Dedicated soundcards - are they worth it?

Reply #30 –
Quote from: Arnold B. Krueger on
My little source impedance calculator analyzes the above as:
[skip]


Arnold, your calculation for ALC861 does not apply, since the output voltage with the 100 Ohm load is affected by the adjustments I made during the RMAA test. As I wrote in my previous post, when testing the loaded line outs, I had to adjust both the master volume (output gain) and the recording level (input gain) to obtain the recorded signal of sufficiently high amplitude and minimum distortion at the same time. The actual line out impedance of ALC861 is about 170 Ohm, as can be seen in my earlier measurement results.

Your calculation for the ALC888 headphone out is correct though, since I did not perform any volume adjustments when testing the output in headphone (buffered) mode.

Dedicated soundcards - are they worth it?

Reply #31 –
Quote from: alexeysp on
Built-in headphone amplifier in ALC888 improves on the crosstalk and distortion with low impedance loads, and does not seem to add any significant noise (the noise figures obtained in the tests are likely due to the ADC noise floor). However, the low frequency roll-off with low impedance loads is quite atrocious, for either line out or headphone out. Supposedly this is due to the small output coupling capacitance. The roll-off in ALC861 tests is much less noticeable.

You can blame this on a cheapskate board manufacturer, as coupling cap size ultimately is their choice. (It's interesting that this would happen on a middle-class Gigabyte board, given that they've had some pretty competent onboard audio implementations. [1]) Realtek datasheets normally suggest 100µ coupling caps for headphone outputs, which with a 100 ohm load + 6 ohms of output impedance should give -3 dB at 15 Hz and -1 dB at ~30 Hz.

It's interesting that you had to turn down rec levels that far for the '888. Could it be that you used the mic in (the red one) with about 10 dB of preamplification still enabled? A similar-vintage ALC262 does offer this setting. In Vista/7, it's in recording device properties under the input level slider. It's not present for the line in (the blue one).

[1] BTW: The P35-DS3 still got a fast-booting AHCI BIOS (1.20 in BIOS F14) - I'm so jealous. My same-generation (iQ35) machine here has to make do with 1.06 (which subjectively takes ages to detect its stuff) and uses a proprietary BIOS that is virtually impossible to mod. Grrr.

PS: Sorry for the late bump, been a bit busy @work lately.

Dedicated soundcards - are they worth it?

Reply #32 –
Quote from: stephan_g on
Quote from: alexeysp on
Built-in headphone amplifier in ALC888 improves on the crosstalk and distortion with low impedance loads, and does not seem to add any significant noise (the noise figures obtained in the tests are likely due to the ADC noise floor). However, the low frequency roll-off with low impedance loads is quite atrocious, for either line out or headphone out. Supposedly this is due to the small output coupling capacitance. The roll-off in ALC861 tests is much less noticeable.

You can blame this on a cheapskate board manufacturer, as coupling cap size ultimately is their choice. (It's interesting that this would happen on a middle-class Gigabyte board, given that they've had some pretty competent onboard audio implementations. [1]) Realtek datasheets normally suggest 100µ coupling caps for headphone outputs, which with a 100 ohm load + 6 ohms of output impedance should give -3 dB at 15 Hz and -1 dB at ~30 Hz.

It's interesting that you had to turn down rec levels that far for the '888. Could it be that you used the mic in (the red one) with about 10 dB of preamplification still enabled? A similar-vintage ALC262 does offer this setting. In Vista/7, it's in recording device properties under the input level slider. It's not present for the line in (the blue one).


The way to make a lot of this moot is to simply add a inexpensive high performance headphone amp such as the Toppinhg NX-1.  It seems to present an easy load for just about any line or headphone output.

Dedicated soundcards - are they worth it?

Reply #33 –
Just make sure you RMAA the whole chain (ALC889 line out->amp->Y splitter->HD518+ALC889 line in). It might differ from the ALC889-only measurement a lot.

Dedicated soundcards - are they worth it?

Reply #34 –
Quote from: Martel on
Just make sure you RMAA the whole chain (ALC889 line out->amp->Y splitter->HD518+ALC889 line in). It might differ from the ALC889-only measurement a lot.


But beware of the fact that many audio interfaces have asymmetrical performance - their playback sides perform far better than their record sides. Loop back testing can provide deceptively poor results when its the playback side that most are far more interested in.

Dedicated soundcards - are they worth it?

Reply #35 –
Quote from: stephan_g on
It's interesting that you had to turn down rec levels that far for the '888. Could it be that you used the mic in (the red one) with about 10 dB of preamplification still enabled?


No, I'm pretty sure the input was configured as line in, without the "boost" option. With the voltage levels used in this test, I had to set the input level slider position in the control panel close to the bottom end in order to avoid clipped recording.  The relative master volume and input level settings were similar for both boards, but the board with ALC861 provides higher absolute voltage. These settings did not, however, exactly correspond to the ADC full scale, there was still some headroom left, about 2 dB according to the RMAA level indicator.