GSMArena Updated Its Audio Quality Tests

[jensigner]

BTW how do you measure the earbud impedance?

For most common earbuds and phones which are mainly resistive at the standard test frequency of 1 kHz, it is pretty simple: Set up a simple voltage divider circuit (one for each channel if you want) with a single extra resistor Rs in series with the buds R_hs, drive the series circuit with a SMALL SIGNAL sine wave at 1 kHz, say ~ 50mV for buds, using a function generator and measure the output voltage compared to the input voltage and apply simple voltage-divider Kirchoff law. The series resistor value is not critical ... just pick a value in the general area of the headphone impedance for a more accurate calculation. I use a USB scope (Digilent Discovery) for this and just measure the peak voltages but you could also just use a DVM (since most DVMs have a reasonable response at 1 kHz). A scope lets you monitor the signal quality (to make sure you aren't clipping or causing other distortion) and also allows you to see any non-resistive components of the earbuds which shows as a phase-shift in the two scope signals V3 and V2. Note that the internal resistance (Rint) of the function generator doesn't enter into the calculation:

 

[jensigner]

Here are some preliminary RMAA measurements on my (Canadian) S4. Firstly I should say that irrespective of any technical measurements, I think the S4 headphone-out sounds amazing and I am very pleased with it. I'm showing my measurements for discussion purposes and to try to understand why there are significant differences to those published at GSMArena as noted near the start of this thread. My results below show (a) the RMAA results for the entire measurement chain without the S4 but using a test-signal generated by the sound-card line-out (left column) and (b) the RMAA results (right column) using the S4 to playback the same 16bit/44.1 kHz test wav content . The stereo crosstalk with the S4 is of concern. Also my RMAA result for THD with the S4 is 0.033% or 10 times higher than the GSM results assuming those are % figures. The IMD+Noise and frequency response is comparable. To compare with the GSM results it would be useful to know the S4 Play level that was used (I used 7/15). I noticed that during idle (no playback) the overall spectral noise floor as measured by the RMAA spectrum analyzer is considerably lower ... but when a pure tone is played, the entire spectral noise floor is raised by close to 10 dB. It appears that the playback process raises the noise floor!

 

[Reuben Chew]

Here are some preliminary RMAA measurements on my (Canadian) S4. Firstly I should say that irrespective of any technical measurements, I think the S4 headphone-out sounds amazing and I am very pleased with it. I'm showing my measurements for discussion purposes and to try to understand why there are significant differences to those published at GSMArena as noted near the start of this thread. My results below show (a) the RMAA results for the entire measurement chain without the S4 but using a test-signal generated by the sound-card line-out (left column) and (b) the RMAA results (right column) using the S4 to playback the same 16bit/44.1 kHz test wav content . The stereo crosstalk with the S4 is of concern. Also my RMAA result for THD with the S4 is 0.033% or 10 times higher than the GSM results assuming those are % figures. The IMD+Noise and frequency response is comparable. To compare with the GSM results it would be useful to know the S4 Play level that was used (I used 7/15). I noticed that during idle (no playback) the overall spectral noise floor as measured by the RMAA spectrum analyzer is considerably lower ... but when a pure tone is played, the entire spectral noise floor is raised by close to 10 dB. It appears that the playback process raises the noise floor!
http://www.jensign.com/GalaxyS4/RMAA_S4.jpg

Thanks for the analysis! I was wondering if you could do a test with low-impedance earphones? It seems that there's a known audio clipping/distortion issue that low-impedance earphones users are facing, at the bass range. More info could be found here: Sound crackling/distortion with certain headphones? - xda-developers

I was wondering if you could shed any light on the issue and why there may such a problem with the S4!

 

[jensigner]

Thanks for the analysis! I was wondering if you could do a test with low-impedance earphones? It seems that there's a known audio clipping/distortion issue that low-impedance earphones users are facing, at the bass range. More info could be found here: Sound crackling/distortion with certain headphones? - xda-developers

I was wondering if you could shed any light on the issue and why there may such a problem with the S4!

For reference, my S4 is a SGH-I337M (Canadian model ... same setup as US model I believe) with Qualcomm Snapdragon 600 Soc and Qualcomm Krait 300 CPU.
After quickly scanning that xda discussion, I have just posted there to see if anyone has made quantitative measurements.
It looks like the issues are all related to low-impedance (< 14ohm?) headphones and drive level particularly at lower frequencies but sure, I can check that using a dummy resistor load of say 12 ohm. I don't have really low impedance headphones, so I could just place a 12 ohm shunt resistor across my 32 ohm (stock) earbuds and just listen which is subjective. Since I also have access to an S3, I can do swap/compare tests. Note that earlier in this posting I reported measurements of the S3 output impedance as ~ 3ohm. Also I tested the output of the S3 WITH 10 ohm loads at 440 Hz and distortion only occurred at the really high level of 14/15. (Haven't tested the S4 yet).
To quantify things, I would just look at the headphone output on a scope and playback some pure sine wave tracks at various frequencies (say 60Hz, 1kHz, etc.). I could also do a more complete RMAA analysis as above, but first I need to understand why my RMAA measurements differ from GSMArena results (they use full audio volume for their tests) and what the problem referred to in the review below under "Update 11 April ..." means (I have also sent this question to GSMArena directly):
Samsung Galaxy S4 review: Supernova - GSMArena.com

 

[Reuben Chew]

For reference, my S4 is a SGH-I337M (Canadian model ... same setup as US model I believe) with Qualcomm Snapdragon 600 Soc and Qualcomm Krait 300 CPU.
After quickly scanning that xda discussion, I have just posted there to see if anyone has made quantitative measurements.
It looks like the issues are all related to low-impedance (< 14ohm?) headphones and drive level particularly at lower frequencies but sure, I can check that using a dummy resistor load of say 12 ohm. I don't have really low impedance headphones, so I could just place a 12 ohm shunt resistor across my 32 ohm (stock) earbuds and just listen which is subjective. Since I also have access to an S3, I can do swap/compare tests. Note that earlier in this posting I reported measurements of the S3 output impedance as ~ 3ohm. Also I tested the output of the S3 WITH 10 ohm loads at 440 Hz and distortion only occurred at the really high level of 14/15. (Haven't tested the S4 yet).
To quantify things, I would just look at the headphone output on a scope and playback some pure sine wave tracks at various frequencies (say 60Hz, 1kHz, etc.). I could also do a more complete RMAA analysis as above, but first I need to understand why my RMAA measurements differ from GSMArena results (they use full audio volume for their tests) and what the problem referred to in the review below under "Update 11 April ..." means (I have also sent this question to GSMArena directly):
Samsung Galaxy S4 review: Supernova - GSMArena.com

Hey I'm actually 'trenzterra' on xda-developers, but I thought I'd reply you here instead...

Okay the thing is I don't have the necessary equipment (and I'm not an engineer by any means so...) to do such an analysis. The only thing I have is a Sound Blaster Z sound card and a line-in cable, but I think the sound card itself adds some distortion so my RMAA analysis wouldn't be that relevant (though there doesn't seem to be any noticeable distortion in the test wave generated). I've no idea how to test it with earphones attached with the current equipment I have.

I guess it would be really great if you could try to replicate the performance with low impedance earphones. Maybe then will Samsung be convinced to do something. The affected frequencies are mostly in the below 1kHz range. If I EQ the bass range down there's no more distortion.

P.S. BTW I won't be too hopeful of GSMArena replying. I emailed them one week ago on their opinion of this problem and I haven't received even a single acknowledgement from them.

 

[jensigner]

I have completed some initial tests of the S4 and S3 using different resistive loads (10ohm-33ohm) and checking the output for different frequencies and different Player volume levels. My results CLEARLY demonstrate an oscillation "artifact" for headphone loads less than ~ 14 ohm. Direct comparison with the S3 SHOWED NO SUCH PROBLEMS. The details of my measurements (which anyone can repeat) are given below. Feel free to pass on to Samsung folks
Samsung Galaxy S4: Problem With Low-Impedance Headphones

 

[jensigner]

P.S. BTW I won't be too hopeful of GSMArena replying. I emailed them one week ago on their opinion of this problem and I haven't received even a single acknowledgement from them.

I haven't received feedback either. I have done quite a few RMAA tests over the past several years so I have some experience. If I can't reproduce the GSMArena RMAA results for the S4 them i'm inclined to not believe their results, or that their testing configuration doesn't really reflect the audio quality heard when using standard earphones. If they can't provide the details of the problem that should be viewed as suspicious

 

[Reuben Chew]

I have completed some initial tests of the S4 and S3 using different resistive loads (10ohm-33ohm) and checking the output for different frequencies and different Player volume levels. My results CLEARLY demonstrate an oscillation "artifact" for headphone loads less than ~ 14 ohm. Direct comparison with the S3 SHOWED NO SUCH PROBLEMS. The details of my measurements (which anyone can repeat) are given below. Feel free to pass on to Samsung folks
Samsung Galaxy S4: Problem With Low-Impedance Headphones

Wow, thank you so much for your in-depth analysis! The strange thing, however, is that most of us experience something slightly different: There is distortion starting from a certain volume level, which gets more severe as the volume increases. For my case, at volume 15/15, the problem is most noticeable while at 11/15, the problem is much less noticeable.

Also, the problem seems to manifest itself more when playing a real audio track rather than a test tone. When playing a test tone, I get a 'noise' at the start of the track, but that's about it. I wonder if the sudden change in frequencies could be a contributing factor in this?

EDIT: I just re-read your article again. So could it be that in songs, where the volume in certain parts of the track varies, when it drops down to the magical 11/15 it experiences this problem?

In this case, do you think this problem could be suitably addressed via software? By the way, it would be great if you could forward your test results to prominent tech sites so that Samsung will be pressured to do anything. I've contacted some over the past few days, but not one of them even bothered to give me a reply. Kinda makes me think Samsung paid them off or something! With your objective test results I am more hopeful though.


EDIT2: Also, with even lower impedance earphones (Sony XBA series for example, which are generally 8ohms and below), some have reported the distortion at even lower volumes, such as at the 30% level. I was wondering if you had the ability to conduct tests with an even lower impedance load, and see if the volume level where this happens changes. Thank you!

 

[jensigner]

EDIT: I just re-read your article again. So could it be that in songs, where the volume in certain parts of the track varies, when it drops down to the magical 11/15 it experiences this problem?

That is what I'm thinking. Perhaps the problems noted are not the same but they may well be related. I can offer to post one of the test-tones I use and others can verify what I observe. Any music will of course be complex and have varying frequency components at time-varying amplitudes which will cause the oscillation above to manifest it in a fairly complex manner. So for example a typical sound track played at full volume (15/15) would have a LOT of components in the 11/15 amplitude range so you'd expect a lot of that oscillation . A music track played back at amplitude 11/15 would have most of its amplitude component at lower effective amplitudes so would show little oscillation. I can check the problem at lower resistances say at 4 and 8 ohm load.

 

[Patrik_swe]

I have S4 and tried stock earbuds with eartips from Yamaha EPH-100. With these eartips (double flange) the Samsung earbuds sound much better.

EPH-100 - Headphones/Earphones - Audio & Visual - Products - Yamaha United States

I lost the small plastic bag with original eartips and I only have the pair mounted from factory. With these sound is thin with no bass, low volume even at max 15.

In my opinion Samsung earbuds sounds good sound if you get good seal.

When using Yamaha EPH-100 earbuds with S4 I have problem with crackling audio at higher volumes, over 10. I usually never listen higher than 7-9 and this issue does not affect me so much with Yamaha EPH-100.

EDIT: I have tested Yamaha EPH-100 more and I hear crackling now and then at volume 7-9 depending which song I play.

I have a pair of SoungMagic E10 and they work at higher volume. Other earphones "A-Jays One+" have crackling audio at higher volume.

EDIT:

I have tested SoundMagic E10 again and they also have problem with crackling but only at highest volume 14-15. Since they are senistive I never play that loud, more like max 10.

I mailed Samsung support in Sweden and they replied like this (translated through Google translate):

"Hi Patrik

Thank you for your e-mail with Samsung Electronics Nordic.


Sorry to hear that you have problems with your headphones. However, the fact is that it really is only Samsung Original which we can guarantee will work, but in most cases, it works with almost all headphones. There has recently been a minor update to the Galaxy S4.
Unfortunately, so as long as your original headphones works so there is nothing wrong with the phone itself. Should you have other headphones, you get partial experiment and see which ones work, as there is no general rule for what works / does not work.

Please contact us again if you have more questions.

Sincerely

Samsung Support"

They write about update but I have installed it but this will not fix audio issue.

They also write "but in most cases, it works with almost all headphones". Strange, of my three pair none works 100%. One of them can be described as "high- end" (Yamaha EPH-100) but the rest are low cost that common people might buy and have issues with.

I have mailed them again and asked them if they know why this crackling occurs and several other questions but didn't got answer (as I thought).

I don't think Samsung support understand this problem.

 

[Reuben Chew]

That is what I'm thinking. Perhaps the problems noted are not the same but they may well be related. I can offer to post one of the test-tones I use and others can verify what I observe. Any music will of course be complex and have varying frequency components at time-varying amplitudes which will cause the oscillation above to manifest it in a fairly complex manner. So for example a typical sound track played at full volume (15/15) would have a LOT of components in the 11/15 amplitude range so you'd expect a lot of that oscillation . A music track played back at amplitude 11/15 would have most of its amplitude component at lower effective amplitudes so would show little oscillation. I can check the problem at lower resistances say at 4 and 8 ohm load.

Hey I got back my phone from the service centre, and yep the same problem still happens. It doesn't seem as apparent this time, however, but I decided to load a 150Hz test tone on the phone, and it crackled ONLY at volume 13. So yes, I can corroborate your findings, though it happens at a different volume level.

 

[NZtechfreak]

I already posted this at Head-Fi, but here is my experience with this:

Did a blind and volume-matched test with the One and S4 (i9505 international version with the same DAC as the one). Listened on 12 and 16ohm IEMs and 32ohm cans.

First interesting thing was that the Samsung stock player is noticeably louder than PowerAMP (not yet tested other third party players). Around 10dB louder if PowerAMP has direct volume control off and ~5dB louder if PowerAMP is using direct volume control. Since we wanted to use the same player in both devices we had to drop the HTC One volume down three notches for them to match closely (only one step to match with the S4 stock player).

Turned off Beats on the One.

Noted with interest that when blinding was removed and we could see which phone we favoured in our notes we both preferred the S4 sound over the One overall, better instrument separation and clarity, less fatiguing and harsh. Generally preferred the bass on the One, although the One was quite boomy. There was one MAJOR problem with the S4 though - crackling with prominent low bass frequencies. This was absent on the T5p at 32ohm, noticeable on the 16ohm IEMs, and very noticeable on the 12ohm IEMs. I ripped the 12ohm ones from my ears after about 5s, the crackling made them unbearable. If this is output impedance related as it appears to be this would put the S4 (i9505) output impedance somewhere in the 4-8ohm range, making it impossible to recommend to anyone using more sensitive IEMs.

 

[Reuben Chew]

Hey guys, I have uploaded a YouTube video to better demonstrate the problem: Samsung Galaxy S4 i9505 Audio Distortion with Low Impedance IEMs - YouTube

If you could approach your local Samsung rep with this video it'd be great.

Also, if you could re-share my Facebook post: https://www.facebook.com/reubenchew/posts/10151428267560966

If it goes viral maybe Samsung would be forced to act.

 

[jensigner]

I have added more information to this web page, including audio spectral plots showing the problem (for headphone impedances less than ~ 14 ohm). Also I have measured the output impedance of the S4 to be ~ 2.5 ohm, almost identical to that of the S3. The page below also contains a zip archive of a few audio test files for those who want to do their own tests with controlled sine wave tracks:
Samsung Galaxy S4: Problem With Low-Impedance Headphones

Included in the web page article is an RMAA measurements summary showing the outstanding audio performance and very low noise of the S4. The tests were performed with 33 ohm pure resistive loads to factor out any specific headphone artifacts. Included here for easier comparison with GSMArena results above. The stereo crosstalk I measure is partially due to the preamp used and will be updated:

 

[jensigner]

...

It turns out that the Snapdragon Galaxy S 4 has the best audio quality of any phone they've tested. I didn't expect that from Qualcomm considering the performance we've seen so far.

This change makes my decision a lot more difficult. I'm still going to wait for Google I/O and go into a store with a set of earbuds to hear for myself, but this phone just went from "definitely not" to "I'm listening...".

My independent RMAA audio quality tests on my S4 (SGH-I337) support the really exceptional audio performance of the S4 with headphone impedances of 32 ohm or higher. My RMAA results without a headphone load (open) are almost identical to the GSMArena results above. However under 32 ohm pure resistor load, the THD and crosstalk degrade but are still very low. Details are at the bottom of this page:
Samsung Galaxy S4: Problem With Low-Impedance Headphones

 

[Reuben Chew]

Thanks! Someone over at XDA claims that his UE900 (impedance of 30ohms) experiences the same problem, but only at 60Hz and 250Hz at only volume 11 (1kHz is unaffected). It's very strange since according to innerfidelity the lowest impedance that the UE900 goes is 20 ohms and should not experience this issue by right.

 

[jensigner]

An RMAA comparison of the S4 vs S3 audio quality, performed at 33ohm headphone load and at a volume level of 14/15:

The S3 is very good (except for crosstalk) but the S4 is significantly better in the numbers with an exceptionally low noise level of -100dB.
Would you actually hear the difference? The very pickiest audiophile listener might hear a very slight difference in tone coloration due to the higher IMD+Noise of the S3.
Based on my own listening tests (jazz, classical, rock) using the stock 35 ohm buds and my Sennheiser HD598 phones (50ohm), both the S4 and S3 provide a fantastic audio listening experience with headphones in this impedance range.

 

[Patrik_swe]

I contacted Samsung several times (one chat and two mail) and got mixed answers about this issue.

They clearly state that Samsung in-ears work and they do for all people.

I asked them of advice for impedance when purchasing earphones for Galaxy S4. They answered "choose LOW impedance". What I understand this issue is caused by low impedance earplugs.

When chatting with one support technician I got advice my 3,5 mm jack is broken or do a factory reset.

No one I contacted never heard of this problem with crackling audio but in the last mail response I got they wrote that they should check it up.

 

[jensigner]

I'd like to bring you up to date on this issue:
Based on quite a few i9505 owners, there is a problem with the headphone audio output FOR LOW IMPEDANCE HEADPHONES <~ 16 OHM. The problem doesn't appear to occur for the i9500. There is a very long discussion at XDA describing experiences. Those who contacted Samsung support on this have reported no progress. I have contacted Samsung Mobile support several times but have received no useful response. They apparently are aware of my measurements but haven't acknowledged the problem.
GSMArena have published a comparison of i9500 and i9505 showing almost identical and superb headphone audio performance. However those tests were performed using headphones with > 30 ohm impedance where the problem doesn't occur. My independent tests for 33 ohm resistive headphone loads support this outstanding audio result.
FURTHER, objective comments posted to GSMArena on that article by myself and others, which didn't seem to violate any of the posting guidelines there, have been quickly removed for no obvious reason without any explanation. Form your own conclusions. So what does this all mean?
If you are planning on getting an S4 i9505 version and intend to use good headphones with low impedance, you will have a problem. If you are happy with the stock (35ohm) earbuds or use headphones with impedance > ~ 20 ohm, you will likely be very very happy with the headphone audio. My Senn HD598 (~50 ohm) are just outstanding with my SGH-I337 (similar to i9505).

imo this looks like a Samsung design problem in implementation of the Qualcomm Snapdragon 600 audio. Since other smartphone use the same processor, I'd say that it isn't an intrinsic problem with the audio associated with that processor.
You can read about this problem also Samsung Galaxy S4: Problem With Low-Impedance Headphones
Also, if you are considering any forthcoming Samsung smartphones that use that Snapdragon 600+ platform and you really care about headphone audio quality, do your homework ... and test it in-store with low-impedance phones.

 

[Patrik_swe]

I'm very happy with S4 audio. The Adapt sound feature is worth a try for different results with your specific earbuds.

The only problem is crackling audio but this issue should be solved with latest firmware. I had terrible crackling with Yamaha EPH-100 but after firmware upgrade I can't hear any faults.

Read more here:

http://forums.androidcentral.com/sa...s4-audio-problems-what-should-samsung-do.html

Good work "jensigner"!