A little test on how loud some players really are

[dfkt]

Yeah, they're all more or less the same.

[CrowdAds]

[lestatar]

Indeed. I never knew this thread existing - I too would be curious to see how the Cowon J3 would place among this group.

dfkt, any chance for an update with new gear you might have on hand now, 1.5 years later?

^ never mind

[dfkt]

I think the only thing that changed is that Rockbox on the Clip/Clip+ in the meantime has higher output than when I did this test.

The non-PMP Cowons all have the same SoCs, same DACs, same everything, there really isn't anything noticeably different.

[NwAvGuy]

This is an issue that I bring up in the Testing Methods post on my blog. The accepted practice is to measure the maximum actual (not relative as done here) output at 1% THD (which is widely considered the "edge" of clipping) driving a realistic and known load.

But, sadly, this is very difficult to do with only RMAA and a sound card are your main tools (see my RMAA blog post). And when you're measuring portable players you can't directly control the "input" level (which is usually a fixed level test track).

As I discuss on my blog, this is best done with an audio analyzer to display real-time THD information so you can gradually increase the volume on the player until you either reach the maximum volume setting or exceed 1% THD--whichever happens first. You then measure the actual (not relative) output into the load and you know the actual maximum undistorted output of the player.

This is exactly what I've done in the review data I've posted on my site. The Clip+, for example, doesn't clip at all even playing back a 0 dBFS test signal (the loudest signal that you can have in a digital sound file) even while driving a relatively worst case 15 ohm load. It's maximum output at 15 ohms is 489 mV RMS.

Because the Clip+ has an unusually low output impedance it's output voltage doesn't vary much with changing loads. So you can generally estimate the power output from the formula P = (V * V)/R. In this case, with a 16 ohm load, you get (.489 * .489)/16 = 14.9 mW.

The Apple iPod Touch 3G, by comparison, could manage a very similar 474 mV into the same 15 ohm load before going to into heavy clipping at the next higher volume setting. It, however, has a much higher output impedance so will do somewhat better into higher impedance loads than the Clip+ and worse into lower impedance ones. You have to account for the output impedance if you want to estimate power into other loads besides the one used for the test (unless the output impedance is close to zero).

The Cowon i9 measured 624 mV into 15 ohms. This is 24.3 mW into 16 ohms. To convert the Cowon's advantage over the Clip+ to dB, you use 10 log(24.3 mW/14.9 mW) = 2.1 dB. So the Cowon, into typical 16 ohm headphones, will play 2.1 dB louder than the Clip+. Not a huge difference but audible.

This is all further complicated by the fact many players don't have the same maximum output (i.e. below 1% THD) at all frequencies. Some, for example, can't even get close at low frequencies and others do poorly at high frequencies. So, if you want to be more accurate, you run a THD vs Frequency plot and find the max level where the player doesn't exceed 1% THD at all audible frequencies. That, however, is more time consuming because the plots don't run in real time. Ideally it would be nice to use something like pink noise, but it's much more difficult to measure distortion with pink noise.

You can find more info in the Testing Methods link on my blog (and in the Clip+ review):

http://nwavguy.blogspot.com/2011/02/...g-methods.html

[dfkt]

The intention of this post wasn't to exactly measure the output power of certain players, but to show that there's really not much difference in perceived loudness between various devices, contrary to what people might interpret into manufacturer's specs.

None of the players I tested clip at their maximum volume, or have over 1% THD. As I already said before, those tests won't mean much in the grand scheme of things, but in relation to each other it sure gives some clues that the arbitrary mW specs of the manufacturers don't mean much - or at least, that people shouldn't believe that a "74mW" Cowon D2 is actually louder than a "40mW" Cowon X5.

[kasrhp]

Nice test dfkt. I always knew (and obviously heard here) that the Clip lineup was on the mellow side.

[NwAvGuy]

Originally Posted by dfkt

The goal of this post wasn't to exactly measure the output power of certain players, but to show that there's really not much difference in perceived loudness between various devices, contrary to what people might interpret into manufacturer's specs.

None of the players I tested clip at their maximum volume, or have over 1% THD.

I completely agree about the manufacture's specs being misleading in many ways. Many also assume that say 30 mW is "twice as loud" as 15 mW which, even if the power ratings are accurate, is far from being true. Doubling the power only gets you 3 dB more volume ( 10*log(30/15) = 3 dB ), which most perceive as a relatively "slight" increase.

It's worth noting that for something to be perceived as roughly "twice as loud" you need 10 dB more SPL (volume). And that requires 10 times as much power ( 10*log(10/1) = 10 dB ).

So for a player to be perceived as "twice as loud" as the Clip+, it would need about 150 mW into 16 ohms. And I'm not aware of many portable players that even claim that much power let alone actually deliver it at low distortion.

It's also hard to believe *none* of the long list of players clipped in your test? Were you playing a 0dBFS file? If not, all you have managed to do is compare their relative *gain* to each other at maximum volume. Which sometimes has little to do with how loud they will get playing real undistorted music into a real load.

Players (like the iPod) typically build in more gain than is usable with a music track that hits 0 dBFS so users can crank up the volume higher on something that's recorded at a lower level--or the quiet sections of classical music. So it's very common for a player to reach clipping at max volume playing back something recorded close to the digital limit (say a highly compressed Lady Gaga track).

If, as you mentioned, the chip/amp used in the player is capable of even higher output than what the manufacture considers a safe maximum, then indeed, it won't clip at all (like the Clip+).

Another effect of clipping is it compresses the music and adds a lot of distortion products to the music. Both of these things make the music *seem* much louder than it really is. That's one reason live concerts often push the levels into hearing damaging territory. As long as the sound remains "clean" many people don't perceive it as being "really loud".

Sadly, we're psychologically conditioned to associate "loud" with "compressed and distorted". So it's quite possible Player A may SEEM much louder than player B. When in reality it's just that Player A is being driven well into clipping.

[The DarkSide]

At least I've had the verification that my old clip was not as loud as my Cowon D2,...satisfaction!!!

[dfkt]

I added the Samsung R0, Archos 32, and the Rockboxed Clip+ (with improved volume level compared to the old version) to the test in the first post.

Originally Posted by NwAvGuy

It's also hard to believe *none* of the long list of players clipped in your test? Were you playing a 0dBFS file?

I've tested all the players with RMAA, and none clip at maximum volume (Rockbox at +6dB does usually clip, I just added that for reference, only the 0dB Rockbox tests are relevant). Yes, the test file was 0dBFS, as stated in the first post.

[Lagoo56]

Oh now I feel special for owning a 32GB Cowon O2.That was the best $144 deal ever hehe.

[Marvin the Martian]

I would assume that when doing these tests that all EQ's and enhancements were disabled....which brings up a question regarding the Cowon players with their assortment of audio tweaks.

If you go through the JetEffect presets with the i9, for example...BBE, BBE VIVA, VIVA 2. BBE Headphone, etc.....some of them clearly add to the perceived volume level, some more so than others. So that makes me wonder if on the "Normal" setting where everything is disabled, when you max it out in performing these types of tests, are you really getting the full output of the player, or is some held in reserve for use with these enhancements?

For that matter, will trying the tests with those enhancements enabled simply push the player to the point of clipping?

Just some random thoughts at 04:33 EST. Now I'll go back to bed.

[BruceBanner]

I owned the cowon o2, x5 and d2 as well as clip. clip_ and fuze. I HATED the o2 but have to admit that fella was L-O-U-D! the sq for audio was very very good, shame about hte rest of the players features.
For me the sansa clip paired with the right headphones gives all the volume i need and believe me i like my music LOUD (PS i hardly play it less than 6db RBed with superfi 5 IEMS).

Bruce