CHANGSTAR: Audiophile Headphone Reviews and Early 90s Style BBS
Lobby => Amp and DAC Measurements => Topic started by: ultrabike on September 18, 2014, 06:58:30 AM
-
So here is some commentary and RMAA measures of the Sansa Zip ... and the Creative Zen which I bought a while ago based on CNET and Amazon rave reviews:
http://www.cnet.com/products/creative-zen/
These are mini SS type MP3 players which have been heard by quite a few and may be used as reference for what to expect when looking at measurements for that same reason... Which is why I think NwGuy started his blog with the Clip+ and the iPhone. Well that and to take a dump on RMAA. The dScope is not cheap after all, and once the purchase is done you better believe that the dScope is awesome. Make a blog entry if that's what it takes... *sigh* and then came AMB, and HC, and HF, and the O2, and the ODAC and so forth...
Can't really complain too much, cuz I actually read Nw's speech on the Clip+. And given that Zip and Clip+ seem to share audio HW I decided to put like $30 where my mouth was to buy one, in part based on his recommends... and it's good!
This thing, after Rockbox, sounds fantastic for the price. I sometimes use it as a car player by either plugging the usb port or the headphone out to the car Aux in... and I think it beats the built in CD player. A fantastic little gadget that any cheap audio-enthusiast should have.
So anyhow, quite a few folks and their mom's have made measures of the Zip and Clip+, so what gives... here are a few more. The following are to 32 ohms and at 0.421 Vrms (according to my BK Precision 2709B purchased per Nate Maher's recommends (http://www.innerfidelity.com/content/2012-holiday-headphone-gift-guide-nate-maher))
Frequency Response
(http://www.changstar.com/index.php?action=dlattach;topic=1776.0;attach=7583;image)
Like Nw said, pretty awesome, and maybe some Class D thing w/o caps.
THD
(http://www.changstar.com/index.php?action=dlattach;topic=1776.0;attach=7585;image)
Nw measured this on the Clip at 0.186 Vrms instead of his usual 0.4 Vrms... which is fine. I think he came up with 0.05% into 15 ohms. Well I did it at 0.421 into 32 ohms and got 0.025%. The load is indeed easier so that might explain why I got less even when voltage was higher. Spectrums are kind of similar.
Output Impedance
Nw looses his mind about this one... and RMAA does not duit!!! But fear not. My $100 bucks multimeter can... He said 1 ohm, I got 0.61 ohms.
Maximum Output
Yup, RMAA cannot measure that one either, cuz PC's usually don't come with quality integrated multimeters. So Nw determined that max output was somewhere around 0.489 Vrms, and Clip+ didn't clip! Well, using Rockbox it turns out that one can go +6 dB which measures 0.845 Vrms, and Zip still doesn't clip.
(http://www.changstar.com/index.php?action=dlattach;topic=1776.0;attach=7587;image)
I get about 0.029% distortion, instead of Nw's 0.06356(2681258158)%, proly cuz I used 32 ohms load instead of 15 ohms.
IMD distortion
So he gets 0.03% (0.180 Vrms into 15 ohms) which he decides is inaudible based on his tape hiss speech (which I dunno if true). I get 0.038% (0.421 Vrms into 32 ohms). And yes plots look kinda similar.
(http://www.changstar.com/index.php?action=dlattach;topic=1776.0;attach=7589;image)
So, didn't do DAC linearity or Jitter spectrum. Maybe next time. Thinking maybe REW for that one. Dunno.
Nw did some RMAA stuff w 0.180 Vrms into 15 ohms which does look different from his dScope stuff. I may put together a 15 ohms load and do this again to check the IMD numbers he got. Maybe something is up with RMAA... or Nw's Benchmark ADC1.
*CREATIVE ZEN*
Now comes the Creative Zen as an example of what can go wrong with some DAPs. The maximum voltage I could get out of this DAP was 0.048 Vrms. One is going to need some very efficient cans. I had to do some trickery to get RMAA to even detect it.
Impedance? like 0.94 ohms.
Frequency Response (32 ohm load)
(http://www.changstar.com/index.php?action=dlattach;topic=1776.0;attach=7591;image)
... Yes, it sounds like that. Maybe it's got some caps at the output or something.
Distortion is hard to characterize given that at 0.048 Vrms it may be bellow the noise floor. Hard to judge by the plots.
So summary: Clip+/Zip rocks, Creative Zen (discontinued) SUCKS! RMAA also rocks. Disclaimer: Used my $150 (or less depending on sales and Black Friday conditions) 2i2 as interface + $100 multimeter, so proly all that I did is wrong. Also, no animals were harmed in the testing.
-
Interesting to see a Rockboxed version also has a higher output voltage.
Makes you wonder why Sandisk limited the output and 'threw away' a good 6dB !
My clip sport measures 0.482V (@ 32 Ohm) and Fuze does 0.55V @32 Ohm (= 10mW) and both have about 0.7 Ohm output R.
Maybe I should Rockbox my Fuze ?
On the other hand acc. to this guy here: http://marlene-d.blogspot.com/2012/10/review-sansa-clip.html
Rockbox works fine with 44.1 but appears to be 'broken' on 48kHz with Clip+.
I am wondering if you get the same results on 48kHz with the clip zip.
-
Been using a rockboxed fuze for a long time with fiio E6 through LOD (lineout dock). Kinda troublesome when I'm mobile so I switched to Nokia N8 instead.
-
Thank you very much for the measurements!
Question: Is it audible that FR of Creative Zen differs for about 0.65db (at max) in 100Hz to 19.5khz area? Also, is the channel imbalance audible (I guess it's not...)?
-
Yuck, don't like the look of the THD spectrum. .025% is good so long as it's free of higher order components.
-
Interesting to see a Rockboxed version also has a higher output voltage.
Makes you wonder why Sandisk limited the output and 'threw away' a good 6dB !
My clip sport measures 0.482V (@ 32 Ohm) and Fuze does 0.55V @32 Ohm (= 10mW) and both have about 0.7 Ohm output R.
Maybe I should Rockbox my Fuze ?
On the other hand acc. to this guy here: http://marlene-d.blogspot.com/2012/10/review-sansa-clip.html
Rockbox works fine with 44.1 but appears to be 'broken' on 48kHz with Clip+.
I am wondering if you get the same results on 48kHz with the clip zip.
I recommend Rockbox. Will see what I can do to check 48kHz.
Thank you very much for the measurements!
Question: Is it audible that FR of Creative Zen differs for about 0.65db (at max) in 100Hz to 19.5khz area? Also, is the channel imbalance audible (I guess it's not...)?
It's just a bit lean, depending perhaps on the impedance of your cans.
Yuck, don't like the look of the THD spectrum. .025% is good so long as it's free of higher order components.
Yeap. It's not perfect. Nw gave the mini3 a much harder time for 0.016% @ 0.4Vrms into 15 ohms, and ultrasonics.
-
Yuck, don't like the look of the THD spectrum. .025% is good so long as it's free of higher order components.
IME this doesn't necessarily have to sound bad or unpleasant. The Metrums had high order distortion components as well, though I know their sound comes down to a personal preference thing.
-
I was alone in thinking the Metrum had a bit of digititus, a bit of nastiness in the upper treble. Even though it was laid back tonally. From my time comparing, I recall the Gungnir and the gamma2 both having a smoother and easier treble quality.
-
The guy I sold my Hex to said he thought it sounded rough compared to the PWD, so you might not be alone. I dunno, not something I could pick up on no matter how much I tried. I guess we'll just need to have a fist fight over it. (Now, put a 0dB 15KHz tone through it, and you'll think you became a 56k modem yourself!)
-
Here is another article by CNET that I remember reading before buying the Zen...
http://www.cnet.com/news/audiophile-mp3-players-by-the-numbers/
Apparently CNET does have (or had) an Audio Precision. W/O a load the Zen doesn't measure too bad. Using mid to high impedance cans may also measure (and sound) OK, but then you have a trade off given it cannot put out too much voltage swing. Proly there is a sweet spot with these (or similar) players. This particular player is discontinued anyhow.
As far as their distortion, noise and SNR numbers... Dunno the levels used across the players, so hard to say. The Zen doesn't go too loud and maybe artificially limited so as to measure good at "full-scale". Again, dunno.
It also kinda goes to show that one can get great or horrible results depending on measurements conditions, regardless if using a dScope, AP or whatevers.
-
Lol, Cowon builds in an EQ. Lots of portables with bass roll-off.
I'm looking at scopes right now, but ones with differential inputs are kind of pricey. No - not getting a dscope. there are some thigns about the dscope specs which seem kind of iffy.
-
Yeah. Differential stuff tends to be pricey. Good probes in general can be.
-
On the other hand acc. to this guy here: http://marlene-d.blogspot.com/2012/10/review-sansa-clip.html
Rockbox works fine with 44.1 but appears to be 'broken' on 48kHz with Clip+.
I am wondering if you get the same results on 48kHz with the clip zip.
Yes. Something might be horribly wrong with 48 kHz under Rockbox. Will report back. Need to use ARTA also, since RMAA currently doesn't handle well 2i2 ASIO drivers past 44.1 kHz. It may introduce some slight unexpected gain also in 44.1 kHz.
I got results more similar to Nw this time with ARTA, even with the 32 ohm and ~0.4V levels, so that's that. Don't think RMAA is bad, but I guess one needs to learn the quirks of the tool.
Even with with the stock FW @ 48 kHz and either way with 44.1 kHz, standard IMD calculation seems to neglect certain crap in the treble region. THD might definitively be much higher with two tones than IMD calculations might suggest.
Will report back tomorrow...
-
When it is dual boot see how it performs on the original FW as well.
My Fuze works very well on 48kHz (I have quite a few 96/24 files converted to MP3-320 which have become 48/24) it performs measurably better on 48 compared to 44.1
On the Sansa's the resolution will only be <16 bits though (limit of the DAC chip) but on some players the full 24 bit depth is reached with MP3.
-
Yup, dual booted for a while this thing today...
Well, rather disappointed at what turns out to be SMPTE IMD 60Hz 7kHz. It seems the % values reported typically include only the 2nd and 3rd order effects. Some talk about it here:
http://www.ap.com/kb/show/414
The problem I see is that the Zip and the Clip throw a LOT of curd at higher order IMD products, even when not broken by Rockbox + 48kHz cuz in that case shit hits the fan.
Furthermore, the Zip is a biatch to measure due to the pitch error issue. This pitch error is BOTH FW and rate dependent. Some of the measurement packages expect the 2nd and 3rd order effects at certain frequencies. Well, the pitch has the effect of moving the distortion components a few Hz off and things can look uber-great or uber-bad. No clear way to sync things up with a DAP since it's clock is not slaved to the interface ADC. Did the best I could there so here it is.
First a few corrections. According to my grain of salt measures...
Zip Rockbox max voltage out: 0.957 Vrms
Zip Stock max voltage out: 0.556 Vrms
Some bit about bits
Zip Rockbox: It can take 24 bits, but it measures virtually identical to 16 bits. Like SD said, best guess is the resolution is limited to 16 bits and the FW does something to make 24 bits play by removing LSBs.
Zip Stock: Placed a few 24 bits tones but the Zip refused to play them.
Pitch and 48 kHz
The pitch on the stock FW seems a bit more off than when using Rockbox (both are off BTW). However, 48kHz files work with the stock FW. Rockbox fucks things up with 48kHz files... Funny thing is that even if Rockboxed Zip manures a 48kHz file, SMPTE IMD reported % distortion doesn't look too bad given curd doesn't happen to hit expected 2nd and 3rd order IMD distortion location. IMD + Noise might catch it though.
So for 393mVrms @ 32 ohm here are some THD (1kHz) and IMD (60Hz + 7kHz 4:1) measures.
Stock FW @ 44.1kHz
THD: 0.0317%, IMD: 0.133%
(http://www.changstar.com/index.php?action=dlattach;topic=1776.0;attach=7600;image)
(http://www.changstar.com/index.php?action=dlattach;topic=1776.0;attach=7602;image)
Stock FW @ 48kHz
THD: 0.0316%, IMD: 0.133%
(http://www.changstar.com/index.php?action=dlattach;topic=1776.0;attach=7604;image)
(http://www.changstar.com/index.php?action=dlattach;topic=1776.0;attach=7606;image)
Rockbox FW @ 44.1kHz
THD: 0.0276%, IMD: 0.122%
(http://www.changstar.com/index.php?action=dlattach;topic=1776.0;attach=7608;image)
(http://www.changstar.com/index.php?action=dlattach;topic=1776.0;attach=7610;image)
Rockbox FW @ 48kHz
THD: 0.0276%, IMD: 0.122% (Awesome, yeah right)
(http://www.changstar.com/index.php?action=dlattach;topic=1776.0;attach=7612;image)
(http://www.changstar.com/index.php?action=dlattach;topic=1776.0;attach=7614;image)
As far as the IMD results with 48kHz with Rockbox, d2 and d3 are negligible compared to the weirdness around 3kHz, 800Hz, and pretty much all the crap everywhere.
-
Here are the 44.1 kHz IMD results (Stock FW) but zoomed around 7kHz.
(http://www.changstar.com/index.php?action=dlattach;topic=1776.0;attach=7616;image)
It seems d5, d7, and d9 look more significant than d2, and IMD SMPTE will proly ignore those.
Even when "working" at 44.1 kHz, the Zip get's quite noisy in the treble area due to IMD stuff. Sure, the ultrasonics might not matter that much (Nw didn't extend that reprieve to the mini3), but there's still a lot of crap between 10 and 20 kHz.
Again, this is not a bad player for the price. I absolutely love it. But I also don't think it's perfect. And perhaps neither are my measurements or indeed the standard measurements.
-
Thanks....
Nice measurements and confirmation to not Rockbox mine .... ('cause of quiet a few 48kHz files mixed with 44kHz on my players)
I noticed some weird things when reproducing squarewaves with the Sansa's as opposed to the FiiO X3.
Should turn it into an article and post my findings.
The Sansa clip sport has HIGH amounts of 500kHz residu in there, the Fuze does not.
The attachments have scope images of a few tests with various players and tones.
They are 'raw' plots but their filenames show its relation, maybe some technical minded folks can make something out of it. ;)
I noticed during my efforts in measuring my Fuze that distortion drastically lowers (at least those 'sideband alike' issues) when NOT loading the HP out.
As I use always use my Fuze with an external amp there is no load and max amplitude issues are solved.
Despite its shortcomings I still like my Fuze but prefer the X3 (which froze on me yesterday again even with the latest FW)
-
Yup, with the Zip things seem to clean up quite a bit also into no load (and high impedance).
Into no load, Zip+Rockbox 44.1 kHz IMD drops to (legitimate) 0.044%, and much of the garbage is gone. In that case d2 and d3 dominate indeed (unlike the 32 ohm case). THD drops to 0.01% as shown here:
(http://www.changstar.com/index.php?action=dlattach;topic=1776.0;attach=7619;image)
(http://www.changstar.com/index.php?action=dlattach;topic=1776.0;attach=7621;image)
If my car (and maybe most cars) present a high impedance at it's AUX IN, that may explain why I like it a lot for trips... Well than and the fact that one does not have to deal with changing CDs.
It's quite a tiny device. I'm positively surprised it can put out the voltage it puts. Dunno about the current or in fact anything about what was done in the amp section.
-
So, since this was a Sansa Zip and Creative Zen thread here are some comparos w 32 ohms and ~0.068 Vrms (max Creative Zen can provide)
Frequency Response (Zip is red / Zen is white)
(http://www.changstar.com/index.php?action=dlattach;topic=1776.0;attach=7623;image)
THD (1 kHz) (Zip is red / Zen is white)
(Zen is 0.0032% THD)
(Zip is 0.0537% THD)
(http://www.changstar.com/index.php?action=dlattach;topic=1776.0;attach=7625;image)
SMPTE 60 Hz 7 kHz 4:1 (Zip is red / Zen is white)
(Zen is 0.012% IMD)
(Zip is 0.249% IMD)
(http://www.changstar.com/index.php?action=dlattach;topic=1776.0;attach=7627;image)
On paper...
+ Distortion: The Zen seems to have the upper hand.
+ Power Ouput: The Zen max output swing is about 0.068 Vrms into 32 ohms or 0.14 mW (slight correction from first post). Zip can do 0.556 Vrms into 32 ohms or 9.67 mW (cleanly w stock FW, Rockbox can do 28 mW but distortion increases). So Zip can get much louder.
+ Output Impedance: Zip is like 0.61 ohms and Zen is like 0.94 ohms.
+ Frequency Response: Zip is ruler flat in the audio range. Zen rolls off a bit depending on load. The lower the impedance of the cans, the more the roll off.
Both are OK players... Well... I still don't like the load impedance dependent roll of of the Zen and it's limp dick max power output, but distortion looks good. The Zen can play .m4a video which is kind of cool. Rockboxed Zip can play some video games, but the screen is a fraction the size of the Zen and much lower quality. I think the Zen is more restricted in audio format support.
Based on some results, the Rockboxed Zip suffers when playing 48 kHz and 96 kHz (Ok with 44.1 kHz and 88.2 kHz) files which I largely don't have, but if one does, it is a consideration vs sticking to the stock FW. The stock FW however will not play 24 bit files at all. Rockbox at least removes the LSBs and let's them out reasonably well. Zen does not recognize FLAC, but can play 48 kHz, 88.2 kHz and 96 kHz WAVs Ok.
Zen is AFAIK no more.
-
There hasn't been a Rockbox official release for over 18 months, however the daily builds have seen a lot of improvements relevant to the tests described in this thread. On targets whose hardware supports both 44.1 and 48KHz without sample rate conversion you can now manually toggle between the two rates. Rockbox's SRC was acknowledged to be bad and has been reimplemented from scratch by saratoga (RB dev) and much improved. I would guess that the same tests performed on a current firmware would show different results and that the differences may be audible, but I don't know.
The Clip+, Clip Zip and Fuze+ running current or recent daily builds will each show these benefits. The Fuze+ is actually different hardware from the Clips and doesn't have the pitch error at all. It is hideous to use with the Sansa firmware but very decent with Rockbox (touchpad controls become actually usable).