[SparkOnShore]

Like I said, ignorance is bliss!

I agree. Audiophoolishness as well!! Best wishes!

 

[yuriv]

Thanks for the note. I know about that clipping (I read Archimago too)). Yes, it was ruled out during recording of SE test sequences. The sound signature (artifact signature indeed) of iPod 5th is due to Wolfson chip architecture. Mid frequencies degrade to more extent than highs. Odd harmonics prevail in the distortion.

Some more thoughts on those old iPods, now that you've ruled out the clipping issue:

• It looks like you're testing with a 32-ohm dummy load on each channel, at 150mVrms. From what I can remember, the performance of the 3.5mm output gets a lot worse with a low-impedance load like that. The distortion figures are much higher, for example. But that's really the built-in headphone amp's fault, not the DAC's. Newer iPhone and iPad headphone amps are a lot better here.
  
• If the testing chain really looks like it does from your picture, then you're really testing DAC -> Headphone amp -> output R + blocking cap (even though the R+C really should be part of the amp itself). At a 32-ohm load, you're going to see (and hear) the effect of that capacitor on the frequency response. It's quite noticeable, much more so than the differences between DACS.
  
• If you really want to characterize the performance of the DAC, take the line output from the 30-pin connector. Of course you'll have to do away with the 32-ohm dummy load. Or, if you dial back the volume so that it's at unity gain instead of overdriving the headphone amp, you can hook up the 3.5mm jack to a typical line in (10 k+). The performance should be a lot better because you're not seeing the effect of the headphone amp being asked to drive a difficult load.
  
• I also noticed that even with a high impedance load, the noise and distortion go way up when the iPod is charging via the 30-pin connector. The distortion was up by a factor of 10, IIRC.
  
• Again, what's really noticeable in real-world use, much more than the DAC's performance itself, is the headphone amp. Its self noise isn't as good as the newer Lightning or USB-C dongles. You can really notice it on ultra sensitive IEMs. The bass roll-off from the blocking cap is also very noticeable on low impedance loads.

I'll have to dig up my old scratched up iPod to see if the hard drive still fires up, lol.

 

[Sterling2]

couldn't you just use the lightning headphone adapter dongle?

I needed a device which could be plugged into the cigarette lighter to keep the iPhone charged, as well as a DAC function.

 

[Sgt. Ear Ache]

I needed a device which could be plugged into the cigarette lighter to keep the iPhone charged, as well as a DAC function.

gotcha

 

[gregorio]

[1] Done it already, and difference is night and day. Have you? [1a] or don't and save a bunch of money. Ignorance is bliss.
[2] Also, you have to have the gear to hear the difference.

1. I have, and so have countless thousands of others. That leaves two possibilities, either: A. You did NOT do "a correct blind test" or B. Your hearing is "night and day" different to all other human beings and therefore you are an alien, a machine or a mutated human being.
1a. Or, do an incorrect test and spend a bunch of money on equipment that is audibly identical. Ignorance is expensive bliss!

2. Using the very best, highest fidelity equipment, the difference cannot be heard. Logically therefore, the "gear to hear the difference" must be relatively poor, low fidelity gear. But then of course you run into another logical problem, how can you describe a low fidelity DAC as "good", compared to a higher fidelity DAC, that is "bad"?

Agreed on this front. Every piece of gear has a bit of color. Even the supposed absolutely neutral stuff. This is what makes the great gear stand out.

The same logical problem as point #2 above. While every piece of gear has some level of colouration, in some/many pieces of gear that level is below the threshold of audibility and it's therefore audibly transparent, IE. Has perfect fidelity (as far as human hearing is concerned). Under the designed for conditions, nearly all average/modestly priced cables, DACs and amps fall into this category. Therefore, for some piece of gear to "stand out" it must have a level of colouration above audibility, not be audibly transparent and be of lower fidelity, very significantly lower fidelity if the difference is "night and day"! In other words, you are effectively saying that "what makes great gear stand out" (in these categories) is low fidelity?

Conclusion: thanks to Wolfson audio chip with relatively high level of waveform degradation iPod 5th (Video) has its own distinctive sound signature (slightly harsh mids, transparent highs).

From what I can tell you didn't test the Wolfson chip, you tested the combination of Wolfson chip + amplifier and therefore you have no basis for your conclusion because it could be entirely "thanks to" the amplifier and have absolutely nothing to do with the chip. And, as already mentioned by others, it could be entirely thanks to the amplifier ONLY when under an inappropriate (not designed for) load. Do you have any objective evidence to support your conclusion?

G

 

[SergeSE]

You might want to double check that with a quick listening test. Because I've done careful listening tests with all of my iPods, iPhones and Macintosh computers, including ones with Wolfson chips and going all the way back to the 8500AV, and they all sound identical. Something else might be causing the distortion you measured, or perhaps it is a form of distortion that isn't audible.

I have some doubts that quick listening test will be appropriate for the case. At least I will need high impedance and high sensitivity headphones. Not to say about providing true blindness of the test, which is not easy indeed. That's where the music-based audio metric can be helpful. All those artifact signatures, Df levels and related math help to find the cases where listening tests can be safely substituted with measurements. Measurements are repeatable and less complicated than listening tests, they must be used for evaluation of sound quality (in combination with listening tests) for sure. What exact measurements should be used is another question. I proposed df-measurements for the purpose as more appropriate/informative than traditional set of parameters - THD, freq.responce, etc.

Anyway the listening test of iPod 5th is impossible at the moment as the player was not mine and I don't have it now )).

Some more thoughts on those old iPods, now that you've ruled out the clipping issue:

• It looks like you're testing with a 32-ohm dummy load on each channel, at 150mVrms. From what I can remember, the performance of the 3.5mm output gets a lot worse with a low-impedance load like that. The distortion figures are much higher, for example. But that's really the built-in headphone amp's fault, not the DAC's. Newer iPhone and iPad headphone amps are a lot better here.
  
• If the testing chain really looks like it does from your picture, then you're really testing DAC -> Headphone amp -> output R + blocking cap (even though the R+C really should be part of the amp itself). At a 32-ohm load, you're going to see (and hear) the effect of that capacitor on the frequency response. It's quite noticeable, much more so than the differences between DACS.
  
• If you really want to characterize the performance of the DAC, take the line output from the 30-pin connector. Of course you'll have to do away with the 32-ohm dummy load. Or, if you dial back the volume so that it's at unity gain instead of overdriving the headphone amp, you can hook up the 3.5mm jack to a typical line in (10 k+). The performance should be a lot better because you're not seeing the effect of the headphone amp being asked to drive a difficult load.
  
• I also noticed that even with a high impedance load, the noise and distortion go way up when the iPod is charging via the 30-pin connector. The distortion was up by a factor of 10, IIRC.
  
• Again, what's really noticeable in real-world use, much more than the DAC's performance itself, is the headphone amp. Its self noise isn't as good as the newer Lightning or USB-C dongles. You can really notice it on ultra sensitive IEMs. The bass roll-off from the blocking cap is also very noticeable on low impedance loads.

I'll have to dig up my old scratched up iPod to see if the hard drive still fires up, lol.

From what I can tell you didn't test the Wolfson chip, you tested the combination of Wolfson chip + amplifier and therefore you have no basis for your conclusion because it could be entirely "thanks to" the amplifier and have absolutely nothing to do with the chip. And, as already mentioned by others, it could be entirely thanks to the amplifier ONLY when under an inappropriate (not designed for) load. Do you have any objective evidence to support your conclusion?

G

Actually I agree with all your points (except may be the next to last by bigshot, can't check it now). I consider DAC and amp as a single stage because today they are designed together as a SoC solution in most cases (“Wolfson chip”). Even more, I'm trying to treat a source of sound as a black box in my measurements. Proper audio measurement procedure should not account the way a sound was processed inside a device. And I use the 32 Ohm resistive load roughly simulating real-life usage scenario. So I don't see any sense to measure a DAC separately in portable players.

It seems to me that the main challenge today while designing headphones amplifiers in portable players is power consumption efficiency and trade off between the latter and sound quality.

 

[bigshot]

It looks like you're testing with a 32-ohm dummy load on each channel, at 150mVrms. From what I can remember, the performance of the 3.5mm output gets a lot worse with a low-impedance load like that. The distortion figures are much higher, for example. But that's really the built-in headphone amp's fault, not the DAC's.

When I did my test, I compared line out to line out amped through the same amp.

I have some doubts that quick listening test will be appropriate for the case. At least I will need high impedance and high sensitivity headphones. Not to say about providing true blindness of the test, which is not easy indeed.

It isn't hard at all. The old iPods had true line out through their dock. All you do is run the output of the line out through a preamp to level match into an amp with switchable inputs. Then take your reference line out and run it into the same switchable amp. Get a friend to switch for you and record your choices. If you are able to discern a difference, swap the preamp you're using to level match to the reference instead of the iPod and run the test again to make sure the preamp isn't coloring it. I didn't get that far... My choices were clearly total guesses. Of course you could just get two identical preamps too.

You can't depend solely on measurements and try to relate that to audibility in the abstract. You have to use real human ears. People in hifi forums who don white coat and pretend to be scientists tend to fall prey to doubling down on "what ifs" and "just to be safes". They end up thinking that things that are completely inaudible in the real world might just possibly be audible if the circumstances are right. They're correct of course, but who lives in an anechoic chamber? I sure don't! I just want to know if something is audibly transparent in the way that I use it... listening to music in my living room. It isn't hard at all to set up a test to determine that.

 

[Voxata]

I'm saying it had some different identifiable characteristics. For example the SU-8 and DX7s Bal out to thx 789 sounds different in specific ways.

 

[SergeSE]

You can't depend solely on measurements and try to relate that to audibility in the abstract. You have to use real human ears. People in hifi forums who don white coat and pretend to be scientists tend to fall prey to doubling down on "what ifs" and "just to be safes". They end up thinking that things that are completely inaudible in the real world might just possibly be audible if the circumstances are right. They're correct of course, but who lives in an anechoic chamber? I sure don't! I just want to know if something is audibly transparent in the way that I use it... listening to music in my living room. It isn't hard at all to set up a test to determine that.

In music-based audio metric such relationship between measurements and audibility can be researched and used in practice. In traditional audio metric this relationship can be traced only by audio professionals with experience. For consumers the interpretation of traditional measurements is made mostly by marketers.

I'm sure that both approaches objective and subjective are valuable for assessment of perceived sound quality. We just need to find the right combination of them.

To have some "safe margins" is quite reasonable due to complex nature of human perception of sound and its quality and due to our not complete knowledge of the issue. Also we don't need to be 100% efficient while manufacturing and choosing the right audio equipment. Some extra quality is psychologically comfortable when we talk about Music and shows some respect to audiophiles. In the end audiophiles do not try to sell you usual audio equipment as premium one, manufacturers do.

 

[Steve999]

In music-based audio metric such relationship between measurements and audibility can be researched and used in practice. In traditional audio metric this relationship can be traced only by audio professionals with experience. For consumers the interpretation of traditional measurements is made mostly by marketers.

I'm sure that both approaches objective and subjective are valuable for assessment of perceived sound quality. We just need to find the right combination of them.

To have some "safe margins" is quite reasonable due to complex nature of human perception of sound and its quality and due to our not complete knowledge of the issue. Also we don't need to be 100% efficient while manufacturing and choosing the right audio equipment. Some extra quality is psychologically comfortable when we talk about Music and shows some respect to audiophiles. In the end audiophiles do not try to sell you usual audio equipment as premium one, manufacturers do.

You made claims of an audible difference in equipment without doing any reliable listening to back it up, or evidence or data that the difference, however much it might approach the audible, would actually be audible to you or anyone. I really like a lot of your work but for me that is going one step too far. It takes us down the rabbit hole.

As a for example, I looked up @Voxata ‘s equipment he listed in his last post. It had been carefully measured on another forum I find credible. Apparently it all measures extremely well for its price point. He seems to have chosen carefully. If having something that measures really well for the price point gives you pride in ownership and it doesn’t break your budget at all I’m all for it. It’s not my cup of tea but I get it. But as the reviews made clear all characteristics that were measured and were not quite as good as the very best measurements they had ever encountered or were otherwise remarkable were far lower than the known thresholds of audibility.

Now, will this iPod you discuss sound exactly the same? If by measurements it’s an edge case I think it would be best to say we don’t know, rather than to speculate based on certain aspects of the measurements that it would be audibly non-transparent for its intended purpose and further speculate as to the purported minor non-transparent characteristics of the sound that would result. That’s why I think you went one step too far there, from someone with absolutely no dog in this fight.

 

[bigshot]

In music-based audio metric such relationship between measurements and audibility can be researched and used in practice. In traditional audio metric this relationship can be traced only by audio professionals with experience.

Most of the published studies on thresholds of audibility are done with tones, not music. In practice the thresholds are as much as an order of magnitude lower when you are using music than with tones. (PS I'm a professional with experience.)

You should just do some simple listening tests yourself. I know in audiophile forums people attribute all kinds of coloration to DACs, But I've done controlled listening tests of various DACs (Wolfson, Cirrus, Sabre, etc.) All the ones I've tested are audibly transperent. If you took the time to do some tests yourself, you would know this too.

 

[Logan75]

Good quality

 

[Sterling2]

Here's just a sidebar getting us out of the weeds of minutia for a minute. Although I have an iPad with analog output, that output is only used for headphones and it's rare for me to use headphones, since, in my private environment, I can enjoy my JBL L100t3's speakers without intruding on anyone's ears. So, the iPad's DAC is moot. I do however use my iPhone 7+ via Airport Express to listen to stereo music in my iTunes Library from my high quality home theatre system. This means to music pleasure is as easy on my ears as it is convenient. In fact, while I'm not a prophet, it's not a stretch to speculate that the future of home, as well as portable music enjoyment will be an iPhone Wi-Fi or Bluetooth wireless transmission to powered speakers. The need for control amps, AVR's, pre-pros, and all manner of source components will expire because the iPhone does indeed produce/deliver a high enough quality sound to be seen as sufficient by most folks on the Planet.

 

[Whazzzup]

Hugo far improves listening experience of an iPhone. Not even close

 

[SergeSE]

You made claims of an audible difference in equipment without doing any reliable listening to back it up, or evidence or data that the difference, however much it might approach the audible, would actually be audible to you or anyone. I really like a lot of your work but for me that is going one step too far. It takes us down the rabbit hole.

As a for example, I looked up @Voxata ‘s equipment he listed in his last post. It had been carefully measured on another forum I find credible. Apparently it all measures extremely well for its price point. He seems to have chosen carefully. If having something that measures really well for the price point gives you pride in ownership and it doesn’t break your budget at all I’m all for it. It’s not my cup of tea but I get it. But as the reviews made clear all characteristics that were measured and were not quite as good as the very best measurements they had ever encountered or were otherwise remarkable were far lower than the known thresholds of audibility.

Now, will this iPod you discuss sound exactly the same? If by measurements it’s an edge case I think it would be best to say we don’t know, rather than to speculate based on certain aspects of the measurements that it would be audibly non-transparent for its intended purpose and further speculate as to the purported minor non-transparent characteristics of the sound that would result. That’s why I think you went one step too far there, from someone with absolutely no dog in this fight.

Good question. To be honest there is not enough info about correlation between measurable degradation of waveform (Df, dB) and audibility of that degradation. Such info can be derived from results of various listening tests. I have only preliminary figures based on analysis of HydrogenAudio listening tests of codecs and beta-testing of 30+ portable players on SE site. At this point of research I'm pretty sure that 10dB improvement in reproduction of real musical signal is huge and must be audible. 10dB is the difference in measurements between early portable players like iriver E100, iphone 3G and modern top smartphones by Apple, Huawei and Xiaomi or iPod classic 2009 for example. Then, looking at df-slide with measurements of iPod 5th we can see that 1kHs Sine is degraded higher than 12.5kHz Sine (more harmonics) and odd harmonics are prevail over even ones (relative position of two triangle signals which is also confirmed by traditional spectral view of harmonic products). So it's quite easy to say how exactly that degradation will be audible - harsh mids. But in general I agree with your objection - relation of df-measurements and audibility should be researched better. I wish I had more time for this but I do what I can.

Most of the published studies on thresholds of audibility are done with tones, not music. In practice the thresholds are as much as an order of magnitude lower when you are using music than with tones. (PS I'm a professional with experience.)

You should just do some simple listening tests yourself. I know in audiophile forums people attribute all kinds of coloration to DACs, But I've done controlled listening tests of various DACs (Wolfson, Cirrus, Sabre, etc.) All the ones I've tested are audibly transperent. If you took the time to do some tests yourself, you would know this too.

I think my reply to Steve999 also suits your objections. I will just add that I'm trying to use results of third party listening tests for the research, not mine. The reason is clear I think.

[merged]