• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Battle of Battery Operated Portable Headphone Amps

jaytrinitron

Member
Joined
Jan 27, 2018
Messages
40
Likes
4
I'm going to provide a suggestion here (again) that would make these reviews more helpful for determining sound quality: 19+20kHz CCIF IMD tests, and also a 6kHz and 20kHz THD+N v power sweeps. These SINAD measurements and what-not are interesting, but I'm not convinced they say much about sound quality (and there seems to be an awful lot of focus on them). Particularly so since virtually all music out there is only CD quality anyway. Practically speaking, the most useful measurements being done right now are probably the THD vs power to see how much power the amps can crank out before clipping.

On top of proper IMD tests and high frequency power sweeps, CEA burst tests might be interesting to see if any of these portables have any extra gas in the tank on actual music. Just a thought. I don't mean to sound like I'm complaining. This stuff is incredibly useful and helpful. Thanks to a combination of digging through this site and RAA I just picked up the Xduoo XD-05 (https://reference-audio-analyzer.pro/en/report/amp/xduoo-xd-05.php#rw14). Just a hair less juice, but the DAC is built in, and pretty good. Plus, you can roll op-amps! :rolleyes:

You are the second person to suggest SINAD is not relevant for a DAC, and that perhaps the testing methodology needs to be revised. The first guy was the main designer from Pure Audio UK who stated he would only be sending his new product to Amir if Amir "revised his testing methodology" and stopped using SINAD as a measurement for a DAC (I assume now since most crack 96 dB, and in normal listening environs they don't even need to do that anyway). I thought it was just a guy scared of what the measured results of his device would look like (and it could still very well be that) and dismissed it but perhaps there is some truth to these claims of the measurement standards needing revising to more accurately reflect actual listening, though I am very much a layman and am not suggesting Amir's work isn't exceedingly helpful and very much appreciated. I haven't read everything here and it is possible Amir has already allayed these concerns elsewhere, though I did not find anything with some cursory googling. Does anyone else care to chime in?

Also, I appreciate the link to the site, always nice to get more measurement sources.
 

Headphonaholic

Active Member
Joined
Apr 8, 2018
Messages
222
Likes
283
You are the second person to suggest SINAD is not relevant for a DAC, and that perhaps the testing methodology needs to be revised. The first guy was the main designer from Pure Audio UK who stated he would only be sending his new product to Amir if Amir "revised his testing methodology" and stopped using SINAD as a measurement for a DAC (I assume now since most crack 96 dB, and in normal listening environs they don't even need to do that anyway). I thought it was just a guy scared of what the measured results of his device would look like (and it could still very well be that) and dismissed it but perhaps there is some truth to these claims of the measurement standards needing revising to more accurately reflect actual listening, though I am very much a layman and am not suggesting Amir's work isn't exceedingly helpful and very much appreciated. I haven't read everything here and it is possible Amir has already allayed these concerns elsewhere, though I did not find anything with some cursory googling. Does anyone else care to chime in?

Also, I appreciate the link to the site, always nice to get more measurement sources.
SINAD is as I understand it (still learning here so feel free to correct me) is merely metric we can use to get an idea of the worst case performance of a product. It's not the end all be all but it does indicate the general performance of something. To avoid measurement because of any one particular test to me suggests trying to cover up or hide something. Yes the argument can be made that it over simplifies things and people may take it the wrong way but it doesn't mean the test is bad. Furthermore this site is about finding not only the performance of products but the best ones. Sure you can argue that anything past CD quality is unnecessary but that isn't what the tests are about, it's about finding well engineered things.
 

jaytrinitron

Member
Joined
Jan 27, 2018
Messages
40
Likes
4
SINAD is as I understand it (still learning here so feel free to correct me) is merely metric we can use to get an idea of the worst case performance of a product. It's not the end all be all but it does indicate the general performance of something.
I have no idea tbh, but at least here it is argued that 85dB is enough at least audibility wise (although the "strict" or "engineering" limit still calls for 120). I have seen the importance of SINAD for DACs called into question in other places, too, though the more technical side is a bit hard to follow for me. I should have phrased my post differently though, I see no problem with including SINAD if the people and Amir want it. But perhaps there should be additional tests related more to how people actually listen like the poster I replied to suggested (FWIW, I have no idea if he's right or not or whether those measurements actually relate the way he says they do, but using measurements that reflect the way we actually listen makes sense to me).
To avoid measurement because of any one particular test to me suggests trying to cover up or hide something.

Agreed.
Sure you can argue that anything past CD quality is unnecessary but that isn't what the tests are about, it's about finding well engineered things.

Of course, but the devices should be well engineered in relation to their purpose. To inflate an artificial number while not actually improving performance is not good engineering, I don't think (Not suggesting that this is what SINAD is, like I said I'm merely wondering about the truth to the claims others have made).
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,388
Location
Seattle Area
You are the second person to suggest SINAD is not relevant for a DAC, and that perhaps the testing methodology needs to be revised. The first guy was the main designer from Pure Audio UK who stated he would only be sending his new product to Amir if Amir "revised his testing methodology" and stopped using SINAD as a measurement for a DAC (I assume now since most crack 96 dB, and in normal listening environs they don't even need to do that anyway). I thought it was just a guy scared of what the measured results of his device would look like (and it could still very well be that) and dismissed it but perhaps there is some truth to these claims of the measurement standards needing revising to more accurately reflect actual listening, though I am very much a layman and am not suggesting Amir's work isn't exceedingly helpful and very much appreciated. I haven't read everything here and it is possible Amir has already allayed these concerns elsewhere, though I did not find anything with some cursory googling. Does anyone else care to chime in?
I have addressed that in my Measurement Tutorial: https://www.audiosciencereview.com/forum/index.php?threads/understanding-audio-measurements.2351/

1548619234046.png


The last sentence is key. I have yet to see any evidence that doing poorly in THD+N improves sonic fidelity of any product.

The usual reason for poor THD+N is mistaken audiophile notions translated into design (e.g. lack of feedback), or simply bad design. Both of these need to be brought out through measurements like THD+N.

My tests have a suite of other measurements so if someone thinks they do well elsewhere, the results will show up.

What they don't do is paper over bad designs. Companies who are afraid of my measurements typically have either never measured their gear as I do (measurement gear is very expensive) or have measured it and know they don't perform well. In time, members will offer the same product for testing so it is unwise for them to take this approach. Let the measurements show the issues they can help resolve. Sticking one's head in the same is not constructive.
 

jaytrinitron

Member
Joined
Jan 27, 2018
Messages
40
Likes
4
I have addressed that in my Measurement Tutorial: https://www.audiosciencereview.com/forum/index.php?threads/understanding-audio-measurements.2351/

View attachment 20930

The last sentence is key. I have yet to see any evidence that doing poorly in THD+N improves sonic fidelity of any product.
Thank you for the link and explanation. I was not suggesting that doing POORLY in THD+N improves sonic fidelity (I should've phrased my post differently), but that I have seen people argue that doing WELL in THD+N (over a certain threshold) does not qualify sonic fidelity properly (but may help clarify other things, which is fine.) I mostly just included the "SINAD is not relevant for a DAC" statement because that is what the guy actually said; I don't necessarily think it's true and clearly according to you and based on what I understand it shows something regardless of how people interpret it, and is therefore worth doing.
My tests have a suite of other measurements so if someone thinks they do well elsewhere, the results will show up.
I see. What do you say to @ryanmh1's comment in this thread about including 19+20kHz CCIF IMD tests, and also a 6kHz and 20kHz THD+N v power sweep to relate more to actual sound quality? I have no idea if he's right or not, I have just seen it cropping up in more and more places that certain measurements may not relate well to actual sound quality/the way people actually listen (I am sure at least some of this is audiophile hogwash).
Let the measurements show the issues they can help resolve. Sticking one's head in the same is not constructive.
Agreed, and you've already proven many times that some DACs and amps are very poor at reproducing music transparently, no matter how much a company and subjective reviews tout them as excellent.
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,388
Location
Seattle Area
What do you say to @ryanmh1's comment in this thread about including 19+20kHz CCIF IMD tests, and also a 6kHz and 20kHz THD+N v power sweep to relate more to actual sound quality?
I am not a fan of CCIF IMD test because music spectrum does not at all have full amplitude 19 and 20 kHz spectrum. The SMPTE test I run has a low and mid-high frequency tones that are more realistic. And demonstrate low frequency distortion where music is loudest. Here is a random track from my library:

1548651854663.png


By the time you get to 20 kHz, your amplitude is down 55 dB lower than 1 kHz.

As to running other tests, I just don't want to overwhelm readers with test after test. Pretty soon people stop reading the reviews if it has so many tests. And I have yet to see additional tests show anything all that meaningful. Once I run the dashboard, the story is told with some rare exception.

Furthermore, dual-tone tests are difficult to related to audibility.
 

ryanmh1

Member
Joined
Jan 11, 2019
Messages
93
Likes
117
I am not a fan of CCIF IMD test because music spectrum does not at all have full amplitude 19 and 20 kHz spectrum. The SMPTE test I run has a low and mid-high frequency tones that are more realistic. And demonstrate low frequency distortion where music is loudest. Here is a random track from my library:


Furthermore, dual-tone tests are difficult to related to audibility.

I am not a fan of CCIF IMD test because music spectrum does not at all have full amplitude 19 and 20 kHz spectrum. The SMPTE test I run has a low and mid-high frequency tones that are more realistic. And demonstrate low frequency distortion where music is loudest. Here is a random track from my library:

[...]

Once I run the dashboard, the story is told with some rare exception.

Furthermore, dual-tone tests are difficult to related to audibility.

Amir -- I think you're doing some good work here, but I would argue that dual tone testing can be very related to audibility--far more so than THD testing. High frequency dual tone testing, in particular, can reveal the presence of nasty stuff that is completely unrelated to the musical signal, and thus arguably far more damaging than merely harmonic distortion. THD testing and dynamic range testing is somewhat akin to assessing how a car performs based on how it does on a dyno, and ending there. It's useful if all you want to do is drag race, but it possible to test a lot more things that will really tell you how the DUT "handles", so use an inapt analogy.

While the CCIF IMD test has no distortion within the audio band--that's not the point of the test. What it does do is tell you whether a lack of linearity is resulting in potentially audible garbage lower in the audio band. And this stuff is real garbage. I think Bruno Putzeys has an article out there somewhere about the nCore and why they substituted CCIF IMD results over 6kHz for the "real" THD results. Those points are valid. Stereophile has used the test for ages. In a 1980 AES paper, Richard Cabot (who I believe started Audio Precision) concluded that "[t]he distortion products generated in this test are usually very far removed from the input signal. This positions them outside the range of the auditory system's masking effects. If a test which measures what the ear might hear is desired, the CCIF test is a good candidate." Arguably, it was the desire to be able to view the results of test like this which precipitated Audio Precision and its spectrum analyzers. You may have read the Cabot article at some point, but if not, you can get it for free on ResearchGate: https://www.researchgate.net/public...n_of_Nonlinear_Distortion_Measurement_Methods. Also useful are Doug Self and Bob Cordell's books on audio power amplifier design. They talk a lot of circuit linearity.

For a practical example of the importance of testing at high frequencies, see https://www.stereophile.com/content/emotiva-xpa-gen3-two-channel-power-amplifier-measurements. Look at all the stuff through Figure 8. Those are basically the tests you run regularly, with the exception of figure 6. And you know what? It appears to ace them. But Fig. 6 is enough to make you go, "Hmmmm....". Now look at the CCIF IMD test: :eek: That's why I think a good review that is trying to link up measurements with audible performance must involve a high power frequency sweep or a CCIF IMD test on every DUT. Else, you're just kicking the tires and racing between stoplights. Useful data, but not enough to make an informed purchasing decision or draw meaningful judgments about sound quality.

Note, though, that the frequency sweep alone is not enough. The nCore doesn't do that well on the sweep, but it still does okay on the CCIF test: https://www.stereophile.com/content/theta-digital-prometheus-monoblock-power-amplifier-measurements. Just because distortion is rising at high frequencies quite a bit, doesn't necessarily mean that the non-linear distortion is going through the roof. Often, but not always. The inverse, though, very rarely occurs.

Now, my practical examples are power amps (and unusual examples I've come across in the past, which I why I mention them here), but the same set of principles holds true on headphone amps or DACs. If you test for this stuff, you will have a far better idea of which products perform and which do not, in ways that are actually sonically meaningful. As interesting as the SINAD drag race is, for 97% of all recorded music, it's largely irrelevant since it's already vastly better than the source material. 90dB or 120dB SINAD will almost never have any audible consequence, in a vacuum. But are any of these great products botching it over 1kHz, and thereby smearing and mucking up stuff lower in the spectrum? For a significant number of products, I really don't know. That holds true for both of the products reviewed here. While this chart: https://www.audiosciencereview.com/...ated-portable-headphone-amps.6312/post-140671 at least suggests both of these products are fairly clean, it's certainly not a conclusion anyone can draw on the basis of the data presented. The only conclusions that can actually be reached on the basis of the data presented is that they both have adequate dynamic range to reproduce CD quality audio, and that they don't suffer from harmonic distortion problems. Whether they suffer from NON-harmonic distortion problems is anyone's best guess.
 
Last edited:

ryanmh1

Member
Joined
Jan 11, 2019
Messages
93
Likes
117
You are the second person to suggest SINAD is not relevant for a DAC, and that perhaps the testing methodology needs to be revised. The first guy was the main designer from Pure Audio UK who stated he would only be sending his new product to Amir if Amir "revised his testing methodology" and stopped using SINAD as a measurement for a DAC (I assume now since most crack 96 dB, and in normal listening environs they don't even need to do that anyway).

SINAD is relevant, but don't forget that SINAD is really nothing more than THD+N at 1kHz, expressed differently. I can understand why a manufacturer wouldn't want to get caught in that drag race. Harmonic distortion just isn't that audible. I'm an objectivist to the core, but my own experience at measuring things vastly more complex than DACs taught me that it is very, very easy to get lost in weeds if you measure the wrong stuff. Well, if sound quality is the concern. If the nth degree of technical proficiency on x parameter is the prime concern, then, well, have at whatever.

I don't think the methodology is bad at all. I just think there are important tests that say a whole lot more about sound quality than THD+N, and there have been a number of products panned or praised for a bad or good THD+N with other characteristics being more or less disregarded. Noise or a high second harmonic isn't going to do much damage if other aspects of the product are exemplary. I'd take a portable with a big output and higher THD+N any day over a clean THD+N with weak output, or one with a really poor IMD figure. So far as the NX3s and the O2, the tests didn't show any potentially audible differences. But that's doesn't mean one product might have a problem. Would CCIF or multitone testing show anything? It's certainly possible. That's why I like the Russian RAA website. There's enough tests there to identify any potential issues. But they've been at this a very long time. Amir's test suite seems to be getting better quite regularly.
 
OP
amirm

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,388
Location
Seattle Area
Amir -- I think you're doing some good work here, but I would argue that dual tone testing can be very related to audibility--far more so than THD testing. High frequency dual tone testing, in particular, can reveal the presence of nasty stuff that is completely unrelated to the musical signal, and thus arguably far more damaging than merely harmonic distortion. THD testing and dynamic range testing is somewhat akin to assessing how a car performs based on how it does on a dyno, and ending there. It's useful if all you want to do is drag race, but it possible to test a lot more things that will really tell you how the DUT "handles", so use an inapt analogy.
My tests include dual-tone Intermodulation tests. It is almost in every review (although I skipped it here). Here is one from the last DAC I just tested:

index.php


The two tones comply with SMPTE IMD test of 60 Hz and 7 kHz:

1548737894693.png


This is the spectrum:

1548738125029.png


This is much more representative of music with higher amplitude for lower frequency. And both signals are readily in the spectrum of real music. This is what I explained in my post.

BTW, the benefit of dual tone testing is NOT that it shows something different than THD. A non-linear transfer function operates equally on one tone or two. CCIF system was designed because in olden days it was difficult to measure harmonic distortion of high frequencies since it extends to tens of kHz. With the dual high-frequency tones in CCIF, you can find the offending distortion products in-band. I have no such restriction in my measurements. Indeed for THD+N versus Frequency I use 90 kHz bandwidth which amply includes harmonic products.

Since many of us can't hear high frequencies let alone their distortion products, I tend to not put as much weight on them. That is another argument in favor of SMPTE IMD test that I do run.
 

ryanmh1

Member
Joined
Jan 11, 2019
Messages
93
Likes
117
My tests include dual-tone Intermodulation tests. It is almost in every review (although I skipped it here).

BTW, the benefit of dual tone testing is NOT that it shows something different than THD. A non-linear transfer function operates equally on one tone or two. CCIF system was designed because in olden days it was difficult to measure harmonic distortion of high frequencies since it extends to tens of kHz. With the dual high-frequency tones in CCIF, you can find the offending distortion products in-band. I have no such restriction in my measurements. Indeed for THD+N versus Frequency I use 90 kHz bandwidth which amply includes harmonic products.

Since many of us can't hear high frequencies let alone their distortion products, I tend to not put as much weight on them. That is another argument in favor of SMPTE IMD test that I do run.

For more, see Leinonen, Otala, and Curl, A Method for Measuring TIM, http://hifisonix.com/wordpress/wp-content/uploads/2010/10/Measuring-TIM.pdf.pdf, and Cordell, A Fully In-Band Multitone Test for TIM. http://www.cordellaudio.com/papers/multitone_test.pdf. It's also fair to note that the SMPTE test was invented in the "olden days" to for the same reasons. It lacks sensitivity to a number of distortion products which the CCIF test catches. The age of the test or why it was invented is not material. Cordell: "If a spectrum analyzer is used... to look at the '2A-B' products ... the [CCIF] test will be sensitive to odd-order products." He also discussed why the SMPTE test is not. The SMPTE test is great, but it's not the end of the story. The CCIF test remains one of the most easily understandable visual methods for showing nonlinear distortion in an audio device.

Oh, well. I've cited enough authorities to make the point, and they (and I) could be entirely wrong, given advances in measurement technology, although I can't explain how that would have changed what types of distortion a particular test does not does not cause. I'll leave you with this, and then call it enough on this topic: https://reference-audio-analyzer.pro/en/report/amp/xduoo-xd-05.php#rw44. Compare the 30.5 ohm SMPTE tests and CCIF test for the Xduoo XD-05 portable after adjusting to the same level. All the garbage is -115dB on the SMPTE test. On the CCIF tests its 10x higher about -95dB. Now do the same exercise for the Sabaj D4: https://reference-audio-analyzer.pro/en/report/amp/sabaj-d4.php#srw27. The CCIF and SMPTE show trash at the same levels. Probably inaudible in both cases, but there are clear differences worth noting in the context of this discussion. I'd love to know whether the same is true for these portables.

Edit: Found back one of Bob Cordell's posts on the topic referring to his old JAES paper: https://www.diyaudio.com/forums/sol...opology-audio-amplifiers-363.html#post3768087. Basically explaining that with spectral analysis, the other test he proposed in the 1980 paper was rendered moot, and the CCIF test reigns supreme. [I'll grant the weird ESS hump is something worth finding, and understanding... That's unusual, and clearly not just the SMPTE test doing what it usually does and mirroring the THD results].

Sorry if I'm coming off too strident. It's your forum, and your gear. I'm just trying to point out some good ways to help achieve what I think you're trying to do. The more rigor and thoroughness in the testing, the less mindless subjectivists will be able to point to obvious flaws, and be forced to admit they just like unpredictable distortion generators. Which is fine. They just need to admit it.
 
Last edited:

Roen

Addicted to Fun and Learning
Joined
Oct 14, 2018
Messages
681
Likes
247
At this point, has anything come out to dethrone the O2 as the best dual mode solution? Did the Grace M900 out-measure it?
 

Cahudson42

Major Contributor
Joined
Sep 21, 2019
Messages
1,083
Likes
1,556
Can anyone recommend a very clean reasonably priced 5v power supply for the NX3s that lets the amp perform when charging as well - or almost as well - as it does on battery?

Or is the poorer performance solely due to the NX3s internal charging circuitry, and a cleaner external supply would be useless?

Edit 10/21/2019 - I asked the above based on the review and comments, anticipating a definite degradation listening when powered - while awaiting my NX3s to arrive.

Now that it has, I can't personally detect a large difference while listening - powered or battery. Both sound to me the same - great! Powered is by an Amazon 1.5a plug, I believe came with a 10" Fire table

The source is the 3.5mm from a Trek K88 tablet (Snapdragon 617, wdc9xx DAC, Dolby enabled flat. Full volume.) NX3s volume with HE400i and low gain falls just short at knob max. On hi gain, knob is 1pm when right for me. Plenty of headroom left. It can be increased to ear-shattering levels, while not - to me - obviously distorting. This is my portable setup, though I will likely replace the K88 with an LG V20. Player is Amazon music HD/UHD - mostly 16/44.1.

Desktop setup is Fire 7, D10, and Liquid Spark. This definitely has greater dynamic range, which I notice on Classical, not vocals. Otherwise, I can't hear much difference.

So far, the NX3s seems about the best $55 I've spent on audio, excepting the $30 for the Fire 7 2019 when it was announced. Very happy with the NX3s. Thank you, Amirm for this review!
 
Last edited:

HeadphoneFan

Member
Joined
Nov 17, 2018
Messages
61
Likes
10
I’m using the nx3s connected to my d30 and can’t seem to hear clipping on high gain. Is my hearing just so bad that I can’t hear it? Or is it reasonable to assume that the nx3s can take 2 volts at high gain without clipping?

The Fiio a5 can put out 845 mw of power at 32 ohms whereas the nx3s can put out 2x 545 mw according to provided specs. Fiio a5 high gain = 13 dB, nx3s high gain= 8.8 db. Fiio a5 input sensitivity is 5.6 volts at low gain and 1.1 volts at high gain. No input sensitivity specs provided for nx3s.

Any thoughts on this? I am contemplating getting a topping a30 but the nx3s can drive my he400i to ear-piercingly loud levels at high gain, connected to a d30. Not sure an a30 is necessary.
 

Skr1231

Member
Joined
Feb 12, 2020
Messages
9
Likes
2
I don't quite understand how you could use NX3 without connecting it with USB to your computer
 

Veri

Master Contributor
Joined
Feb 6, 2018
Messages
9,596
Likes
12,036
I don't quite understand how you could use NX3 without connecting it with USB to your computer
Well it is a battery-powered amplifier, right? It should run with just line input.
 

Veri

Master Contributor
Joined
Feb 6, 2018
Messages
9,596
Likes
12,036
So how do you get the data from PC to convert?
This is Topping NX3, it gets no data from PC. Are you mistaken with other product, Topping NX4 maybe?
It's just an amplifier, not a DAC / converter.
 

Skr1231

Member
Joined
Feb 12, 2020
Messages
9
Likes
2
This is Topping NX3, it gets no data from PC. Are you mistaken with other product, Topping NX4 maybe?
It's just an amplifier, not a DAC / converter.
Oh,sorry,that explains a lot. Thanks
 

Asylum Seeker

Senior Member
Joined
Feb 12, 2020
Messages
414
Likes
295
Location
Guatemala
....

I have a philosophical problem with plugging an iem into a dedicated amp. I know there are very good iem's. I know most would benefit from good, clean, abundant power. But then there's the principal of the matter. A man has to have his principals.

...

I don't understand what the philosophical problem and principle is...
 

Asylum Seeker

Senior Member
Joined
Feb 12, 2020
Messages
414
Likes
295
Location
Guatemala
My testing is proof that those sensitivity numbers/computations just don't work accurately. There is also the issue of running out of gain if the source material is not recorded to max levels.
Agree.
 
Top Bottom