Using a dongle dac to power speakers (no amp)

[amanieux]

added two 56 seconds sound samples comparing dawn 4.4 vs aiyima a07 ( 2 songs + pink noise for each ) :
https://drive.google.com/file/d/1eT-DH1_rj4FC4rk-uwTaPT1BVfm2go4f/view?usp=sharing

fr graph of pink noise on both solutions,

finally a short 5 seconds extract (first dawn4.4, then same 5 seconds from aiyima a07, hard to tell them appart)

 

[IPunchCholla]

added two 56 seconds sound samples comparing dawn 4.4 vs aiyima a07 ( 2 songs + pink noise for each ) :
https://drive.google.com/file/d/1eT-DH1_rj4FC4rk-uwTaPT1BVfm2go4f/view?usp=sharing

fr graph of pink noise on both solutions,

finally a short 5 seconds extract (first dawn4.4, then same 5 seconds from aiyima a07, hard to tell them appart)

Given the impedance differences, that is about what I would expect (from my limited knowledge): increased bass and decreased treble in the dongle.

 

[amanieux]

Given the impedance differences, that is about what I would expect (from my limited knowledge): increased bass and decreased treble in the dongle.

does not show up on the graph, do you hear a difference in trebble and bass in this attached 5seconds wav file between the dongle and the amp ?

 

[IPunchCholla]

does not show up on the graph, do you hear a difference in trebble and bass in this attached 5seconds wav file between the dongle and the amp ?

It is on the graph. Look at the dB difference from your null to the peak of the bass. There is a few dB larger difference on the moon drop. There is also a few more dB roll off in the highlights relative to the mids in the moondrop. If you normalized them at 1 kHZ (and overlayed them), this would be more obvious.

 

[amanieux]

I just tested this with my old Fiio Olympus 2 and it worked pretty well
Of course I would not use my system in this way (simply because I need way more power) but as an experiment it is great!

I can imagine it could work perfectly for a DIY 'head-speaker' using 1-2 inch full range drivers

did sound scale up well on your e10k ? on dawn 4.4 above 80% sounds gets harsh and i hear less bass

 

[ppataki]

did sound scale up well on your e10k ? on dawn 4.4 above 80% sounds gets harsh and i hear less bass

It was the same with the Fiio too
I guess there isn't enough power

 

[DVDdoug]

If it works for you, great! But you are potentially "stressing" the headphone output so it's not something I'd recommend to others...

Hifi experts told me to not use headphone amps to power speakers there would be impedance mismatch that would screw up my frequency response

You don't want to "match" the impedance. The output impedance should be low relative to the load and a 1-Ohm output impedance into 8-Ohms isn't too bad. 4-Ohm speakers is "pushing it" but the impedance might still be OK,

The real (potential) problem is - Impedance and resistance are "the resistance to current flow." At 8-Ohms you are "pulling"4 times as much current as with 32-Ohms, or 8 times as much at 4-Ohms, and more than it's designed for.

See my attached video, sound is cleaner than going through my aiyima a07

Of course it sounds "clean"! It should sound clean with the amplifier too (as long as the amp isn't defective or over-driven) but when you lower the volume you also lower the distortion so any distortion becomes harder (or impossible) to hear. You also won't hear all of the little-subtle musical details. An amplifier will amplify any existing noise, plus it will add SOME noise and distortion of its own. It doesn't necessarily audibly degrade the sound but the mythical "straight wire with gain" doesn't exist.

An Analogy - If you have a 16-bit file and you attenuate it by 48dB, you are only using 8-bits so you only have 8-bits of resolution. (This is why people argue against digital volume control.) But you don't hear any loss of resolution. You also lose resolution during quiet passages or during a fade-out. At the end of a fade-out you could be down to 1 or 2 bits but it's so quiet you can't hear ANYTHING so you can't hear the terrible resolution!

If you have 8-bits of usable resolution and you re-amplify it (digitally or electronically) you WILL hear the loss of resolution as quantization noise.

 

[amanieux]

It is on the graph. Look at the dB difference from your null to the peak of the bass. There is a few dB larger difference on the moon drop. There is also a few more dB roll off in the highlights relative to the mids in the moondrop. If you normalized them at 1 kHZ (and overlayed them), this would be more obvious.

here is the graph of the 5 seconds extract normalized, you are right there is a differemce in trebble on the graph, about the difference in bass. not so sure because my b&w cdm2 used in this test don't go below 70hz

 

[amanieux]

If it works for you, great! But you are potentially "stressing" the headphone output so it's not something I'd recommend to others...


You don't want to "match" the impedance. The output impedance should be low relative to the load and a 1-Ohm output impedance into 8-Ohms isn't too bad. 4-Ohm speakers is "pushing it" but the impedance might still be OK,

The real (potential) problem is - Impedance and resistance are "the resistance to current flow." At 8-Ohms you are "pulling"4 times as much current as with 32-Ohms, or 8 times as much at 4-Ohms, and more than it's designed for.


Of course it sounds "clean"! It should sound clean with the amplifier too (as long as the amp isn't defective or over-driven) but when you lower the volume you also lower the distortion so any distortion becomes harder (or impossible) to hear. You also won't hear all of the little-subtle musical details. An amplifier will amplify any existing noise, plus it will add SOME noise and distortion of its own. It doesn't necessarily audibly degrade the sound but the mythical "straight wire with gain" doesn't exist.

An Analogy - If you have a 16-bit file and you attenuate it by 48dB, you are only using 8-bits so you only have 8-bits of resolution. (This is why people argue against digital volume control.) But you don't hear any loss of resolution. You also lose resolution during quiet passages or during a fade-out. At the end of a fade-out you could be down to 1 or 2 bits but it's so quiet you can't hear ANYTHING so you can't hear the terrible resolution!

If you have 8-bits of usable resolution and you re-amplify it (digitally or electronically) you WILL hear the loss of resolution as quantization noise.

dawn 4.4 level is 60% (but sinad is 117db), aiyima a07 level is 5% for matching loudness (but sinad is 82 db) so how do we compare how "clean" they are respectively ? noise floor from my samples certainly come from zoom h1 (recording gain at 60%) because it is a entry level recorder, with my ears i could not hear any noise coming from both

about loosing resolution because volume is low, my ears are 30-40cm away from the speakers so don't close distance compensate for lower volume (same perceived loudness as higher volume but listener further away)

about stressing the headphone output, you may have a point as the specs don't mention max or min impedance supported (iems this product is designed for are typically between 16 and 32 ohms so 8 ohms is not a typical impedance) but it was still cold to the touch after a few minutes test at 60%

 

[JayGilb]

Seems likes an interesting experiment. Have fun !

 

[amanieux]

Seems likes an interesting experiment. Have fun !

yes but not only for curiosity, it is also to understand if we need an amp if we only use speakers near field ( i know some people listen to music on a couch in a large living room so amps are required but i spend 8 hours a day sitting at a desk with a laptop so this where i listen to music and i must not be the only one in such a position and speakers offer much a better experience than headphones or iems on the crucial soundstage and imaging aspects of music reproduction, with speakers music is all around you, with headphones and iems music is only between your two ears, i also happen to live in a appartment with neighbours all around and i also guess i am not the only one in such situation so with near field (or super near field in my use case) you can lower volume for the same perceived loudness and music details )

finally removing the amp from the audio pipeline not only reduces quality loss as you remove an element in the chain of degradation but it also reduces cost so you can put more money where it matters

the curious thing is that studio monitors designed for near field (so low volume) have powerful amps, it does not make sense unless their drivers are very hard to drive which is quite unlikely

 

[IPunchCholla]

here is the graph of the 5 seconds extract normalized, you are right there is a differemce in trebble on the graph, about the difference in bass. not so sure because my b&w cdm2 used in this test don't go below 70hz

The low end is less than I was expecting, but there is a subtle increase from 200 to 600. Not sure if it would be audible or not, but in several spots looks like about 1 db more.

 

[VintageFlanker]

dawn 4.4 level is 60% (but sinad is 117db), aiyima a07 level is 5% for matching loudness (but sinad is 82 db)

Doesn't work that way. Like, at all... The A07 has 82dB SINAD under 5W/4Ω/1kHz. The Moondrop, "117dB" at full output, with no load. Apples and oranges.

Also, "matching loudness" isn't a thing, unless being verified to be at the same output level with a DMM.

on dawn 4.4 above 80% sounds gets harsh and i hear less bass

... because it's clipping under such circumstances, which is to be expected when used improperly like here. By the way, "80%" also means very little when we don't know the conditions (continuous full scale sine wave? I pretty much doubt it). If it's a musical signal : between x or y recording the differences could be just enormous.

 

[amanieux]

Doesn't work that way. Like, at all... The A07 has 82dB SINAD under 5W/4Ω/1kHz. The Moondrop, "117dB" at full output, with no load. Apples and oranges.

Also, "matching loudness" isn't a thing, unless being verified to be at the same output level with a DMM.


... because it's clipping under such circumstances, which is to be expected when used improperly like here. By the way, "80%" also means very little when we don't know the conditions (continuous full scale sine wave? I pretty much doubt it). If it's a musical signal : between x or y recording the differences could be just enormous.

i also expected it to clip at some point, but at 60% which matches practical loudness ( what you can hear in the audio sample) it does not clip, that was suprising, don't you think so ?

 

[amanieux]

It can barely do 100 mW into 16R, at 8R is probably half of that. Distortion wise your also nowhere hear the DAC’s optimum, you probably get around a 90 SINAD before clipping. Still respectable, but with only 50 mW, it’s not going to be enough for speakers: It will get you to about 81 dB peak if your lucky. With about 15 dB crest factor that would mean average volume would be about 65 dB at best.

average volume seems to be more like 75db, remember that my ears are only 1 feet away ( the closer you are the lower you volume can be and that is good to limit untreated room reflections, the limit on a 2 way speaker is that you have a minimum distance for the tweeter to merge with the woofer and it seems to be about 1 feet

 

[MaxwellsEq]

dawn 4.4 level is 60% (but sinad is 117db), aiyima a07 level is 5% for matching loudness (but sinad is 82 db) so how do we compare how "clean" they are respectively ? noise floor from my samples certainly come from zoom h1 (recording gain at 60%) because it is a entry level recorder, with my ears i could not hear any noise coming from both

What is the background noise level at your listening position - ideally measured C-weighted slow response sound meter. Alternatively record the background noise with your normal device and present this on the same graph scale.

 

[amanieux]

What is the background noise level at your listening position - ideally measured C-weighted slow response sound meter. Alternatively record the background noise with your normal device and present this on the same graph scale.

i forgot to do this test for background noise, all i can tell is that 1 feet away my jbl 305 in the same setup has a very audible tweeter background hiss but in this test i could hear nothing even putting my ear 5 cm away from the drivers

 

[MaxwellsEq]

i forgot to do this test for background noise, all i can tell is that 1 feet away my jbl 305 in the same setup has a very audible tweeter background hiss but in this test i could hear nothing even putting my ear 5 cm away from the drivers

I didn't mean when the speakers are powered. I meant - turn everything off and measure the noise where you sit. It's a good idea to do two different times such as mid morning and late evening.

This will establish what the minimum noise you can detect from your electronics and the maximum dynamic range you can realistically expect.

 

[AnalogSteph]

Anything suitable for driving IEMs without driving you insane with hiss has to have vanishingly low noise in normal-sensitivity speakers.

The hearing threshold for independent L/R white noise in headphones is about 20 dB SPL (and arguably higher in IEMs due to noise resulting from blood flow). Something that is subjectively noise-free in 125 dB / 1 V IEMs has to fare better than -105 dBV (~5.6 µVrms, or 5 dB lower noise than a Benchmark AHB-2 which scores 113 dB SINAD on Amir's 5W / 4 ohm chart, only the Topping LA90(D) at 120 dB could beat that).

The hearing threshold for white noise in speakers (using both ears) in decently quiet surroundings is about 4 dB SPL. A regular old 85 dB / W 4 ohm speaker has a sensitivity of 88 dB @ 1 m ref. 2.83 Vrms, or 79 dB SPL / 1 Vrms @ 1 m, or 89 dB SPL / 1 Vrms at 1'. It would hit audibility at 3 cm (!), at which distance however you would have to resort to sticking your ear to the tweeter which brings the hearing threshold up substantially (at least said 20 dB SPL for headphones).

I didn't mean when the speakers are powered. I meant - turn everything off and measure the noise where you sit. It's a good idea to do two different times such as mid morning and late evening.

This will establish what the minimum noise you can detect from your electronics and the maximum dynamic range you can realistically expect.

The microphones available to regular consumers are not terribly suited for that kind of task, and then you still have to interpret the result correctly even if you have access to say a RØDE NT-1. You might see 30-40 dB Z-weighted but that in no way means that you won't be able to detect 4 dB SPL white noise.