miniDSP 2x4 HD Sweetspot-audio Mod Review

[RandomEar]

Honnestly i dont find that the mods degrade performance. It is true that the noise is marginally imcreased (totally harmless), but this comes with a significant reduction of harmonic above H2. Once again i find noise an overated parameter.

The difference in distortion is irrelevant here because it is below the noise floor (approx. -91 dB) of the modded device. The FFT is somewhat misleading in that regard, because the noise floor looks lower than actually is (-> FFT gain).

This is a lab test, so not all egde cases are checked. But from the measurements shown by amir, we have to assume that you will never get any benefit from that lower distortion because the noise is always dominating. Therefore, this isn't a mod with tradeoffs. It will perform a bit worse than the original in any application and it costs money.

 

[Toni Mas]

The difference in distortion is irrelevant here because it is below the noise floor (approx. -91 dB) of the modded device. The FFT is somewhat misleading in that regard, because the noise floor looks lower than actually is (-> FFT gain).

This is a lab test, so not all egde cases are checked. But from the measurements shown by amir, we have to assume that you will never get any benefit from that lower distortion because the noise is always dominating. Therefore, this isn't a mod with tradeoffs. It will perform a bit worse than the original in any application and it costs money.

Yes but noise is noise and distortion another thing correlated to the Signal, while noise is not... Should make a difference, and if the fft provides detailed information about all the harmonic components, why only retain the strongest one as if the rest were not audible, especially if a sinad 90-100db is not low enough to consider distortion inaudible ?

 

[Sokel]

Yes but noise is noise and distortion another thing correlated to the Signal, while noise is not... Should make a difference, and if the fft provides detailed information about all the harmonic components, why only retain the strongest one as if the rest were not audible, especially if a sinad 90-100db is not low enough to consider distortion inaudible ?

Distortion (depends which distortion of course) can be inaudible high up to 40's.
About -60db is (was) the pro standard for many years and no one complained about it.

Noise on the other hand...

 

[Toni Mas]

Distortion (depends which distortion of course) can be inaudible high up to 40's.
About -60db is (was) the pro standard for many years and no one complained about it.

Noise on the other hand...

Well, low order products like H2 and H3 are generally considered harmless, while higher order products might be quite audible at much lower levels. See Geddes work and proposal of harmonic products weighting. Regarding noise, if it is low enough to be unnoticable, it is low enough... Turntables rumble, vinyl surface noise, tape hiss, psu noise, etc...

 

[mocenigo]

While I generally agree, so many active speakers hiss. I think HD is overrated since so many of us fail to be able to hear distortion until it is large.
I sure can hear noise, although we can argue if it interferes with anything but quiet passages. I showed noise-audibility on the PA5II thread, although the DAC was SOTA. Would have been interesting to see how much worse a 2x4 and a Sweetspot audio version would have degraded!

I almost lost love for music listening when I switched to a used vintage NAIM setup, because of the hiss *and* hum. At first I thought I would get used to it, I was enthusiastic about the bumpy bass, the warm midbass and mid, and "PRaT" and whatnot. Those preamps and amps are absolute crap, but I was in the subjectivist realm then. But, with time, that hiss and the to a minor extent also the hum, started to annoy me more and more.

Then I moved to Abrahamsen Audio — essentially Electrocompaniet "original clones". Much better sound, and in fact I also realised that it was also much more accurate: less distortion and less noise, except for a 50hz+100hz induced by the power supply in the audio signal. Since my speakers were quite sensitive (96Db), this was audible, albeit barely, from the listening position. But they annoyed me. I finally decided to try two NC500 based monoblocks by Apollon (Hi, Tibor!) and I finally was able to focus on the music and on the music only. Now I am 100% sold on powerful, no-noise, no-distortion, amplifiers that do their job, i.e. amplify the signal faithfully, and nothing else. If I hear anything more than the faintest hiss with my ear a few cm close to the 100Db sensitive AMT, or the faintest hum with my ear close to the woofers, and if this noise is still audible from the distance of a foot, I declare (!) that there is something wrong in my system.

Clearly, HD and IMD under 0.1% should not be a problem. Not only it is inaudible per se, but it still submerged by the distortion of the transducers.

 

[RandomEar]

Yes but noise is noise and distortion another thing correlated to the Signal, while noise is not... Should make a difference, and if the fft provides detailed information about all the harmonic components, why only retain the strongest one as if the rest were not audible, especially if a sinad 90-100db is not low enough to consider distortion inaudible ?

Sorry but I don't understand what you're asking.

Noise is noise and distortion is distortion, but if anything, noise is worse. If you have noise at say -40 dB, it will mask most of the signal below and will be very obvious while listening. Distortion at -40 dB will be barely audible for most people, if at all.

Again: All I'm saying is that the assumption that this modded miniDSP is a compromise with some strengths compared to the original is not correct. It will perform worse in any practical application.

 

[RandomEar]

Well, low order products like H2 and H3 are generally considered harmless, while higher order products might be quite audible at much lower levels. See Geddes work and proposal of harmonic products weighting. Regarding noise, if it is low enough to be unnoticable, it is low enough... Turntables rumble, vinyl surface noise, tape hiss, psu noise, etc...

But you will never hear the distortion in this device because it's lower than the noise! All arguments about what type of distortion is preferable are meaningless, because any of it will be drowned in noise on this thing.

 

[mocenigo]

The covid issues probably have made hifi manufacturers create worst and more worst products.

Since it is a mod by a german manufacturer, I would say this it the wurst product.

 

[Toni Mas]

Noise is noise and distortion is distortion, but if anything, noise is worse.

Sorry but don't agree. I can hear tape hiss, vinyl surface noise, etc... without being bothered in excess because though audible it has no correlation with the musical program. Distortion is worse .... when audible, because it spoils the signal itself, and it has to be quite low to avoid a special sound signature to be impressed to the sound.

Btw, most of the time masking does not mean silencing.

 

[Toni Mas]

But you will never hear the distortion in this device because it's lower than the noise! All arguments about what type of distortion is preferable are meaningless, because any of it will be drowned in noise on this thing.

Sorry again but I don't agree. Noise applies some masking effect, but masking does not mean silencing and all the harmonic content matters.

If you mix some music with white noise, what amount of noise will be necessary to get the music totally buried in noise?

 

[Killingbeans]

10dB louder than the music peaks?

 

[RandomEar]

Sorry but don't agree. I can hear tape hiss, vinyl surface noise, etc... without being bothered in excess because though audible it has no correlation with the musical program. Distortion is worse .... when audible, because it spoils the signal itself, and it has to be quite low to avoid a special sound signature to be impressed to the sound.

Btw, most of the time masking does not mean silencing.

I think you're vastly overestimating the dB values of the specific noise and distortion you're talking about in this example. Test yourself at which point distortion stops being audible to you on your system:

Listening Test

www.klippel.de

Then play pink noise at that volume level and compare. I think you'll be extremely surprised how loud that noise will be. On my system (near field listening), I had to push noise far, far below -80 dB to kill audible tweeter hiss. Distortion with real music becomes essentially inaudible to me around -36 dB with the test linked above, maybe -42 dB on a real good day and at elevated SPL. That's easily 40 dB of difference! I encourage you to try it.

 

[Toni Mas]

I think you're vastly overestimating the dB values of the specific noise and distortion you're talking about in this example. Test yourself at which point distortion stops being audible to you on your system:

Listening Test

www.klippel.de

Then play pink noise at that volume level and compare. I think you'll be extremely surprised how loud that noise will be. On my system (near field listening), I had to push noise far, far below -80 dB to kill audible tweeter hiss. Distortion with real music becomes essentially inaudible to me around -36 dB with the test linked above, maybe -42 dB on a real good day and at elevated SPL. That's easily 40 dB of difference! I encourage you to try it.

Thanks for the inputs, i am curious to try.
But i am not discussing noise audibility, but the audibility of a signal under the noise level. I mean same as these -80 dB of noise might be audible in your tweeters why should this noise mask an H3 or H5 at -83dB making these HD products inaudible?

 

[Toni Mas]

10dB louder than the music peaks?

I have just done the following test with Audacity, playing simultaeously
Track 1 White noise amplitude 0.8 fscale
Track 2 sine wave 1.000 hz same amplitude.

Track 2 clearly audible mixed with strong noise.

I attenuate track2 40dB and play again: same strong noise and track2 now inaudible.

I attenuate track2 only 30dB and play again: no noticeable change.

I attenuate track2 only 20dB and play again: i can now clearly hear the 1khz tone though the noise sounds louder.

Conclusion?

 

[olieb]

White noise amplitude 0.8 fscale

What does that mean?
In REW white noise above -9dBfs gets a warning about the signal will be clipping.

 

[Toni Mas]

What does that mean?
In REW white noise above -9dBfs gets a warning about the signal will be clipping.

Never used Rew, though if DSP should be applied it is generally safer to stay under full scale. In this case no processing nor clipping to be considered.

 

[Killingbeans]

Conclusion?

Sounds like what should be expected.

Artifact Audibility Comparisons

This article describes the audibility of audio artifacts.

ethanwiner.com

Try again with music instead of a 1kHz tone, and it probably gets a lot harder. A recording of clear human speech should sit somewhere between that.

 

[MAB]

I think you're vastly overestimating the dB values of the specific noise and distortion you're talking about in this example. Test yourself at which point distortion stops being audible to you on your system:

Listening Test

www.klippel.de

Then play pink noise at that volume level and compare. I think you'll be extremely surprised how loud that noise will be. On my system (near field listening), I had to push noise far, far below -80 dB to kill audible tweeter hiss. Distortion with real music becomes essentially inaudible to me around -36 dB with the test linked above, maybe -42 dB on a real good day and at elevated SPL. That's easily 40 dB of difference! I encourage you to try it.

This is my experience too. The stock MiniDSP is not a high HD device, the value of HD improvement is dubious. It will play a role in noise when used as an active crossover in high sensitivity systems. While a passive filter will reduce noise (like a capacitor on a tweeter, or padded down with resistors), passive is not an option for all applications.

I even demonstrate audible residual noise with a PuriFi amp and a high efficiency compression driver right in the human ear's most sensitive region (not tweeter!!!), and clearly audible noise with amps that have noise roughly equal to this modified MiniDSP. I likely can even tell the amps I used apart blind, for instance the 25 year old Bryston sounds like a harsh buzzing noise to me, different timbre than the other amps, and would really bug me. It occurs to me, I should have tested the noise with a high efficiency midrange since the noise extends down to midrange.

I am not even sure HD went down.
2nd degrades, 3rd improves. 4th degrades from nothing to =120dB, good news for Sweetspot is I won't hear this. Hard to call 1-2dB improvement on 5th harmonic a big win for higher order HD. 7th HD is unchanged, the higher order are better but at going from ~127 to ~130dB down is the limit of the AP, and a parsec away from my ability to hear.

I will admit that I haven't explored the actual audibility of higher order HD, maybe I will get out a VST again and see if I really I can hear higher HD as obnoxious, but I also know I don't operate well at 120dB down. And I can hear noise very clearly, it bugs me since I listen to music with quiet passages where noise intrudes. And Sweetspot Audio has taken a device with great HD and adequate noise and good HD, and made the HD slightly different but not really much better at all, and added noise.

 

[olieb]

Never used Rew, though if DSP should be applied it is generally safer to stay under full scale. In this case no processing nor clipping to be considered.

You did not answer the question. There is no constant amplitude in white noise only statistical fluctuation. If the rms-level would be 0.8 (aka -2dB) then the signal would be clipping big time.
So I assume the noise level is in the range ≤-10dBfs. And you can hear a -20dB tone beneath that. No surprise here that was pretty much the guess of @Killingbeans.
And this is with high level signals. Now try this 60 or 80dB lower.

 

[Toni Mas]

This is my experience too. The stock MiniDSP is not a high HD device, the value of HD improvement is dubious. It will play a role in noise when used as an active crossover in high sensitivity systems. While a passive filter will reduce noise (like a capacitor on a tweeter, or padded down with resistors), passive is not an option for all applications.

I even demonstrate audible residual noise with a PuriFi amp and a high efficiency compression driver right in the human ear's most sensitive region (not tweeter!!!), and clearly audible noise with amps that have noise roughly equal to this modified MiniDSP. I likely can even tell the amps I used apart blind, for instance the 25 year old Bryston sounds like a harsh buzzing noise to me, different timbre than the other amps, and would really bug me. It occurs to me, I should have tested the noise with a high efficiency midrange since the noise extends down to midrange.

I am not even sure HD went down.
2nd degrades, 3rd improves. 4th degrades from nothing to =120dB, good news for Sweetspot is I won't hear this. Hard to call 1-2dB improvement on 5th harmonic a big win for higher order HD. 7th HD is unchanged, the higher order are better but at going from ~127 to ~130dB down is the limit of the AP, and a parsec away from my ability to hear.
View attachment 352497

I will admit that I haven't explored the actual audibility of higher order HD, maybe I will get out a VST again and see if I really I can hear higher HD as obnoxious, but I also know I don't operate well at 120dB down. And I can hear noise very clearly, it bugs me since I listen to music with quiet passages where noise intrudes. And Sweetspot Audio has taken a device with great HD and adequate noise and good HD, and made the HD slightly different but not really much better at all, and added noise.

I simply said that i found a marginal improvement in HD products, nothing special nor worth the upgrade, but as Amir insisted on degradation due to a small increase of noise, i disagreed.

Btw, another test i did with Audacity years ago was mixing a pure 1khz tone with one attenuated pure harmonic from H2 to H7 and had to accept the claims of people like Nelson Pass or Geddes, that higher harmónics are far more audible that H2 or H3