• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). There are daily reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Topping D90LE - "Valve" and "Transistor" effects measurements

Gorgonzola

Addicted to Fun and Learning
Forum Donor
Joined
Jan 27, 2021
Messages
784
Likes
1,035
Location
Southern Ontario
The the debate on some subjectivist audiophile sites is whether 2nd & 3rd order distortion at levels seen here in 'Valve' and 'Transistor' modes is (a) somehow audible per se, or (b) these levels affect sound only by '"masking" nasty, higher order harmonics. Those of the former opinion cite findings, (I don't say fact), that 2nd order distortion is inaudible below 2-3%.

In case of the D90LE, distortion & noise are very low apart, and that includes higher order harmonics, from the 2nd & 3rd order purposely introduced. Thus using the D90LE's facilities a (subjectivist) audiophile might go some ways towards resolving this question, at least to his/her own satisfaction. Although, of course, up and downstream components had better be low distortion as well.

My money, (at short odds), would be on experienced, critical audiophiles hearing some effect from 2nd and/or 3rd order at the levels produced by the D90LE. This might or might not be provable with blind ABX testing.
 

Toni Mas

Active Member
Joined
Jun 9, 2022
Messages
124
Likes
113
We should be aware that "typical tube sound" (whatever that is) from a real hardware amp might be more than just stable and frequency independant harmonic distortion which is what that DAC emulates, most likely with the H2/H3 distortion compensation of the ESS DAC chip used (as does the ESS version of the RME ADI-2 DAC).
I am also curious to know how these harmonics are created depending on level. In other words is the clipping behaviour of tubes ( more or less soft clipping or hard clipping) is also simulated, or the same % of harmonics is added at any level?
 

lashto

Addicted to Fun and Learning
Forum Donor
Joined
Mar 8, 2019
Messages
841
Likes
450
Nice feature, potentially simulating "tube distortion" while retaining high SNR.
Surely nice to have but this is more like a half-feature.
The "real deal" will be to have full access to the H2/H3 levels so the user can test & chose as he pleases. Having a few predefined presets is only a good first step
 

KSTR

Major Contributor
Joined
Sep 6, 2018
Messages
2,274
Likes
5,010
Location
Berlin, Germany
Surely nice to have but this is more like a half-feature.
The "real deal" will be to have full access to the H2/H3 levels so the user can test & chose as he pleases. Having a few predefined presets is only a good first step
You can do this with the ESS version of the RME ADI-2 DAC FS using different presets, see https://www.rme-audio.de/downloads/adi2dacr_e.pdf , page 28
 

KSTR

Major Contributor
Joined
Sep 6, 2018
Messages
2,274
Likes
5,010
Location
Berlin, Germany
I am also curious to know how these harmonics are created depending on level. In other words is the clipping behaviour of tubes ( more or less soft clipping or hard clipping) is also simulated, or the same % of harmonics is added at any level?
Harmonics generation always depends on level. It's a simple static transfer function alteration (thus frequency independant, but level-depedant).
Soft clipping is not implemented.
 

Fidji

Active Member
Joined
Feb 17, 2019
Messages
170
Likes
338
Location
Firenze, I
Interesting and informative. I still wonder how would you choose a mode that has 10exp6 more distortion on a fine tuned DAC. Is it maybe to justify your previous $20K spent on a tube preamp/amp combo that glows in the dark and sounds horrible?

No, you are doing here the audiophile finetuning of the system.

You have "warm" tubes for 20k then you obviously need to engage Transistor
You have "cold" AB - cured by Tube mode.

Don't ask me why. You have people adjusting the sound with warm or cold ethernet cables. e.g rack with glass is often described as "cold sounding", one with the wood painted to dark colors - yes, you guessed it - "warm"

And you will not find too many guys with 20k tubes buying Topping - for them it is "analytical, liveless sound"
 

ZgSTar

Member
Joined
Oct 26, 2019
Messages
58
Likes
58
I always thought it would be a cool product idea for someone to release a custom distortion generator. Ideally digital in and out so it could be placed in-line before your DAC but analog could also work.

Something that lets you dial in specific 2nd, 3rd, 4th, 5th, etc order distortion just to play with how it effects sound.
Plenty of AU plug-ins for that if you want to play with it!
 

lashto

Addicted to Fun and Learning
Forum Donor
Joined
Mar 8, 2019
Messages
841
Likes
450
Senseless to add H2/h3 to an audio signal without being able to configure transients which are what makes a signal become (real, natural, analog) music.

oh, the wonderful "transients". Everyone talks about those lately but nobody seems to be able to say what exactly they are. Even the wiki page you linked says that the acoustic transients are an "idea":
Transient response – a common electrical engineering term that may be the source of the idea of an acoustic "transient"

Saying something like "transients are what makes a signal become music" is just a giant leap of faith. (not saying it's not true, just that there is no proof either way)
And before you get a transients-setting in your DAC you'll need a precise (technical) definition of the transients, a way to measure those, studies to clarify their exact (audible) effects and so on... Let's just say that the transients-setting will not happen tomorrow :)

P.S.
or maybe I am missing something and someone can link a few papers/studies that already clarified those "transients".
 
OP
R

Rja4000

Major Contributor
Forum Donor
Joined
May 31, 2019
Messages
1,957
Likes
2,498
Location
Liège, Belgium
I am also curious to know how these harmonics are created depending on level.

Here you go:
As you see, below, the level of the Second Harmonic increases with the DAC level:

2022-06-23 18_01_25-Greenshot.png


Or, using a more "classical" view (similar to Amir's IMD plot), with THD in %,:
THD % should decrease proportionaly to signal, until it starts raising above the noise.

Here-below, we see that, in "Valve" mode, it increases faster than level:

2022-06-23 18_06_35-Greenshot.png
 
Last edited:

Toni Mas

Active Member
Joined
Jun 9, 2022
Messages
124
Likes
113
Here you go:
As you see, below, the level of the Second Harmonic increases with the DAC level:

View attachment 214277

Or, using a more "classical" view (similar to Amir's IMD plot), with THD in %,:
THD % should decrease proportionaly to signal, until it starts raising.

Here-below, we see that, in "Valve" mode, it increases faster than level:

View attachment 214278
Nice, thanks!
I was asking because the measurements shown were taken close to 0db full scale, and Imo valve sound best if not "overdriven" and with a lot of dynamic range.
And btw, clipping is not simulated.
Also state of the art "sinad Champions" are not so brillant in the -40 -20 db range, and distortion decreasing with level maybe kind of weird, at least no really deserves the "transparent" term...:rolleyes:
 
Last edited:
OP
R

Rja4000

Major Contributor
Forum Donor
Joined
May 31, 2019
Messages
1,957
Likes
2,498
Location
Liège, Belgium
distortion decreasing with level maybe kind of weird
It's just the way we represent it.

As you see in the first plot above, the H2 remains mostly at the same absolute level, whatever the signal level, in "standard" mode.
Only in the highest 20dB or so, you see a slow increase.
That's what makes this DAC exceptionally good !

What we see as H2 below -20dB is, in fact, the limit of what we can detect in the noise.
The noise floor level is almost 30dB ABOVE the H2.

That's the reason the THD % decreases with the level :
The noise hardly changes, so our detection threshold remains the same... but the signal increases. So our % of threshold level vs signal level decreases.

It's only when the actual distortion level starts to increase enough to be more clearly distinct from noise that we see the plot curving up.

And here, in "standard" mode, we hardly see that.
 
Last edited:

Toni Mas

Active Member
Joined
Jun 9, 2022
Messages
124
Likes
113
It's just the way we represent it.

As you see in the first plot above, the H2 remains mostly at the same absolute level, whatever the signal level, in "standard" mode.
Only in the highest 20dB or so, you see a slow increase.
That what makes this DAC exceptionally good !

What we see as H2 below -20dB is, in fact, the limit of what we can detect in the noise.
The noise floor level is almost 30dB ABOVE the H2.

That's the reason the THD % decreases with the level :
The noise hardly changes, so our detection threshold remains the same... but the signal increases. So our % of threshold level vs signal level decreases.

It's only when the actual distortion level starts to increase enough to be more clearly distinct from noise that we see the plot curving up.

And here, in "standard" mode, we hardly see that.
Yes it's true the absolute level of H2 is almost constante, hence falls the %thd
 

Vini darko

Major Contributor
Joined
Jun 1, 2020
Messages
2,167
Likes
3,173
Location
Dorset England
Suprised topping is adding this functionality , seems contrary to thier whole usp.
I suppose it's keeping up with the Jones's (smsl) who have had this feature in thier higher price dac's for a long time (sound colour)
 

Toni Mas

Active Member
Joined
Jun 9, 2022
Messages
124
Likes
113
Suprised topping is adding this functionality , seems contrary to thier whole usp.
I suppose it's keeping up with the Jones's (smsl) who have had this feature in thier higher price dac's for a long time (sound colour)
The problem with perfection is that It cannot be an usp... Perfection just kills the business... You need to add some kind of topping...;)
 

PeteL

Major Contributor
Joined
Jun 1, 2020
Messages
2,749
Likes
2,953
oh, the wonderful "transients". Everyone talks about those lately but nobody seems to be able to say what exactly they are. Even the wiki page you linked says that the acoustic transients are an "idea":


Saying something like "transients are what makes a signal become music" is just a giant leap of faith. (not saying it's not true, just that there is no proof either way)
And before you get a transients-setting in your DAC you'll need a precise (technical) definition of the transients, a way to measure those, studies to clarify their exact (audible) effects and so on... Let's just say that the transients-setting will not happen tomorrow :)

P.S.
or maybe I am missing something and someone can link a few papers/studies that already clarified those "transients".
I will not really go and dig for white papers, but I have been working audio and engineering for a long time and it's not some vague concept. It's just a very short percussive peak now how short does it need to be to be a transient yeah maybe it needs clarifying, but anyway, It's the idea, if you look at a music file, typically the loudest parts will be percussive peaks and bass notes, bass notes because of the response of the ear that force mixing engineers to push those, and transients. The thing is, transients being so short, if they are clipped you won't hear that has obvious distortion, It's just that it will suck out the dynamics, the music will be "compressed"or "limited".
 

pma

Major Contributor
Joined
Feb 23, 2019
Messages
3,616
Likes
7,712
Location
Prague
Another important aspect is that this added distortion can easily lead to alisasing components for high frequency signals which obviously a real hardware does not do. I have not checked whether the ESS compensation is free of this but I do know that (especially older) plugins generate significant aliasing as they do not properly apply the required upsample&filter --> distort --> filter&downsample process
Yes. And they add non-existing IMD products - try to apply some distort SW to 19+20k IMD , make H2 only and see the result.
 
Top Bottom