• 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!

ESS THD ‘Hump’ Investigation

confucius_zero

Addicted to Fun and Learning
Joined
Nov 22, 2018
Messages
541
Likes
345

Herbert

Addicted to Fun and Learning
Joined
Nov 26, 2018
Messages
527
Likes
435
No one out there to try to cure the hump on the Khadas Tone Board?
I might be wrong here, but: The hump can be cured by a correct
resistor to capacitor ratio in the lowpassfilter attached to the opamp.
To get back to 2V, resistor/capacitor values after the opamp have
to be changed as well.
Or is this approach too simple and I am missing something else in the signal path?
 

Herbert

Addicted to Fun and Learning
Joined
Nov 26, 2018
Messages
527
Likes
435
Amir mentioned in some thread that it will unlikely that "boutique" ES9018 DACS like the DIY Twisted Pear Audio Buffalo
would ever make it to audiosciencereview for testing. Here comes a proof:
https://audiophilestyle.com/forums/topic/58670-does-a-dacs-output-change-as-it-warms-up/
The thread starter put a Buffalo (I assume Buffalo II) through his AP analyzer with bad results.
In retrospective, this seems logica. The analog output stage, called IVY, was designed for general Purpose and not
exclusively for the ES9018...
Somehow good to know as my subjective Impression was that this DAC (with IVY Output stage) was the best DAC I ever listened to.
But I believe in testing/measuring, not listening.
I still own two Khadas boards. Though the hump was obviously not audible, it would be great if someone could figure a mod based on Soncoz / Ben´s findings to minimize the hump on the Khadas. Just for the sake to know that the Khadas board ould be then perfect…
All the best. Herbert
 
Last edited:
  • Like
Reactions: trl

scott wurcer

Major Contributor
Audio Luminary
Technical Expert
Joined
Apr 24, 2019
Messages
1,501
Likes
2,821
Somehow good to know as my subjective Impression was that this DAC (with IVY Output stage) was the best DAC I ever listened to.
But I believe in testing/measuring, not listening.

I hope one sees the problem with statements like this, what if by listening you mean a properly controlled DBT and you still prefer that DAC. Measurement is only half of the story, without a clear hypothesis and experiments to prove it it's not science.
 

Herbert

Addicted to Fun and Learning
Joined
Nov 26, 2018
Messages
527
Likes
435
There is no problem with my statement, as it perfectly reveals HiFi-Vodoo and the tricks of the brain:
My brain says with the Buffalo II:
Whoa, this runs on three shunted Class-A Power supplies,
you shielded every cable, what a warm, clear, strong sound!

With the Khadas:
Oh well, a single power supply, only 5V and there is the hump...

But objectively, the Khadas is better...

I compare this experience with my first live classical concert:
After my ears were used to many well recorded classical recording, I went to a live concert,
Beethoven´s 5th. Harsh, shrill, lack of transparecny - but wait: Ooops, this is how a live
Orchestra sounds in a concert hall that has above average acoustics…

Anyway, anyone out ther t adress the hump on a Khadas?
I asked now several times and I am not able to do the math...
 
Last edited:

Guetta

Member
Joined
Feb 12, 2020
Messages
80
Likes
7
There is no problem with my statement, as it perfectly reveals HiFi-Vodoo and the tricks of the brain:
My brain says with the Buffalo II:
Whoa, this runs on three shunted Class-A Power supplies,
you shielded every cable, what a warm, clear, strong sound!

With the Khadas:
Oh well, a single power supply, only 5V and there is the hump...

But objectively, the Khadas is better...

I compare this experience with my first live classical concert:
After my ears were used to many well recorded classical recording, I went to a live concert,
Beethoven´s 5th. Harsh, shrill, lack of transparecny - but wait: Ooops, this is how a live
Orchestra sounds in a concert hall that has above average acoustics…

Anyway, anyone out ther t adress the hump on a Khadas?
I asked now several times and I am not able to do the math...
I think that on Khadas you will never experience the issue caused by that hump... or maybe you have?
 

eliash

Senior Member
Joined
May 29, 2019
Messages
407
Likes
209
Location
Bavaria, near lake Ammersee
There is no problem with my statement, as it perfectly reveals HiFi-Vodoo and the tricks of the brain:
My brain says with the Buffalo II:
Whoa, this runs on three shunted Class-A Power supplies,
you shielded every cable, what a warm, clear, strong sound!

With the Khadas:
Oh well, a single power supply, only 5V and there is the hump...

But objectively, the Khadas is better...

I compare this experience with my first live classical concert:
After my ears were used to many well recorded classical recording, I went to a live concert,
Beethoven´s 5th. Harsh, shrill, lack of transparecny - but wait: Ooops, this is how a live
Orchestra sounds in a concert hall that has above average acoustics…

Anyway, anyone out ther t adress the hump on a Khadas?
I asked now several times and I am not able to do the math...

Regarding the ESS noise hump, there was some discussion going on before with some other DAC test (don´t remember which one), but my strong suspect to control it, is a clean analog power supply. Clean means really clean, as clean as the output itself, because any noise on the analog power supply comes out 1:1 on the output, e. g. when playing a DC test signal. With an AP tester there is some other test signal used, probably a sine wave. When ramping up DC in a test signal, slowly increasing from 0 to e.g. 0.5 max output and holding it there, the noise on the output increases dramatically, even when using a low noise voltage regulator for that analog supply. It needs to be a low noise opamp sourcing the ESS chip (that is true for the ESS9018s, but I suppose it hasn´t changed for its successors)
 

Herbert

Addicted to Fun and Learning
Joined
Nov 26, 2018
Messages
527
Likes
435
Well, I would lke to trick my brain like this:
Whoa, since some genius at audiosciencereview minimized the hump by swapping the feedback
resistors in the output stage, the Khadas is the best DAC ever.

Seriously gentlemen, I have no analyzer to do this. And I do not understand the math presented
on the Soncoz DACs. But I assume as any opamp has a high input impedance, the math might be tranferable...
https://www.audiosciencereview.com/...ow-to-fix-ess-hump-on-sgd1-and-la-qxd1.10502/
 

LTig

Master Contributor
Forum Donor
Joined
Feb 27, 2019
Messages
5,760
Likes
9,442
Location
Europe
Regarding the ESS noise hump, there was some discussion going on before with some other DAC test (don´t remember which one), but my strong suspect to control it, is a clean analog power supply. Clean means really clean, as clean as the output itself, because any noise on the analog power supply comes out 1:1 on the output, e. g. when playing a DC test signal.
This would be very bad DAC, having a PSRR of 0 dB.o_O
 

Herbert

Addicted to Fun and Learning
Joined
Nov 26, 2018
Messages
527
Likes
435

eliash

Senior Member
Joined
May 29, 2019
Messages
407
Likes
209
Location
Bavaria, near lake Ammersee
This would be very bad DAC, having a PSRR of 0 dB.o_O

This is an ESS design property, just check it out in their specs. They seem to divide the analog power supply voltage into an output voltage. Personally I think there is nothing wrong with that, you just need to respect it. Shunting the low noise regulators from the analog power supplies whith 3 caps (470µF, low ESR) for +/- each, solved the issue to a tolerable amount...
 

BYRTT

Addicted to Fun and Learning
Forum Donor
Joined
Nov 2, 2018
Messages
956
Likes
2,452
Location
Denmark (Jutland)
@Herbert in my opinion on a EU salary it cant pay back for example use 14 days of one's free time time to get Ben1987 mod working on KTB also its not fun working on those micro sizes, see costs below to get a LA-QXD1 that inclusive ballanced outputs and the box, that said on speakers i could not notice any difference between KTB and a ADI-2 Pro set to linear phase LP filter, i'm too lazy test the two in head phones and just live happy with my KTB inclusive all the way from day one..

Herbert_1.png


Herbert_2.png


Herbert_3.png
 

Herbert

Addicted to Fun and Learning
Joined
Nov 26, 2018
Messages
527
Likes
435
@BRYTT Thank you, but I am completely aware of the QXD1.
But I have two Tone boards already (got them even for 75€ each) and do not see the need
to put anything on the pile of electronic waste. The Khadas fits my needs.
I do not need DSD or balanced outputs.
Just plug and play S/PDIF
I am also used to SMD-soldering...

There might be another solution: Simply checking the other "humpless" designs.

Here is Ben´s solution
https://www.audiosciencereview.com/...and-la-qxd1.10502/#lg=attachment43316&slide=0

This is for a balanced design. But besides the 2.2nF on the right, I assume the values in the middle are to compensate for the change in gain caused by 51R/390R/220pF on the right. Can anyone tell me if I am right or wrong?
 
Last edited:

Herbert

Addicted to Fun and Learning
Joined
Nov 26, 2018
Messages
527
Likes
435
Please, gentlemen, is no one out there who could help me with fixing the hump
on the Khadas Tone board?
I know in the meantime better boards/DACs came out
but I am still stuck to the Khadas because because it is small, very DIY-friendly
and only needs 5 V and 120mA.

I.e. It is perfect for upgrading classical CD-Player,
as getting SPDIF from the standardised Sony / Philips circuits
is pretty easy - and with many eighties players it it even fits into the cabinet.

So any help is appreciated...
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
Has the IMD hump proven to be an audible issue ?

The I/V conversion has gotten 7x higher gain so the second stage needs less gain to get the same output voltage.
The filter frequency on the second stage is made different because that of the IV stage also changed considerably.
 

Herbert

Addicted to Fun and Learning
Joined
Nov 26, 2018
Messages
527
Likes
435
We all want to squeeze the best out of our gear, don´t we?

So Ben1987 broadened the bandwith to cure the hump and raised the gain in I/V 7x?

The last time I played with I/V conversion was with a PCM1704, so it´s been a while. I thought he lowered it...
https://www.audiosciencereview.com/...ow-to-fix-ess-hump-on-sgd1-and-la-qxd1.10502/
I have no understanding what the bandwith here means. But looking at Ben´s post, he does not tell what bandwith he did choose.

Once again, here is the output stage of the Khadas and Bens modifications.
So, as a first step, gain has to be raised?
https://www.audiosciencereview.com/...-on-sgd1-and-la-qxd1.10502/page-4#post-315596
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,891
Likes
35,912
Location
The Neitherlands
I can't say this has to be done in all cases. It probably also is related to the used opamp properties.
There are plenty other applications of the same DAC chip that also do not have the IMD hump so possibly there is a standard application.
You will need to have the actual output stage schematic.
Going from pictures won't help as there are no values written on caps.
Either ask the manufacturer or reverse engineer and desolder the caps and measure them.
 

Herbert

Addicted to Fun and Learning
Joined
Nov 26, 2018
Messages
527
Likes
435
I can't say this has to be done in all cases. It probably also is related to the used opamp properties.
There are plenty other applications of the same DAC chip that also do not have the IMD hump so possibly there is a standard application.
You will need to have the actual output stage schematic.
Going from pictures won't help as there are no values written on caps.
Either ask the manufacturer or reverse engineer and desolder the caps and measure them.

Dear Solderdude - there is schematics, exactly in the Image/link I posted:

https://www.audiosciencereview.com/...-on-sgd1-and-la-qxd1.10502/page-4#post-315596

Because of an error I cannot relink them in this post...
Above Bens solution, below the Khadas schematics

And finding the resistors and caps on the board is fairly easy... even with SMD
 
Top Bottom