• 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

bboris77

Senior Member
Joined
Oct 23, 2018
Messages
444
Likes
931
Amir, can you please explain something about these two graphs? They should be portraying the same phenomenon, but it seems to me that the dB scale on the Khadas one is either wrong or it was measured differently, since the -40 dB and -30 dB tones are peaking at -10 dB and 0 dB respectively.


NAD M51 DAC and digital Pre-amplifier USB IMD Spectrum Measurements.png

1539194952817.png
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,381
Location
Seattle Area
Amir, can you please explain something about these two graphs? They should be portraying the same phenomenon, but it seems to me that the dB scale on the Khadas one is either wrong or it was measured differently, since the -40 dB and -30 dB tones are peaking at -10 dB and 0 dB respectively.
Yes, dbr measurements require me manually setting the reference and it seems I did not do that for the first graph.
 

bboris77

Senior Member
Joined
Oct 23, 2018
Messages
444
Likes
931
Yes, dbr measurements require me manually setting the reference and it seems I did not do that for the first graph.
Oh, so the NAD M51 graph is incorrect? I thought it was the other way around.
 
Last edited:

akras

New Member
Joined
Mar 28, 2019
Messages
1
Likes
0
Are there any measurements for any of the affected chips running in the voltage output mode (balanced output, with a basic RC filter for HF common mode noise), is the hump still there?
 
Last edited:

daftcombo

Major Contributor
Forum Donor
Joined
Feb 5, 2019
Messages
3,687
Likes
4,068
Hi,

I might be completely out of the topic, but here is an excerp of Ken Rockwell's measurements of Apogee Duet 2 (which has a ESS Dac inside):

Low-Level Linearity measurements top
A -60 dB to -120 dB glide of a dithered 500 Hz sinewave into noise sounds fine, with no distortion. With the Ultrasone Edition 8 plugged in the front, I could hear a clean 500 Hz tone out to 0:24, or -108 dBFS, in a 16-bit track (CBS CD-1 track 20).
Apogee tells me that the analog levels are all controlled in the analog circuitry, with the DACs always run at full level, however listening to this and other very low-level sine-wave tracks, increased distortion is quite audible when the headphone gain is set around the -30 to -40 dB range with very low-level test tones (especially on CBS CD-1 Track 19, dithered 1 kHz at -70 dB FS), suggesting that Apogee might be getting some of the gain range by truncating the digital signal to the DAC. I don't know; 63 dB attenuators aren't that difficult to implement, and I do know that I tested the frequency response also at -20 and -40 dB attenuation, and it was exactly as it was at 0 dB attenuation, which is critical for professional monitoring.
I don't know, but I do know that the Apogee Duet 2 sounds great with music, and with high-level and with low-level test tones at the levels you'd listen to them, and that test signals are for dweebs. The fact that I can break it with very low-level tones set to a narrow range of lower output levels simply serves me right: so what if the Duet 2 isn't very good at doing something no one buys it to do?
There is no distortion audible with normal test signals or music at attenuated settings; it sounds great!
ref: https://kenrockwell.com/audio/apogee/duet-2.htm

Could that have to do with a ESS hump?

Cheers.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,381
Location
Seattle Area

johan

Active Member
Audio Company
Joined
Feb 23, 2018
Messages
225
Likes
372
We hope that in 2-3 weeks , we will do very extensive testings on Ess hump and getting rid of it . As per H , special attention was given to the clocks , decoupling of the opamps , feedback R (we will try multiple values) and noise on opamp rails (maximum 1uV).. I think key to solve it its a combinations of all of the above especially clocks (I still think that syncronous clock is a must and all DACs whitout the issues seem to be syncronous) and opamp rails. If you look at any DACs that do not show the hump , they have one thing in common . Very good engineering (along with higher price) . Most companies approach implementing Ess SOC with minimum requirements (to be low priced) and get good THD numbers at 1Khz while exhibiting the hump.

This hump , its clearly solvable. The challenge is to make a unit that its competitive with Topping , Khadas .
 

mgmate

Member
Joined
Mar 27, 2019
Messages
55
Likes
24
Johan,

I guess that
We hope that in 2-3 weeks , we will do very extensive testings on Ess hump and getting rid of it . As per H , special attention was given to the clocks , decoupling of the opamps , feedback R (we will try multiple values) and noise on opamp rails (maximum 1uV).. I think key to solve it its a combinations of all of the above especially clocks (I still think that syncronous clock is a must and all DACs whitout the issues seem to be syncronous) and opamp rails. If you look at any DACs that do not show the hump , they have one thing in common . Very good engineering (along with higher price) . Most companies approach implementing Ess SOC with minimum requirements (to be low priced) and get good THD numbers at 1Khz while exhibiting the hump.

This hump , its clearly solvable. The challenge is to make a unit that its competitive with Topping , Khadas .

Thanks Johan,

I guess that this will not be solved via Firmware upgrade... Can you please confirm this point?

Regards,

Manuel.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,368
Likes
234,381
Location
Seattle Area
We hope that in 2-3 weeks , we will do very extensive testings on Ess hump and getting rid of it . As per H , special attention was given to the clocks , decoupling of the opamps , feedback R (we will try multiple values) and noise on opamp rails (maximum 1uV).. I think key to solve it its a combinations of all of the above especially clocks (I still think that syncronous clock is a must and all DACs whitout the issues seem to be syncronous) and opamp rails. If you look at any DACs that do not show the hump , they have one thing in common . Very good engineering (along with higher price) . Most companies approach implementing Ess SOC with minimum requirements (to be low priced) and get good THD numbers at 1Khz while exhibiting the hump.

This hump , its clearly solvable. The challenge is to make a unit that its competitive with Topping , Khadas .
Thanks for the update. We are all waiting holding our breath that you succeed and tell us how. :)
 

Music1969

Major Contributor
Joined
Feb 19, 2018
Messages
4,636
Likes
2,809
I think key to solve it its a combinations of all of the above especially clocks (I still think that syncronous clock is a must and all DACs whitout the issues seem to be syncronous) and opamp rails. If you look at any DACs that do not show the hump , they have one thing in common . Very good engineering

Pro-Ject Pre Box S2 DAC is synchronous but I can't tell from @amirm 's old measurements (before APx555) if the hump is there.

Would be an interesting re-measure if someone can loan amir one (I'm not in North America sadly).
 

IVX

Major Contributor
Joined
Mar 3, 2019
Messages
1,407
Likes
2,771
Location
South of China, SHZ area, - Слава Україні
Hi there, I do not follow that thread but if anyone interested in the root cause of that "hump" phenomenon I have to show something. The attached plot is power sweep vs THD+N my tiny ES9038Q2M USB->DAC->HPA dongle on real planar headphones loaded Fs=44.1kHz 24b. 100MHz named curve correspond reg#0 value 2’b00: MCLK = XI (default), 25MHz at 2’b11: MCLK = XI / 8. As you can see the blue 100MHz curve has no hump at all, and you can ask me why some DACs uses lower MLCK value? Check this out(note: FFT after 40db notch), harmonics distortions
distribution is quite different for 100MHz vs 25MHz, however THD+N very close for both modes. 25MHz mode has nearly perfect harmonics set - 2, 3, 5.. and nothing more, then 100MHz has lot's of small nails at 15, 17 and 19khz. I made some buttonless "easter eggs" hidden interface for my #9038S USB->DAC->HPA dongle to let user switch between 3 modes:
1) Normal-mode 100MHz
2) Performance-mode 25MHz
3) SE-mode 100MHz+.25% of 2nd harmonic9038S_PWSWEEP_HE400I_HUMP.PNG9038S_norm-3dbfs.png9038S_p_mode-3dbfs.png
 

JohnYang1997

Master Contributor
Technical Expert
Audio Company
Joined
Dec 28, 2018
Messages
7,175
Likes
18,292
Location
China
Hi there, I do not follow that thread but if anyone interested in the root cause of that "hump" phenomenon I have to show something. The attached plot is power sweep vs THD+N my tiny ES9038Q2M USB->DAC->HPA dongle on real planar headphones loaded Fs=44.1kHz 24b. 100MHz named curve correspond reg#0 value 2’b00: MCLK = XI (default), 25MHz at 2’b11: MCLK = XI / 8. As you can see the blue 100MHz curve has no hump at all, and you can ask me why some DACs uses lower MLCK value? Check this out(note: FFT after 40db notch), harmonics distortions
distribution is quite different for 100MHz vs 25MHz, however THD+N very close for both modes. 25MHz mode has nearly perfect harmonics set - 2, 3, 5.. and nothing more, then 100MHz has lot's of small nails at 15, 17 and 19khz. I made some buttonless "easter eggs" hidden interface for my #9038S USB->DAC->HPA dongle to let user switch between 3 modes:
1) Normal-mode 100MHz
2) Performance-mode 25MHz
3) SE-mode 100MHz+.25% of 2nd harmonicView attachment 25726View attachment 25729View attachment 25730
Would you try the 2 files I posted? Thanks.
 
Top Bottom