• 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

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
What do you think about possible buffering with a high input impedance buffer, Ovidiu.
 

syn08

Senior Member
Joined
Aug 9, 2019
Messages
376
Likes
461
Location
Toronto, Canada
Not sure how that could work, in current mode. The output impedance would be the same 200ohm for a current amplifier, unity gain buffer (that is, read the ESS output current as a short, output as a close to an ideal current source). The solution to avoid the hump is to use the voltage mode, but that comes to a performance penalty up front (not sure exactly why, BTW, but confirmed by many implementations).

One to another, if I am correct, it is a faulty current output stage design.
 

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
OK, I see. I thought it was voltage mode with 200 ohm output impedance. 200 ohm output impedance of the current source does not make sense to me. I would expect 200 kohm.
 
Last edited:

Skeptischism

Active Member
Joined
Sep 6, 2019
Messages
229
Likes
124
If only some of the modern current out dacs were actually robust current sources. that hasnt been the case for years though. None of the current crop, from any of the manufacturers are over 1kΩ AFAIK
 

syn08

Senior Member
Joined
Aug 9, 2019
Messages
376
Likes
461
Location
Toronto, Canada
AK4499 has 100ohm output impedance per channel, 36mA full scale output current, and no "hump". The I/V common mode input voltage, if different from the 2.5V default, can be provided to the VCOML/R pins to avoid current imbalances in the I/V stages. They do not recommend paralleling the current outputs, but summing the voltages after the I/V stages (yes, more op amps, but at least they are honest about). See the AKM data sheet.

Clearly ESS screwed the output stages and stays mum about. Given the number of DAC products with the "hump", they didn't even care to proactively inform their commercial customers about this pitfall, while they clearly know about, since the demo boards are running @1V Vcm (instead of the default 1.65V).
 

chris719

Senior Member
Joined
Mar 22, 2019
Messages
373
Likes
423
OK, I see. I thought it was voltage mode with 200 ohm output impedance. 200 ohm output impedance of the current source does not make sense to me. I would expect 200 kohm.

Hopefully I'm not getting this wrong because it was buried in a huge thread at diyaudio and I am too lazy find the quotes directly... but I think the designer once said they did not initially set out to produce a current-mode part, but distortion was always lower when they tested it that way. He said it "cancels a slight on-chip resistor voltage coefficient".

I used to think PCM1792 and AD1955 had "hot" outputs... I guess once you parallel enough current sources to get these high SNR numbers you end up with a poor current source anyway, so I can understand why ESS went to a switched resistor output.
 

syn08

Senior Member
Joined
Aug 9, 2019
Messages
376
Likes
461
Location
Toronto, Canada
Quotes from DIYA

dusfor99;2223575 said:
I dont think we ever claimed a current source, but rather a "current mode". The current mode is simply when the current going in and out the pin of the chip is sensed. This mode has the benefit of cancelling 2nd and 3rd harmonics of some of the internal ananlog circuitry. The "voltage mode" is when the pin of the chip has a voltage on it that is being sensed. While this has the 2nd and 3rd hormics (at the -100dB level or so), some people have even claimed this mode is more "tube-like". Its all personal prefference.

dusfor99;1814598 said:
THE [ES9008] output looks like a voltage source with 781.25 Ohm output impedance. "Current mode" is simple when you put the output to a virtual ground and some level (we use 1.65V) then therefore use the current though the 781.25 Ohm in a IV stage. I is not a typical "current source" or "current sink" like the other chips you mention. You could short the output of the chip to ground or AVCC without damaging the chip.

https://patents.google.com/patent/US7116257
 

jkasch

Addicted to Fun and Learning
Forum Donor
Joined
Feb 26, 2021
Messages
756
Likes
4,775
Are there any known issues with the ES9018K2M? (It's what inside my Parasound Integrated.)
 

JSmith

Master Contributor
Joined
Feb 8, 2021
Messages
5,153
Likes
13,214
Location
Algol Perseus

chris719

Senior Member
Joined
Mar 22, 2019
Messages
373
Likes
423
Did you see my plot from Samuel Groner? The distortion does not appear to be stationary.

That was from the ES9008 or ES9018 though, right? I think there was discussion of at least one issue that was fixed in the newer parts. I don't recall that anyone ever repeated Samuel's test, though.
 

scott wurcer

Major Contributor
Audio Luminary
Technical Expert
Joined
Apr 24, 2019
Messages
1,501
Likes
2,821
That was from the ES9008 or ES9018 though, right? I think there was discussion of at least one issue that was fixed in the newer parts. I don't recall that anyone ever repeated Samuel's test, though.

That could be, I still think some time domain plots are more telling than spectra.
 

syn08

Senior Member
Joined
Aug 9, 2019
Messages
376
Likes
461
Location
Toronto, Canada
That was from the ES9008 or ES9018 though, right? I think there was discussion of at least one issue that was fixed in the newer parts. I don't recall that anyone ever repeated Samuel's test, though.

Where is this Groner test documented? I’m willing to give it a shot on the 9038, now is a good moment.
 

chris719

Senior Member
Joined
Mar 22, 2019
Messages
373
Likes
423
Where is this Groner test documented? I’m willing to give it a shot on the 9038, now is a good moment.

I can't remember what the original source was (never read it), but Scott has posted the time domain shot a couple times. He or obviously Samuel would probably know. I did a quick search for the posts but there are a ton to sift through in the Blowtorch thread etc.
 

Pdxwayne

Major Contributor
Joined
Sep 15, 2020
Messages
3,219
Likes
1,172
How many here can actually hear the hump? How to check? I have gustard x16 with the hump. Thx!
 

JohnYang1997

Master Contributor
Technical Expert
Audio Company
Joined
Dec 28, 2018
Messages
7,175
Likes
18,292
Location
China
How many here can actually hear the hump? How to check? I have gustard x16 with the hump. Thx!
The level of the hump on X16 is already quite low comparing to those that were originally revealed. The hump shows up because the noise performance is very good. Granted it can be treated. I don't think it's audible this time.
 

deafenears

Senior Member
Forum Donor
Joined
Feb 2, 2019
Messages
398
Likes
476
@WolfX-700 just published measurements of the SMSL SU-8s, same chip as the X16 (ESS9068), also showing the hump.
 
Top Bottom