• 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!

Both DAC are transparent as measures by Amir, why one not as clean sounding as the other?

b4nt

Addicted to Fun and Learning
Joined
May 29, 2021
Messages
803
Likes
270
What exactly is top end?
: )

Anyway, brand and model of your speakers? Your amp?

Does the amp mather if used at between <0.05% <0.01% ? Is this audible? Should I ABX?

I was told my Dynaudio Emit M20 are very nice. They are. You just listend to their detailed highs, catched by AKG mics. Would need to do better capture for med/bass, they are very nice also :)
 
OP
Pdxwayne

Pdxwayne

Major Contributor
Joined
Sep 15, 2020
Messages
3,219
Likes
1,172
Does the amp mather if used at between <0.05% <0.01% ? Is this audible? Should I ABX?

I was told my Dynaudio Emit M20 are very nice. They are. You just listend to their detailed highs, catched by AKG mics. Would need to do better capture for med/bass, they are very nice also :)
My $150 Pioneer speakers designed by Andrew Jones also have nice highs.
; )

When I see the word "top end", I imagine it as the top of the line speakers from a major manufacturer. For example, my Paradigm Persona B is "top end" from Paradigm.

; )






.............................................................
Latest summary of comparing original test song:
https://www.audiosciencereview.com/...clean-sounding-as-the-other.24778/post-861227


Comparing another 2 minutes song:
https://www.audiosciencereview.com/...clean-sounding-as-the-other.24778/post-861652
 
Last edited:

b4nt

Addicted to Fun and Learning
Joined
May 29, 2021
Messages
803
Likes
270
@Pdxwayne May you please record at speaker/listener side and share with us what is so bad with the D30 pro?

I'd trust a pair of AKG studio mics :)
 
OP
Pdxwayne

Pdxwayne

Major Contributor
Joined
Sep 15, 2020
Messages
3,219
Likes
1,172
@Pdxwayne May you please record at speaker/listener side and share with us what is so bad with the D30 pro?

I'd trust a pair of AKG studio mics :)
I don't have any good mic. As mentioned, d30pro is not bad. Just "felt" different when brand new. All the null tests, although d30pro nulled the worse, points to minor differences in dB that most people can't sense anyway.



.............................................................
Latest summary of comparing original test song:
https://www.audiosciencereview.com/...clean-sounding-as-the-other.24778/post-861227


Comparing another 2 minutes song:
https://www.audiosciencereview.com/...clean-sounding-as-the-other.24778/post-861652
 

b4nt

Addicted to Fun and Learning
Joined
May 29, 2021
Messages
803
Likes
270
I don't have any good mic. As mentioned, d30pro is not bad. Just "felt" different when brand new. All the null tests points to very minor difference.

You found a probe using DW that comfort your impressions.

If I tweak my speakers probing using AKG mics, I think I would end demonstrating they are acceptable.
 

alek19000

Member
Joined
Jun 27, 2021
Messages
25
Likes
12
Unless experts here object, based on latest USB captures, Node2i Coax captures, and files comparisons after trimming (cut 2 seconds from begin and from end), I am going to conclude the findings as:

*d30pro is not as stable as X16
*d30pro vs orig contains instances where null difference reached potential audible levels, whereas X16 is about 10db better, thus unlikely to have audible difference vs orig.

I read from start to this place . thanks for your work!!!
I am from a software design background, but in the field of cellular wireless communication SOC product.

My guess is : the clock drifting issue, as you test from different USBs, because USB interface needs local clock to demodulate digital signal too, that is the main issue for D30PRO: bad usb affects usb signal receiving.

Then you switched to a better usb port , the main issue became to: d30pro have worse null test against orignal file than others, ( but not in the case against it self in multiple runs),
this may be the design of CS43198, it has its own PLL build on chip , to handle syncing , while other dacs depends on descrete devices on board.
and D30PRO has 4 cs43198, I wonder how the clock syncing among those, if they use pll on each own chip, and chip varies on temperature, production differences , it may cause unsync signals and combing them after analog phase may cause unstable result.
 
Last edited:

b4nt

Addicted to Fun and Learning
Joined
May 29, 2021
Messages
803
Likes
270
The 4 cs43198 are combined to lower the noise in the outputs.
In the D90se, 8 chips are combined.

If that wouldn't work, Amir's tests should show it.
 

alek19000

Member
Joined
Jun 27, 2021
Messages
25
Likes
12
The 4 cs43198 are combined to lower the noise in the outputs.
In the D90se, 8 chips are combined.

If that wouldn't work, Amir's tests should show it.
yes and no.
the other chips are without dpll function , so combine them you need a outside clock signal and dpll to produce a single clock signal to all chips and path , so they are all synced in digital signal.
combine them is like linear time invarant system ,
you add the signal and noise , because the noise is white and adding them will not double the level , you can have " gain" to increase SNR.
it is a common technique in wireless communication, most important one in 3G system, in which Qualcomm invented RAKE receiver which can combine signals from different path and base station. and many many other like HARQ later on.

but for cs43198 it has own DPLL function on each chip which means it may differ for each chip in clock syncing , basically there is an independent clock source for each chip,if these chip's dpll behave differently. and because dpll process needs time to adjust frequency dynamically and lock phase , and it is like a negative feedback process, it breaks the linear time invaraint rules of the singal and system textbook . so adding them adds up the timing errors.


I read from some tearing up d30pro thread , they say the 4 chips are fro L+ L- R+ R - in xlr , so the combine phase is in the amp.

I suggest to test the RCA out, which only use one chip for L and one for R channel, if it is better, my guess is verified!!!
 
Last edited:

b4nt

Addicted to Fun and Learning
Joined
May 29, 2021
Messages
803
Likes
270
the other chips are without dpll function

The D90se chips also have PLLs, for specific applications.

In the cs43198, the PLL may be unused for PCM 441/48 multiples input signals:

cs43198PLL.PNG
 

alek19000

Member
Joined
Jun 27, 2021
Messages
25
Likes
12
The D90se chips also have PLLs, for specific applications.

In the cs43198, the PLL may be unused for PCM 441/48 multiples input signals:

View attachment 145254
For D90se ES9038PRO, the clock can be set to one master chip , and for D70S it has two expensive ACCUSILICON clock source, I guess AKM doesn't have DPLL.
For d30pro , DPLL must be turned on for each chip, which means they operate independently.
 

Attachments

  • 200454.png
    200454.png
    150 KB · Views: 71

b4nt

Addicted to Fun and Learning
Joined
May 29, 2021
Messages
803
Likes
270
If you feed the same clock to 4 cs43198, they shall be in sync also.
 

alek19000

Member
Joined
Jun 27, 2021
Messages
25
Likes
12
If you feed the same clock to 4 cs43198, they shall be in sync also.
Theoretically yes!!!
But here we are in this situation, that E30 outperforms newer D30PRO and I can't think of any other explanation for topping to have this issue.

I am not an expert in DPLL ,I checked this: Phase-locked loop - Wikipedia , and it shows DPLL process needs time to get stable, and may get different stable point for different runs ,as the clock source drifts. So 4 independent DPLL process may lead to this problem: time error adds up .
 

b4nt

Addicted to Fun and Learning
Joined
May 29, 2021
Messages
803
Likes
270
But I do not think the D30 switches on the plls, they seem to be forseen for non 441/48k multiples:

cs43198PLL.PNG
 

alek19000

Member
Joined
Jun 27, 2021
Messages
25
Likes
12
But I do not think the D30 switches on the plls, they seem to be forseen for non 441/48k multiples:

View attachment 145258
from the tear up thread here: [心得] Topping D30 PRO 開箱&拆機 - 看板 Headphone - 批踢踢實業坊 (ptt.cc)

there are no standard 22.5or 24.5 oscillator on board, it is using an Altera CPLD 5M80ZE64C5N programmed oscillator , which I don't think to be 22.5or 24.576MHZ necessarily.
So , "the PLL must be turned on to provide the desired internal MCLK."
I think this feature from cs is for convenient design for like mobile phones, which has different oscillator for RF, and no need for another one for audio.
 

Attachments

  • e2EIgGHl.jpg
    e2EIgGHl.jpg
    111.4 KB · Views: 92
Last edited:
OP
Pdxwayne

Pdxwayne

Major Contributor
Joined
Sep 15, 2020
Messages
3,219
Likes
1,172
yes and no.
the other chips are without dpll function , so combine them you need a outside clock signal and dpll to produce a single clock signal to all chips and path , so they are all synced in digital signal.
combine them is like linear time invarant system ,
you add the signal and noise , because the noise is white and adding them will not double the level , you can have " gain" to increase SNR.
it is a common technique in wireless communication, most important one in 3G system, in which Qualcomm invented RAKE receiver which can combine signals from different path and base station. and many many other like HARQ later on.

but for cs43198 it has own DPLL function on each chip which means it may differ for each chip in clock syncing , basically there is an independent clock source for each chip,if these chip's dpll behave differently. and because dpll process needs time to adjust frequency dynamically and lock phase , and it is like a negative feedback process, it breaks the linear time invaraint rules of the singal and system textbook . so adding them adds up the timing errors.


I read from some tearing up d30pro thread , they say the 4 chips are fro L+ L- R+ R - in xlr , so the combine phase is in the amp.

I suggest to test the RCA out, which only use one chip for L and one for R channel, if it is better, my guess is verified!!!
I did captures of RCA of d30pro a short while back. With the limited captures I got, they are slightly better when nulling against itself (comparing captured 1 vs capture 2) as compared to xlr capture 1 vs xlr capture 2.

However, not necessarily better when nulled against the original file.

Here is one I shared:
https://www.audiosciencereview.com/...clean-sounding-as-the-other.24778/post-861467






.............................................................
Latest summary of comparing original test song:
https://www.audiosciencereview.com/...clean-sounding-as-the-other.24778/post-861227


Comparing another 2 minutes song:
https://www.audiosciencereview.com/...clean-sounding-as-the-other.24778/post-861652
 

b4nt

Addicted to Fun and Learning
Joined
May 29, 2021
Messages
803
Likes
270
On that same picture, I do see two clocks close to that Altera:

Clocks.PNG
 

alek19000

Member
Joined
Jun 27, 2021
Messages
25
Likes
12
I did captures of RCA of d30pro a short while back. With the limited captures I got, they are slightly better when nulling against itself (comparing captured 1 vs capture 2) as compared to xlr capture 1 vs xlr capture 2.

However, not necessarily better when nulled against the original file.

Here is one I shared:
https://www.audiosciencereview.com/...clean-sounding-as-the-other.24778/post-861467






.............................................................
Latest summary of comparing original test song:
https://www.audiosciencereview.com/...clean-sounding-as-the-other.24778/post-861227


Comparing another 2 minutes song:
https://www.audiosciencereview.com/...clean-sounding-as-the-other.24778/post-861652


Thanks !!
and my apology for the RCA output info, I did some searching again ,they say the RCA output is combined the +/- signals after XLR , so that wouldn't rule out the DPLL un-sync case for 4 chips.
from the board there are OP_L+ OP_L- ,which pass to a OPA1612 then OP_L to OUT_L, and OP_L+ to OUT_L+ , OP_L- to OUT_L-.
 

Attachments

  • 0IxZAnTl.jpg
    0IxZAnTl.jpg
    90.8 KB · Views: 77
Top Bottom