• 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!

Topping D90 III - teardown

whydan

Member
Joined
Apr 17, 2021
Messages
23
Likes
72
Hello,

I knew Topping DAC started from the 1st generation of D90 with AK4499EX, and I still keep D90 MQA.
Then 2nd gen D90SE/D90LE change to ES9038pro because fire broke out AKM factory, which I owned for months.
Now the 3rd gen. D90 III still maintain the same famliy ID design with new ES9030pro.
I borrow this D90 III from my friend and considering to get a new one.

Front view, although D90 keep the same ID design on chassis but it still changed from SW button to Knob.
Honestly, I prefer SW button more than knob due to mobility, I know that is a weird reason for a desktop DAC.
7kS2gWV.jpeg


Rear view, XLR/RCA output, I2S/Coaxial/OPT input, BT, USB and AES input, now it has 12V trigger
Topping add 12V trigger design in their production in recently years, it's really useful for end customer.
BysvhpF.jpeg


Bottom side, D90 III use new foot base which is much better than previously rubber one.
GPFfAj5.jpeg


I found that IR sensor is far away from panel during teardown, which means D90 III still has remote problem as D90.
But D90 III add new BT remote control, that maybe a good alternative way to resolve its poor IR design.
UgYgU1J.jpeg


Inside, there are three portions. Power on left side, digital procces in middle and DA convertor on right side.
MDBQxOm.jpeg


It used Meanwell IRM-20-15 15V 20W SMPS and onboard EMI filter made by discrete components
PSu5VVN.jpeg


The source of digital power is TI 5430, which output 5.7V/4.2V as main power rail then convertor to others.
oo28Fgb.jpeg


The source of analog power is used LM317 to reg to 12.4V then convert to pos/neg power of OPA.
sd7o2QX.jpeg


Analog power used TI 5430 to set two power rail 6.2V as ±6.2V from LM317, then go CLC filter to reduce noise.
V0lfoDL.jpeg


SPDIF receiver is AK4118, the MCU ST32 on right side is used for display and overall control.
hHgACQL.jpeg


The USB receiver used XMOS XU316 on the center, and Altera CPLD on below with NDK NZ2520SDA 45M/49M
BT module on the left upper side, the optocoupler of 12V trigger is near connector.
It511fh.jpeg


D90 III used two ES9039SPRO which is more than D90SE with one ES9038 and AVDD5V provided from OPA1612
DAC 1.2V came from the D2D IC on the left lower side. TAITIEN 22M/25M in the center can be switched to DAC.
It's quite different from previously ESS chip since 9018. 9018 request one reference clock and use internal PLL.
I believe that may reduce power consumption by remove internal PLL and cost. ES9038 is quite hot when running.
bZR1jN7.jpeg


I/V used brand new design I/V converter, you may image those 4 modules as discrete OPA from Topping.
The balanced to single end used OPA1612 as usual.
34yZC6S.jpeg


Coupling capacitor used Nichicon FW 220uF 16V, which is not Hi-Fi enough for me but still ok on circuit design.
sMibMM4.jpeg


I skip all test since Amir already provide a specific review on forum.

Let me know if you got any quesiton on this report, thank you.
 
Last edited:
Hello,

I'm working on a multichannel USB to I2S bridged based on stm32 and a simple multi-channel DAS (exES9080). Browsing the forums, I see that many pay some attention to the noise propagation between the USB (especially in case of USW powering the device), and the analog part: multiple power supplies and/or LDO, ferrite beds, digital isolators... I try to get an understanding about how that is important or not.

Would it be possible to derive from the tear down some on of the Topping DAC features implemented to mitigate this noise propagation from the USB? Obvioulsy they don't use the USB VBUS as power supply.

Are there some filtering features at USB connector level (ferrite beds, ...)?

Do they have isolators on USB or data lines?

Some hints about the grounding scheme (single ground plane for all boards, different ground planes... that one may not be easy) ?

Other features?

Best regards,

JMF
 
Hello,

I'm working on a multichannel USB to I2S bridged based on stm32 and a simple multi-channel DAS (exES9080). Browsing the forums, I see that many pay some attention to the noise propagation between the USB (especially in case of USW powering the device), and the analog part: multiple power supplies and/or LDO, ferrite beds, digital isolators... I try to get an understanding about how that is important or not.

Would it be possible to derive from the tear down some on of the Topping DAC features implemented to mitigate this noise propagation from the USB? Obvioulsy they don't use the USB VBUS as power supply.

Are there some filtering features at USB connector level (ferrite beds, ...)?

Do they have isolators on USB or data lines?

Some hints about the grounding scheme (single ground plane for all boards, different ground planes... that one may not be easy) ?

Other features?

Best regards,

JMF
hello JMF,

I saw your project on other forum, let me try to explain your question as I can.
1. as you know, Topping DAC won't use USB VBUS as power supply but they still use VBUS as a trigger. I remember that is required from XMOS chip.
2. they use GND breaker (series resistor between connector ground to DGND on DAC), marked as red in attached pic
1722416194133.png


3. no, only common mode choke on USB data lines. I got some improvement when using Topping USB isolator HS-02 so add adum3166 will be great.
4. for EGND I will say GND breaker. for AGND/DGND probably bead or 0ohm resistor will be fine.
5. if you plan to add USB isolator, probably use high class femto clock on chip side wll be great.
 
hello JMF,

I saw your project on other forum, let me try to explain your question as I can.
1. as you know, Topping DAC won't use USB VBUS as power supply but they still use VBUS as a trigger. I remember that is required from XMOS chip.
2. they use GND breaker (series resistor between connector ground to DGND on DAC), marked as red in attached pic


3. no, only common mode choke on USB data lines. I got some improvement when using Topping USB isolator HS-02 so add adum3166 will be great.
4. for EGND I will say GND breaker. for AGND/DGND probably bead or 0ohm resistor will be fine.
5. if you plan to add USB isolator, probably use high class femto clock on chip side wll be great.

Thanks so much whydan !

This is really helpfull. It demonstrates that "not" using exotic features like isolators does not prevent to reach more than 123 SINAD and achieve top 20:
- decouple USB ground,
- only use Vbus as trigger input, but not power,
- I imagine sound GND management,
- excellent engineering for the rest.

You make my day.

By the way, in your message, you state "for EGND I will say GND breaker". I look at ES9039 datasheet and don't see any. Only DGND and AGND. could you please help on that point?

Best regards,

JMF
 
Last edited:
2. they use GND breaker (series resistor between connector ground to DGND on DAC), marked as red in attached pic
View attachment 383939
Hey @whydan the picture is not clear enough for me to see but what you marked in red there seems to be the connection of the usb shield to ground, as it is commonly made with a resistor+capacitor. Is it possible that usb ground is connected to DAC ground by the ferrite bead (? just guessing) besides the capacitor you encircled? Just curious.
 
D90 III used two ES9039SPRO which is more than D90SE with one ES9038 and AVDD5V provided from OPA1612
DAC 1.2V came from the D2D IC on the left lower side. TAITIEN 22M/25M in the center can be switched to DAC.
Thank you for this teardown.

I suppose the newest topping d90 III works in sync mode. Therefore there are two clock near the DAC chips. According to the es9039pro pdf, If I read it correctly, in sync mode with 22/24M clock only 352/384kHz can be played but topping is able to play 705/768kHz.
How it'is possible? I misinterpret/misunderstand something?


1000001231.jpg
 
Thanks so much whydan !

This is really helpfull. It demonstrates that "not" using exotic features like isolators does not prevent to reach more than 123 SINAD and achieve top 20:
- decouple USB ground,
- only use Vbus as trigger input, but not power,
- I imagine sound GND management,
- excellent engineering for the rest.

You make my day.

By the way, in your message, you state "for EGND I will say GND breaker". I look at ES9039 datasheet and don't see any. Only DGND and AGND. could you please help on that point?

Best regards,

JMF
hi JMF,

EGND means earth gnd, let me quote a circuit from diyaudio for better explanation.
Earth gnd used to connect earth / chassis/ USB shielding, which may have higher EMC noise from outside.
1722476644664.png
 
Hey @whydan the picture is not clear enough for me to see but what you marked in red there seems to be the connection of the usb shield to ground, as it is commonly made with a resistor+capacitor. Is it possible that usb ground is connected to DAC ground by the ferrite bead (? just guessing) besides the capacitor you encircled? Just curious.
Hello MCH,
yep you're right, that is shielding gnd and I guess you are asking for DGND from USB cable.

Actually I don't own this D90III, I borrowed from a friend and returned couple weeks ago.
I will check my D90 MQA later and update once I got it, or below D30pro pic as reference.
E5WXJvV.jpg
 
  • Like
Reactions: MCH
Thank you for this teardown.

I suppose the newest topping d90 III works in sync mode. Therefore there are two clock near the DAC chips. According to the es9039pro pdf, If I read it correctly, in sync mode with 22/24M clock only 352/384kHz can be played but topping is able to play 705/768kHz.
How it'is possible? I misinterpret/misunderstand something?


View attachment 384020
Hi Piaseczek,
Glad you notice that, that confuse me for a while when writing this tear down report.
In beginning I thought those two clock near DAC (22/24M) provide BCK and 48/49M near CPLD provide MCLK.
But I trace back the pin connection between 22/24M and ES9039 on MCLK pin, so I didn't have answer for that.
1722481346398.png
 
hi JMF,

EGND means earth gnd, let me quote a circuit from diyaudio for better explanation.
Earth gnd used to connect earth / chassis/ USB shielding, which may have higher EMC noise from outside.
View attachment 384050
Thanks a lot for the explanation. I will dig the DIYaudio thread you refer to.
 
Hi Piaseczek,
Glad you notice that, that confuse me for a while when writing this tear down report.
In beginning I thought those two clock near DAC (22/24M) provide BCK and 48/49M near CPLD provide MCLK.
But I trace back the pin connection between 22/24M and ES9039 on MCLK pin, so I didn't have answer for that.
View attachment 384054

Just wild guess - maybe 705/768kHz signals are downsampled in XMOS?
 
I am curious if the color of the pcb has an influence on the sound quality?
Try looking for "Kandinsky" in this context, it can certainly be applied to the colors of the circuit boards.
It would also explain why I always found yellow circuit boards to sound a bit harsh and sharp.
 
hi , i have d90 from ak4499 now, is it the worth upgrade to d90 iii sabre? or wait the next generation?

but when the next generation launch ?
 
hi , i have d90 from ak4499 now, is it the worth upgrade to d90 iii sabre? or wait the next generation?

but when the next generation launch ?
hello Fendy,

I didn't have direct comparison between D90 MQA and D90III since I did a fully MOD for D90 MQA.
As I recall, D90 III sounds more balance in high and bass whch got more improvement than original D90 MQA.

But the next gen of D90 III Discrete with DSM may come out on Q3,24 in rumor, I will suggest to wait that one.
Untill some reviewer provides a fully comparision between D90 III Sabre and D90 III Discrete, then pick one you preferred,
 
Back
Top Bottom