Topping D900 - teardown

[Billy Budapest]

I'v been using D900 for 3 weeks, the best sounding DAC in the history of Topping, no need to say more.

Considering there is no way for properly constructed DACs to sound different than each other, there was no need even for you to say that.

 

[Billy Budapest]

In case the sarcasm from the other members went over your head, the reason is all DACs sound the same.

How can DAC's have a SOUND SIGNATURE if they measure as transparent?

I know there are people out there that think cables affect sound, which is much worse, but there really is no response to something like that, but just to smile and nod. But what about people who talk about DACS as if they were headphone drivers or speakers, and talk about the SOUNDSTAGE...

www.audiosciencereview.com

Some very poorly performing DACs could conceivably sound different, but they would have to be really misbehaving in order not to be audibly transparent.

 

[ternam]

What are these 2 chips used for? It seems the left one is for AES3 galvanic isolation? Thanks!

 

[solderdude]

It seems the left one is for AES3 galvanic isolation?

yes, for SPDIF signal galvanic isolation.

Scary

View attachment 461227

It is actually the best way to do this correctly.
The timing, certainly at high speed, of the (I2S) signals is important.
This means the wire length for all signals needs to be the same and is easier to do with cables than with PCB traces without special trace routing on the PCB.
Secondly the impedance of the cable is constant compared to that one would get on a PCB when using board-to-board connectors.
Looks like a good decision from Topping.

 

[ternam]

yes, for SPDIF signal galvanic isolation.

It's tricky that when saying galvanic isolation, it should indicate ground is also isolated from peripheral devices, like computer, sound card, etc.

Per my understanding, the SPDIF ground (outer ring) and the AES3 ground (pin 1) are all connected and are connecting to AC PE. That means, the ground loop can't be eliminated from SPDIF and AES3 inputs, even there is an input transformer inside D900 as indicated in the picture.

Please advise if my understanding is correct. Thanks.

 

[solderdude]

It's tricky that when saying galvanic isolation, it should indicate ground is also isolated from peripheral devices, like computer, sound card, etc.

Per my understanding, the SPDIF ground (outer ring) and the AES3 ground (pin 1) are all connected and are connecting to AC PE. That means, the ground loop can't be eliminated from SPDIF and AES3 inputs, even there is an input transformer inside D900 as indicated in the picture.

Please advise if my understanding is correct. Thanks.

coax 1 and coax 2 are clearly isolated and its signals will go through the pulse transformer.

 

[Billy Budapest]

Why these sound horrible mute chips on the output instead of decent relays? It's not worth the money.

Unless the chips let some residual signal come through, or are susceptible to RF interference, I highly doubt there will be any audible difference upon muting than if relays would have been used. Plus, we are talking about muting here. I don’t really care about sound quality when the unit is muted. Lol.

 

[Billy Budapest]

The USB, AES and I2S inputs of my D90 III Discrete do not sound the same.
It is fairly obvious with quick switching, and yes, I have verified with a multi-meter that the gains are the same.

Internally, the D90 is not handling the inputs that same. Roon should be the same, for all inputs to the Raspberry Pi5/P2AES DietPi is also involved.

There is no testing here for adverse conditions and different sources.

There is bias in those believe that all DACs sound different that believe they sound the same.
There are also rewards for supporting popular positions.

- Rich

Confirmation bias is strong, as is the placebo effect. That’s not a dis on what you are hearing, but rather is a scientific fact.

Whatever can be heard can always be measured. Whatever can be measured can’t always be heard.

 

[Billy Budapest]

Certainty is nice, but the testing at ASR is insufficient to make claims that in use, all DACs sound the same.
There are too many variables and untested modes.

ASR a set of tests, consistently applied that are useful to verify basic design hygiene.

- Rich

The issue isn’t whether there are differences. Rather, the issue is that all such differences, even though measurable, are not high enough in level to breach the limit of human hearing. They are just too low in level, and any modern competently designed DAC should not cross that threshold in any parameter—whether it has been tested for that or not.

I’m not convinced there are audible differences between 16-bit files on up. Why? I don’t think there is recorded music that takes advantage of the dynamic range able to be expressed in 16-bit recording. 96dB swings in output level on something that has actually been recorded? Extremely doubtful. As far as sampling rates, I’m not convinced any differences over 44.1kHz are audible, although they certainly are measurable.

 

[MBO]

I received the D900 today, and I can say with confidence that it has not excited me as much as the D90III Discrete. The D900 certainly sounds better in many ways, but the overall sound setup is similar to an EQ being on, despite the fact that PEQ is off. The D90III sounds more neutral, clear and distinct. The D900 sounds soft, too soft, like through cotton wool. And with PEQ on, the sound can be improved a little, but it still lacks neutrality. I suppose there is some confusion with PEQ... And there is another flaw, when you turn it off, the D900 loses all settings.
The D900 needs a major SW update!

 

[Billy Budapest]

I received the D900 today, and I can say with confidence that it has not excited me as much as the D90III Discrete. The D900 certainly sounds better in many ways, but the overall sound setup is similar to an EQ being on, despite the fact that PEQ is off. The D90III sounds more neutral, clear and distinct. The D900 sounds soft, too soft, like through cotton wool. And with PEQ on, the sound can be improved a little, but it still lacks neutrality. I suppose there is some confusion with PEQ... And there is another flaw, when you turn it off, the D900 loses all settings.
The D900 needs a major SW update!

If you are hearing any differences between the D900 and D90, then maybe PEQ *is* activated but is not showing up as so in the display. In that case, a software update definitely would be needed.

 

[archerious]

Hello,

Some of my friends are quite interesting on Topping D900 since Topping shows on CanJam New York 2024. They start to sell D900 since last month but it's weird that Topping pending D900 for a year cause the date code on bar code label shows 202405.

Front panel, the ID is quite simple with touch panel which used same as another DAC like Centaurus.

Back panel, D900 integrated preamp inside so there are two set of XLR on it without RCA.
I think RCA output can be added if only maintain one set of Toslink/Coaxial.
Rest input I2S, AES, Toslink*2, Coaxial*2, 12V trigger, USB type B and one more Type C.

Bottom plate and foot, D900 used aluminum foot pads this time. You may see counterbore holes reserved for rubber foot, fortunately they kept aluminum foot pads.

The remote made by aluminum like DX9, rest are freebies for first batch.

Inside, there are five portions, the left is power supply, middle-left is digital process and input, middle-right is DSM32 module and right are preamp.

PCB bottom side, this is the first time I saw flex clean on PCB, hope Topping may keep improve their build quality of SMT.

Power supply is Meanwell IRM20-15 isolated power, and there are six set of TPS5420 step down for 12V, 4.3V, ±12V and ±8V.

Digital input, 12V trigger on top, BT module on top center. XMOS XU316 on the middle-left and CPLD in the center. Top right is PEQ IC and middle-right is SRC bridge IC with SPDIF receiver function.

The crystal of XMOS XU316 and CPLD used NDK SDA series, which is ultra low phase noise.
After check PCB trace and found those signals and MCLK came from CPLD which means that CPLD re-clock function works for all input.

The top LQF48 IC is for PEQ but didn't know which IC is. The SRC bridge looks like CT7302 which can be used for DSD converter and SPDIF receiver.

MCU came from Synwit, and ultra low noise LDO SGM2209/SGM2211 on the top.
There is an AK4377 blue board which used for display only.

To avoid any impact of sound quality, it used on DAC to decode both PCM and DSD then used MCU ADC to read the output level which is digital date to show on touch screen, like UV display.

DSD signal connect between DSM32 board and main board with IPX coaxial cable, rest power and control signal used pin header. There are 2 set 32-phase bit shift on DSM32 so it will be 64pc of logic gate on it.

Vref made by OPA, although the top marking is removed but I guess this one is OPA1612.
Rest logic gates also remove top marking, suspect Topping is trying to prevent copycat from competitors.
The resistors on string DAC used normal 0402, although there are some 0402 resistors with low temp drift but I can't identify those difference by visual inspection

The clock on DSM32 board used 11.2898 MHz, but I can recognize which brand of it.

You can see I/V board as a big discrete OP, and of cause all ICs polished. I guess the OPA on I/V may be OPA1652 or OPA1656.

The volume control used 9pcs relays per channel, which provide 256steps from -99 to +8dB. You can change 0.5dB or 1dB per step on setup manual.

There are two OPA on preamp, I guess the tiny one is OPA1612 as power supply and bigger one is OPA1656 as gain stage, the whole design may come from A70pro.

The power of resistors matrix also came from OPA1612.

The AC coupling caps used Nichicon MUSE BP 100uF 25V on direct output of DSM32 board.

The temperature of chassis is around 38deg.C (room temp. 33deg.C) after run-in 8hrs.

Take a group photo for both D90MQA and D900 in the end.

Let me know any you have any question on this report, thank you.

Wait a minute here, I was about to ask who the target audience was then I saw the anime waifu stickers.

Is Topping putting the waifu theme on the box?

Is it a Zeos edition? /s

If I buy a D900 do I get the Waifu stickers?!

My weeb wallet never runs out compared to HiFi wallet. Moondrop has made a lot off of me. Is Topping embracing culture?!

 

[whydan]

Wait a minute here, I was about to ask who the target audience was then I saw the anime waifu stickers.

Is Topping putting the waifu theme on the box?

Is it a Zeos edition? /s

If I buy a D900 do I get the Waifu stickers?!

My weeb wallet never runs out compared to HiFi wallet. Moondrop has made a lot off of me. Is Topping embracing culture?!

I don't know, you should ask your local dealer or just ask Topping service team.

 

[archerious]

I don't know, you should ask your local dealer or just ask Topping service team.

I reached out to them.

 

[ternam]

Topping just copy their designs over and over. D900 is an old product (by old I mean design not release) and at that time Topping used it on all of their devices for signal routing. Topping only realized how to route spdif to xmos a few years later (after D50 III) so in recent designs they abandoned it. But they don't want to change the existing, finished, designs, especially on this premium device, as there's no need to save BOM.

BTW, the chip has two round orientation markings. Very few chip has those. So if you see the two orientation markings, it's definitely LC89058W.

Moreover, if you really have read the datasheet as you said, you will understand LC89058W is the chip they use in D900, based on your teardown photos.
Go to page 63 of the datasheet. you'll find resistors/capacitors/data lines match *exactly* to the reference design.

I just asked Topping technical engineer, the chip used to do PEQ in D900 is XU316.

 

[audiofun]

I just asked Topping technical engineer, the chip used to do PEQ in D900 is XU316.

I don't believe you understand the topic well. the quoted post has nothing to do with PEQ.
LC89058W is only used to route audio signal. no processing is done in the chip.

 

[NekoLaska]

I don't believe you understand the topic well. the quoted post has nothing to do with PEQ.
LC89058W is only used to route audio signal. no processing is done in the chip.

He replied to the wrong person, he meant that the article said that PEQ used a standalone IC, but in fact, the D900 used XU316 to implement PEQ

 

[archerious]

He replied to the wrong person, he meant that the article said that PEQ used a standalone IC, but in fact, the D900 used XU316 to implement PEQ

Love your profile pic.

 

[Deleted member 89542]

I understand from ASR that all well measuring dacs and amps are the same and no one can tell the difference in blind tells. So why do certain manufacturers keep coming out with different dacs and amps. Is is just snake oil marketing or are there genuine differences in the sound between well measuring dacs and amps. Just trying to understand.

 

[archerious]

I understand from ASR that all well measuring dacs and amps are the same and no one can tell the difference in blind tells. So why do certain manufacturers keep coming out with different dacs and amps. Is is just snake oil marketing or are there genuine differences in the sound between well measuring dacs and amps. Just trying to understand.

I’d argue some newer products are adding useful or improving features like remote control for volume, improved build (perhaps metal from plastic), digital screen, PEQ, higher wattage output, streamer support etc.

The DX5 II comes to mind as its cheaper than RME ADI-2 DAC FS but is quite similar.

I don’t understand the appeal of A900 and D900 though. If they had a waifu on the box or stickers or something I might’ve considered buying it just for that but otherwise I see no point.