Review and Measurements of the Topping D70 DAC

[Veri]

It rose up to 112 dB.

Wow, excellent that does redeem the D70 in my book.

Thanks for the measurements! And topping for lending it out!

 

[Veri]

...and then you have the D2 actual top-end flagship , but unless that thing mops the competition thoroughly, things like the RME and Okto will forever be encroaching on it as they offer a serious price/featureset at the price the D2 will be attempting to compete at.

The SMSL VMV D2 will be a multibit ladder DAC so no, it will not top the competition in specs I'm afraid

 

[Moriarty]

Topping seems to be very proud of very low jitter. They use AK4118 receiver (according to Topping: the jitter lowest S/PDIF receiver) and a pair of very fast AccuSilicon 45.1584MHz and 49.1520MHz "high-end femtosecond clocks". Is it really the most advanced jitter solution these days or just some promo slogan?

How low is D70 jitter actually? Comparing to, for instance, RME ADI-2 DAC with its femtoseconds clock RME is also proud of?

 

[Veri]

How low is D70 jitter actually? Comparing to, for instance, RME ADI-2 DAC with its femtoseconds clock RME is also proud of?

Jitter is pretty boring as soon as it hits 'very good' range honestly. As soon as the worst offender spikes are at or under -120dB, they are already below audible jitter!! But if we want to make it a measurement contest here you go.

Fourth place, RME ADI-2 with its proprietary SteadyClock FS. Very clean, note lack of spikes entirely.

3d place, AK4118 receiver with good clocks as seen in the D70. Spikes visible due to very low noise floor.

(pretty sure this graph's comment was meant to say -120dB)

Second place, OKTO DAC8 with heavy ESS jitter suppressing using ASRC and once again good clocks. Two tiny blips.

First place goes to SMSL VMV D1. It has an Altera cyclone FPGA with sole purpose being reclocking, uses the same AccuSilicon clocks as the D70

-170dB at its lowest point, and very, very clean looking at the replacement board (shown in blue).

But like I started off my post. This is measurebating to the extreme. NONE of all four have any excessive spikes at all! You can't hear this!!

 

[Music1969]

Jitter is pretty boring as soon as it hits 'very good' range honestly. As soon as the worst offender spikes are at or under -120dB, they are already below audible!! But if we want to make it a measurement contest here you go.

Fourth place, RME ADI-2 with its proprietary SteadyClock FS

3d place, AK4118 receiver with good clocks as seen in the D70

(pretty sure this graph's comment was meant to say -120dB)

Second place, OKTO DAC8 with heavy ESS jitter suppressing using ASRC and once again good clocks

First place goes to SMSL VMV D1. It has an Altera cyclone FPGA with sole purpose being reclocking, uses the same AccuSilicon clocks as the D70

-170dB at its lowest point.

But like I started off my post. This is measurebating to the extreme. NONE of all four have any excessive spikes at all! You can't hear this!!

Putting audibility aside and just looking at those measurements, isn't the RME the best out of those 4?

 

[shanecoughlan]

This is measurebating to the extreme. NONE of all four have any excessive spikes at all! You can't hear this!!

This may seem glib, but for the most part jitter seems to be a problem in search of a solution with gear hitting the baseline in 2019. Everything apart from horrific is fine.

 

[Veri]

Putting audibility aside and just looking at those measurements, isn't the RME the best out of those 4?

I would say no. The D70 has some spikes because it is shifted downwards -10dB in noise floor. In the RME those spikes are simply buried. Not saying the RME is not exemplary though, because it is.

But pitting the RME versus the DAC8 and the D1, I don't see how you'd rate the RME over them? Or am I missing something

 

[PierreV]

That is not my understanding. See for example: https://cemarking.net/ce-self-certification/

View attachment 27001

View attachment 27002

It's a bit more complicated than that in fact. The EU summary is around 150 pages... but yes, the system can be gamed by suppliers outside of the EU if they really want to. Assuming I am a producer in Veryfaristan and want to sell power amplifiers in the EU for the first time, if I don't work with an approved verification provider, I should send along full documentation demonstrating that I comply with all the harmonized rules. It then falls on the customs relevant services to verify that conformity and either accept or reject the import. Once the product has been approved, it goes into a database and is in theory approved in all EU (and Turkey I believe - but zero experience on that market).

Now, the verification depends a bit on the country doing it, some vendors know that very well and will try to game the system and also does depend on how high risk the product is considered. Customs officials will be more likely to worry about a heating blanket than a cheap DAC for example. If the manufacturer has long terms views, it is better for him not to be rejected as track of those rejections are also kept.

Then, there are some games at the national level. In theory, countries are supposed to allow the importation and sale of stuff that has been validated elsewhere but they can decide to put additional restrictions on their use or require further certification.

A product that has been validated by a third party should have a number along with its CE certification somewhere in the documentation, and that number can mean a few different things such as design validation only, design and production validation only, etc...

You would expect German customs to take their role more seriously than others and on the whole they probably do, but they still managed to certify and export a bunch of well-known diesel engines Ultimately, the game always depends on the honesty of the players...

 

[Snafu]

To me PSU noise is something i don't like at all even it is below -120. When time goes by PSU usually is first to cause problems.

That's why i wonder why D70 doesn't have external PSU like D50, didn't Topping release at least proto type P50 or something ?
With external PSU D50 owners could have used their PSU for it and price would have been a bit lower.

https://www.google.com/imgres?imgur...MXiAhUqlosKHbGbDEoQMwhAKAAwAA&iact=mrc&uact=8

 

[Music1969]

But pitting the RME versus the DAC8 and the D1, I don't see how you'd rate the RME over them? Or am I missing something

Noted regarding lower noise floor. Perhaps I’d put RME #2 behind DAC8 but those two seem to have less low frequency noise (power supply noise?) than the other two?

Just speaking objectively, not audibly.

 

[NielsMayer]

When I ran a 1 kHz spectrum test, there was a spike at 45 kHz. Since the bandwidth of that test is twice as wide (90 kHz), that spike got included in the computation of THD+N in that graph. Since the spike is not audible and its level would go down with the test signal, it is not a very significant impairment. In other words, that graph looks worse than it should.

Since the "HD Audio" spec/label on this class device extends to 40Khz, I believe your tests should standardize to the frequency range of the "HD Audio spec" ( https://www.jas-audio.or.jp/english/hi-res-logo-en ):

• (1) Microphone response performance: 40 kHz or above during recording
• (2) Amplification performance: 40 kHz or above
• (3) Speaker and headphone performance: 40 kHz or above

By those standards the troublesome 45khz spike just squeaks by the limits.

However, for me , it pulled my finger off the ordering trigger and I'll just wait to upgrade my Dx7S. In my case, the DAC is acting as preamp and directly driving a pair of Hypex NC400 monoblocks that I expressly constructed and cabled to be impervious to interference with signals outside the audio band. These are wide-band amplifiers: https://www.audiosciencereview.com/forum/index.php?attachments/1546109265760-png.19593/

IMHO, the reason some people have complained about the sound of the Hypex NC400, especially in the high-end, is because the NC400 is very susceptible to high frequency interference (per reports), including the dangers of aliasing with RF signals near the switching frequency: https://www.diyaudio.com/forums/class-d/321632-hypex-ncore-nc400-input-anti-alias-filter.html

Many of the builds I've seen using the NC400 pay absolutely no attention to "RF engineering" -- including the unit you tested here on ASR. I'd love to have you test a properly done NC400, e.g. with absolute minimal length in untwisted/unpaired output and input cabling -- which includes inserting the wires into the NC400 output lugs with the wires staying pressed together right up to the screws (i.e. twisted output pair stays twisted right to the connection point, with the leads pointing outwards towards the surrounding NC400 screw lugs) rather than doing it the "easy" way.

Getting back to using the D70 --> Hypex NC400 -- the issue with such wideband by design (it's a feature, not a bug) is that the 45 khz spike will be amplified and sent to the speakers. Although there won't be aliasing concerns, given that the whole point of HD audio is to properly care for our "beyond nyquist" psychoacoustic concerns, we certainly want things to be clean up to at least 48Khz or so....

On the NC400, the high end might get pretty screechy, or there might even be "siren sounds" (as reported) when a local AM radio station transmits or the microwave turns on -- but that's more a job of proper cabling and Pin1 design -- something that most NC 400 builds, including the one tested on ASR, have not paid attention to:
https://www.rane.com/note165.html (Pin 1 Revisited, aka use https://www.neutrik.com/en/neutrik/products/xlr-connectors/xlr-cable-connectors/emc-series >> suggested ASR TODO: investigate the in-case "pin 1" length on devices with potentially noisy interiors like DACs and DAPs -- e.g. a long trace on a circuit board is an antenna and/or large value resistor at RF frequencies, not a ground ).
https://www.rane.com/note151.html ] (Grounding and Shielding Audio Devices)

In either case thanks for reporting on the topping D70 with a level of honesty that allays fears you're a topping shill .

-- Niels.

PS: I'd like to see some measurement on the effects of the volume control on the D70 w/r/t reducing bit-depth of the signal. If you read the ESS marketing hype, they may convince you that -- certainly for 16 bit material, and up to a point for 24 bit sources -- the internal 32 bit implementation of the ESS DAC allows all but the greatest levels of attenuation to be done "losslessly". Basically they're using the "excess bits" of their internal 32 bit implementation to act the same as a virtual resistor ladder network attenuation with attenuation of 32-16 or 32-24 bits possible (dep on source bit width).

On the D70, what is the equivalent of the ESS electronic volume control? ( as described here ):
http://www.audiodesignguide.com/DAC32/digital-vs-analog-volume-control.pdf
(ESS: Digital vs. Analog Volume Controls -- Note: "Internally 32 bits, externally 16/24" pp 14.)

PPS: "The -135dB of the ESS Sabre DAC would need an exceptionally low noise analog volume control to beat its internal digital one" (pp.17)

 

[MZKM]

Nice. Do any measurements get negatively affected with -3dB volume? Can't imagine so, but maybe interesting to look at before you return the unit.

Well, output voltage. Depending on your power amp’s input sensitivity, you would lose at most 3dB if the DAC volume is -3dB, aka 1/2 the wattage.

Luckily, most amps have an input sensitivity of ~ 1.2-1.7 Vrms.

 

[NielsMayer]

By the way, I got an interesting reply from Shenzenaudio regarding the D70. In other places a D70 with AptX-HD bluetooth is listed: " "Support ATPX, APTX-II, APTX-HD, LHDC Bluetooth audio transmission." however this unit is not available yet. (In response to my query, the other store with the incorrect info removed the bluetooth info).

Apparently, Topping will look at the current sales reports on the D70 and decide on producing a bluetooth version in the future....

Sounds like such a "refresh" (D70 PRO?) could address both the full-output linearity issue (showstopper if you plan to use it as pure DAC into preamp), the 50-60hz/100-120hz/etc power supply hum (maybe the SMSL D1 design isn't that crazy afterall), and the 45 khz spike. BTW, the pictures i've seen of the internals of the D70 show no shielding between the analog supply/transformer section and the rest of the unit. Even my old and dirt-cheap Emotiva XDA-2mk2 has a separate chassis section for the toroid the switching supply and display sections https://darko.audio/wp-content/uploads/2012/11/emotiva_xda-2_internal1-580x484.png

Also, Topping, if you're reading this, please add a 12v output trigger -- which would be great to have alongside existing auto on/off feature auto-powering other gear, or driving a relay ( e.g. https://dlidirect.com/products/iot-power-relay ).

Also, is the power supply hum "modulating" the jitter on this or any other DAC ASR tests? That could be far less "euphonic" than the normally noise-based jitter. https://www.stereophile.com/reference/1290jitter/ https://d3uzseaevmutz1.cloudfront.n...e_of_spdif_digital_interface_transceivers.pdf

 

[777]

What is the ratio between 1khz test signal and 45khz ?

 

[Feyire]

Getting back to using the D70 --> Hypex NC400 -- the issue with such wideband by design (it's a feature, not a bug) is that the 45 khz spike will be amplified and sent to the speakers. Although there won't be aliasing concerns, given that the whole point of HD audio is to properly care for our "beyond nyquist" psychoacoustic concerns, we certainly want things to be clean up to at least 48Khz or so....

Would the standard digital filters in a DAC, which are typically applied from 20khz, make this a non-issue?

 

[Ron Texas]

The only thing which bothers me is at $500, the D70 is getting a bit expensive for Chi-Fi.

 

[Dominique-T]

Would be nice to compare This one with the SMSL M300, same chipset (AKM AK4497) and symmetrical outputs...(Cheaper).

 

[amirm]

Would be nice to compare This one with the SMSL M300, same chipset (AKM AK4497) and symmetrical outputs...(Cheaper).

I have not succeeded in getting an M300....

 

[777]

Would be nice to compare This one with the SMSL M300, same chipset (AKM AK4497) and symmetrical outputs...(Cheaper).

Gustard A20H has the same AK4497 and we see again the same IMD curves. Topping DX3PRO has AK4493 and again the same IMD curves. RME ADI-2 has AK4490 but doesn't has the same IMD curve. So, there is any clue for all that DAC's ?

There is, of course. The voltage for AK4493 and AK4497 is maximum 5,5v. For AK4490 the same voltage could be 7,2v ! So, what we see at 0dBfs it-s a first sign of the AKM chipset clipping ?

 

[Jefrpol]

A bit off-topic, but some of the questions above got me thinking:

I'll confess that when I read these reviews I go by the test explaining the graphs and not the graphs themselves. So perhaps there are variations that should be readily apparent but aren't to me.

That said, I'm curious: what does a more expensive dac get you? If the measured output, as depicted in the graphs, is similar or better, are you paying more for nicer casework and better customer support? Does the analog implementation change the sound you hear (which you would think is measurable but perhaps not) and is that a factor in the price?

Also, I notice most comments are about how well this dac or another dac would work with some headphone amplifier. Is this dac, for instance, better mated to a headphone amplifier than to a two channel speaker system? In other words, do the measurements not reveal some information useful in one application rather than the other?