Topping D90 Balanced USB DAC Review

[JohnYang1997]

Not the gain generate the noise.

Opamp's input stage noise and the DAC's inherent get amplified.

 

[barrows]

The lower specs are caused by lower output current. Lower output current = higher gain in the analogue output stage = more noise.

i do not understand what you are talking about? I am referring to the DAC chip data sheet (page 17), this has absolutely nothing to do with the I/V, these are specs for distortion and noise for the chip, at different rates of DSD input?

 

[JohnYang1997]

i do not understand what you are talking about? I am referring to the DAC chip data sheet (page 17), this has absolutely nothing to do with the I/V, these are specs for distortion and noise for the chip, at different rates of DSD input?

So how does the company give out specs without testings?

 

[KingFiercer]

Bummer, there must be something wrong with your unit then. Hopefully they can do something about it or exchange it.
Normally I let Audirvana upsample 44.1 to a higher quantity. But to be sure I tried 44.1 without upsampling with my D90 and I don't have that problem over USB.
I hope they get you a new one .
I hope you'll keep us informed, because I am anxious to know how well their customer service is.

It's probably the XO for the 44.1khz/88.2khz was misplaced with the one for 48khz/96khz.

There is answer from Topping: "The result of the experiment by the replacing the wrong clock is that: if the clock is wrong, not only the USB will fail, but other inputs will also fail. However, caution is still recommended to return".
After this response, the seller agreed to a full refund, so the situation was well resolved!
I decided to stay with D70, because with it the sound of my system is nicer/mellower for my ears. D90 has superb perfromance though.

 

[barrows]

So how does the company give out specs without testings?

I must say, I cannot understand what you are posting here? All I am I pointing out is that the performance of the chip degrades with DSD 512 inoput, vs DSD 256, so I advise those folks, like me, who like to oversample in software to high rate DSD, to use dSD 256 as their input rate, rather than 512 (at which rate the chip's specified performance degrades).
I so not see what this has to do with current output of the chip, the degradation is only for the DSD input rate of 512.

 

[MRphotography]

There is answer from Topping: "The result of the experiment by the replacing the wrong clock is that: if the clock is wrong, not only the USB will fail, but other inputs will also fail. However, caution is still recommended to return".
After this response, the seller agreed to a full refund, so the situation was well resolved!
I decided to stay with D70, because with it the sound of my system is nicer/mellower for my ears. D90 has superb perfromance though.

When you say the D70 has a more mellow sound, can you expand on this a little? I actually prefer a more laid back sound but had read the D90 was actually just as smooth as previous AKM implementations with stronger lower frequency or bass performance and a more airy sound? Can you confirm?

 

[barrows]

Can anyone report what the output voltage of the D90 is with DSD input in volume bypass mode? I need at least 4 V output at 0 dBFS, as I oversample in software and this requires a level reduction of about -5 dB to avoid clipping in the oversampling process. Even another couple of dB in the output would be appreciated. I really wish for more like 5 V output with DSD input to have enough gain to suit me.

 

[JohnYang1997]

Can anyone report what the output voltage of the D90 is with DSD input in volume bypass mode? I need at least 4 V output at 0 dBFS, as I oversample in software and this requires a level reduction of about -5 dB to avoid clipping in the oversampling process. Even another couple of dB in the output would be appreciated. I really wish for more like 5 V output with DSD input to have enough gain to suit me.

Output voltage is the same at all modes.

 

[Veri]

Can anyone report what the output voltage of the D90 is with DSD input in volume bypass mode? I need at least 4 V output at 0 dBFS, as I oversample in software and this requires a level reduction of about -5 dB to avoid clipping in the oversampling process. Even another couple of dB in the output would be appreciated. I really wish for more like 5 V output with DSD input to have enough gain to suit me.

Almost all DAC will output 2V rca 4V xlr though. That is the convention.. only few output more (some Aune models, Auralic, ...)

 

[Nicolaas]

Have been using the D90 for several hours while playing 44.1k 16b FLAC files from my Android tablet with BT Aptx and from my Android phone with Aptx HD. And I must admit this sounds very good. My intention was to use the D90 only via wired USB connection from my dedicated laptop but BT Aptx (HD) gives me serious alternative options .

 

[barrows]

Almost all DAC will output 2V rca 4V xlr though. That is the convention.. only few output more (some Aune models, Auralic, ...)

I would not say that. Many DACs have higher output: Benchmark, Mola Mola, EMM Labs, RME, etc, etc. Also, many DACs have 2 and 4 V output with PCM, and then lower output with DSD, often 6 dB lower with DSD, so it is good to check and know for sure before one purchases.

I ask because I like to oversample everything in software to DSD 256. Oversampling in software requires that one apply a level reduction first of between -3 dB and -6 dB in order to avoid clipping in the oversampling process. So the ultimate DAC output will be reduced by that amount. and my amp(s) require 2 V input for full output, and I like to have a bit more than that available for some recordings made at very low average levels, which, here, need a bit more gain to reach satisfying playback levels.

I have had a couple of DACs here to test, which actually did not have enough output level available for my needs, so i like to be sure. It is really frustrating if you have a system set up, when even at full volume, the level is too low for satisfaction! I understand this problem is rare (most users find they might have the reverse problem) but it is real if one has it.

I checked directly with Topping to confirm a couple of things, and the result was good news:

1. The DAC output levels are the same with DSD as they are with PCM: 2 V from the single ended outputs, and 4 V from the balanced outputs

2. The DAC does allow for "Direct DSD Mode" when it is set to "DAC Mode"

So it will have 4 V output and bypass the processing and modulator when in "DAC Mode" and playing DSD. I wish it had a little more gain, but I can live with 4 volts for 95% of the recordings I play.

Here's to hoping that RME might be working on an AKM 4499 DAC, as they will have higher output available.

Here is another thing, this is a bit tweaky, but would probably result in a better system overall. How about a DAC which operates the AKM chip in direct DSD mode, which has 10 V output, and then one can connect it to the new Purifi amp modules, with no input gain stage. The RME ADI-2 has almost enough gain to do this (it would have enough for some systems). With higher voltage running between DAC and amp, the influence of cables on sonics is lower, and noise has less effect on cable transmission as well. This is why pro DACs, and pro line level gear tend to run at higher voltages than consumer gear, to avoid noise pickup on longer cable runs which are used in the pro environment.

 

[Soundstage]

Here is another thing, this is a bit tweaky, but would probably result in a better system overall. How about a DAC which operates the AKM chip in direct DSD mode, which has 10 V output, and then one can connect it to the new Purifi amp modules, with no input gain stage. The RME ADI-2 has almost enough gain to do this (it would have enough for some systems). With higher voltage running between DAC and amp, the influence of cables on sonics is lower, and noise has less effect on cable transmission as well. This is why pro DACs, and pro line level gear tend to run at higher voltages than consumer gear, to avoid noise pickup on longer cable runs which are used in the pro environment.

Do you imply that one should not use standard DACs like the DX7 or the D90 as direct source to the power amp, but always use a preamp in between (even with short cables)?

 

[barrows]

Do you imply that one should not use standard DACs like the DX7 or the D90 as direct source to the power amp, but always use a preamp in between (even with short cables)?

No, not at all. I am big believer in going amp direct as leaving out the extra noise and distortion of a preamp is an advantage. But it is important to make sure the DAC output stage has enough voltage gain, and enough current capability to easily drive the amplifier in question. Thankfully, most contemporary DACs and amplifier combinations are fine running this way, but there are a few exceptions.
Specifically with the D-90, it uses LM 4562 which should be fine driving most amps direct, and I see the D-90 also has very low output impedance, which indicates that the implementation of the LM 4562 is such that it has high current capability. I am almost sold on picking up a D-90 to use here as a commercial reference DAC (I like to build DIY DACs, but having a nice, fine measuring, commercial reference is a good idea for comparisons).

One other interesting thing I find with the build of the D-90, is that they use 4 OPA 1612 in the I/V stage, where most manufacturers would normally use just two (the 1612 is dual OPA). Apparently they seem to be running each OPA with single ended input from the AKM 4499, and then summing their outputs on the way to the LM 4562 output stage (I need to look at the AKM 4499 data sheet to be sure of this). By doing this they probably gain 3 dB of SNR, cool.

Edit, correction: OK, looking at the data sheet I see that the AKM 4499 has two balanced DACs per channel, so there is one OPA 1612 for each of the DACs. So the OPA 1612 is operating with balanced input, and then outputs are summed after the chip. but the other part of this approach still applies, most manufacturers woudl just sum the outputs of the two DACs on each channel, and then use two OPA 1612 for the I/V conversion stage. It is cool thta Topping ups the ante here using 4 OPA 1612 and summing after them, likely gaining 3 dB of DNR in the process.

 

[Soundstage]

No, not at all. I am big believer in going amp direct as leaving out the extra noise and distortion of a preamp is an advantage. But it is important to make sure the DAC output stage has enough voltage gain, and enough current capability to easily drive the amplifier in question. Thankfully, most contemporary DACs and amplifier combinations are fine running this way, but there are a few exceptions.
Specifically with the D-90, it uses LM 4562 which should be fine driving most amps direct, and I see the D-90 also has very low output impedance, which indicates that the implementation of the LM 4562 is such that it has high current capability. I am almost sold on picking up a D-90 to use here as a commercial reference DAC (I like to build DIY DACs, but having a nice, fine measuring, commercial reference is a good idea for comparisons).

One other interesting thing I find with the build of the D-90, is that they use 4 OPA 1612 in the I/V stage, where most manufacturers would normally use just two (the 1612 is dual OPA). Apparently they seem to be running each OPA with single ended input from the AKM 4499, and then summing their outputs on the way to the LM 4562 output stage (I need to look at the AKM 4499 data sheet to be sure of this). By doing this they probably gain 3 dB of SNR, cool.

And how much noise and distortion is added to the signal when one runs the D90 (or any pure DAC) through a headphone amplifier? Isn’t it unfair to compare the SINAD of the D90 for example with the SINAD of the DX7?

 

[barrows]

A headphone amplifier will add as much noise and distortion as it has. I see no reason not to try and run headphones directly from the balanced outputs of the D-90 with a balanced headphone cable though, if 'phones are one's thing. I would think a relatively efficient headphone like the Focal Clear would sound great direct from the balanced outputs of the D-90. Seems like it should drive efficient headphones, but not all, to acceptable levels from the balanced outputs. Just be sure not to use very insensitive headphones that way, and advance the volume slowly, and stop using it if you hear any distortion.
Of course for serious headphone folks a really good dedicated headphone amp is probably in order.

 

[Veri]

A headphone amplifier will add as much noise and distortion as it has. I see no reason not to try and run headphones directly from the balanced outputs of the D-90 with a balanced headphone cable though, if 'phones are one's thing. I would think a relatively efficient headphone like the Focal Clear would sound great direct from the balanced outputs of the D-90. Seems like it should drive efficient headphones, but not all, to acceptable levels from the balanced outputs. Just be sure not to use very insensitive headphones that way, and advance the volume slowly, and stop using it if you hear any distortion.
Of course for serious headphone folks a really good dedicated headphone amp is probably in order.

I tried this before and it's true, additional amplification can at best keep signal in tact, but likely adds its own polution.

On the other hand a lot of DACs are not made for listening directly to its output, but rather as a line level with output impedance 50ohms or higher.. That could only sound 'good' with insensitive headphones, indeed...

 

[Soundstage]

How do I know if I really need a headphone amp? I am only at 95% of the volume of my DAC.

 

[AndyLu]

How do I know if I really need a headphone amp? I am only at 95% of the volume of my DAC.

The only way to know is listening to one. Borrow one from a friend. Nobody here has your ears, your preference and your taste

 

[tonapo]

I would not say that. Many DACs have higher output: Benchmark, Mola Mola, EMM Labs, RME, etc, etc. Also, many DACs have 2 and 4 V output with PCM, and then lower output with DSD, often 6 dB lower with DSD, so it is good to check and know for sure before one purchases.

I ask because I like to oversample everything in software to DSD 256. Oversampling in software requires that one apply a level reduction first of between -3 dB and -6 dB in order to avoid clipping in the oversampling process. So the ultimate DAC output will be reduced by that amount. and my amp(s) require 2 V input for full output, and I like to have a bit more than that available for some recordings made at very low average levels, which, here, need a bit more gain to reach satisfying playback levels.

I have had a couple of DACs here to test, which actually did not have enough output level available for my needs, so i like to be sure. It is really frustrating if you have a system set up, when even at full volume, the level is too low for satisfaction! I understand this problem is rare (most users find they might have the reverse problem) but it is real if one has it.

I checked directly with Topping to confirm a couple of things, and the result was good news:

1. The DAC output levels are the same with DSD as they are with PCM: 2 V from the single ended outputs, and 4 V from the balanced outputs

2. The DAC does allow for "Direct DSD Mode" when it is set to "DAC Mode"

So it will have 4 V output and bypass the processing and modulator when in "DAC Mode" and playing DSD. I wish it had a little more gain, but I can live with 4 volts for 95% of the recordings I play.

Here's to hoping that RME might be working on an AKM 4499 DAC, as they will have higher output available.

Here is another thing, this is a bit tweaky, but would probably result in a better system overall. How about a DAC which operates the AKM chip in direct DSD mode, which has 10 V output, and then one can connect it to the new Purifi amp modules, with no input gain stage. The RME ADI-2 has almost enough gain to do this (it would have enough for some systems). With higher voltage running between DAC and amp, the influence of cables on sonics is lower, and noise has less effect on cable transmission as well. This is why pro DACs, and pro line level gear tend to run at higher voltages than consumer gear, to avoid noise pickup on longer cable runs which are used in the pro environment.

Thanks for posting, this is interesting, although I can't pretend to fully understand it. As you know @barrows, I use HQ Player too to overclock to DSD256 (all content), and I have been running in a new amp this week and noticed a few things to do with the settings in HQ Player. I have been using -16.0db's Vol Min and -3.0db's Vol Max. However I had noticed that the volume dial, in the main screen, has gone red a few times due to clipping, I have therefore adjusted it, and I am currently at -7.0dBFS. I assume I am seeing this clipping as a result of the oversampling process. I am wondering if I should have a think about these settings again, how do you set up yours?

I think I need to understand more about these settings, so perhaps I need to research a little more. I currently have Roon set-up for fixed volume and I usually adjust volume using the amplifier control, or in this case, a logitech harmony remote.

 

[barrows]

I currently do not have a machine for HQPlayer here, (but I am in the process of getting something powerful enough to run the EC modulators). So I am running ROON, and using it to got o DSD 256. ROON has a setting which reduces level before oversampling, I set that to -5 dB to eliminate clipping during the oversampling process. So right there I lose 5 dB. With that setting I still have one recording which clips at peaks: this is a Channel Classics DSD recording, which probably has peaks above 0 dBFS (DSD has some headroom above 0 dB, unlike PCM). I have not seen any other file clip. It seems DSD-DSD might require a little more attenuation before oversampling than PCM. But -5 dB is what I use.

So, let's look at this the simple way, -6 dB halves the output voltage. So say we have the D-90, and we are oversampling to DSD in software, and applying a -6 dB (for easy math) reduction before oversampling. This means the output of the D-90 via XLR at 0 dBFS, will be half its normal output, so it will have 2 V output maximum. (normal FS V=4 V, so .5*4=2 V @ FS). So we see if we are oversampling in software, it is nice to have a DAC which might have a little bit more output voltage to accommodate the needed reduction. Of course this should be matched with the amplifier and speakers. I find that with my amplifier (NC-400) which requires ~2 volt input to reach full output, I like to have a little bit more than 2 volt output form my DAC, in order to accommodate some recordings which are at a very low average level (often Audiophile recordings, recorded with some headroom and large DR). My speakers can take everything that the NC-400 can give, so I like to have every watt of of clean power available. What a lot of people do not understand is that what usually damages speakers is not high clean power, it is when the amp is overdriven (clipping) and loses control of its output: most loudspeakers will not be damaged by brief power peaks over their rating, as long as those peaks are undistorted. (unless we are talking about crazy mis-matches, like a 94 dB speaker with a 1000 watt amplifier or something). Accordingly, I build my DIY DACs such that they output ~4.3 V @ FS. I hope that makes sense.

Being someone who is handy with a soldering iron, I wonder if it might be possible to mod the D-90 for a little more output voltage. Might be as easy as changing the feedback resistors on the output stage OPAs. I would just want another 3 dB of gain.