DSD upsampling trial

airdronian

Radar Member
Site Supporter
I've been curious for a long time about treading the up-sampling to DSD path. Thought I would give it a try, in my usual budget oriented fashion via the Topping D10.

ToppingD10.jpg

There's not much to it - powered from the USB bus, so no wall wart. It can also serve as a USB/SPDIF converter. Quite small, but solid feeling. Plugged it in, and the Mac Mini recognized it right away. It is said to reproduce PCM to 32/384 and up to DSD256, no drivers required for macOS.

It played Spotify at 384 khz PCM, which seemed to be controlled from Audio Midi that detected the max setting. Seemed OK. Playing a redbook file at native sampling rate I was unsure of the audio quality - quite a different presentation than the Bifrost. But that's not really why I brought this in.

I had downloaded several DSD files for testing. However, so far I cannot get them to play as DSD - it down samples to PCM 32/352. I also tested with up-sampling PCM to DSD - this too did not work. Settings changes in Audirvana Plus demonstrated it needs to do DoP in order to get this to work. So with the native DSD handling set to DSD over PCM standard 1.1 it shows the DAC is capable of DSD128. So currently I am up-sampling any of the library files to DSD128. It will now also play the native DSD128 files.

And like I've read here at the Haven, the DSD up-sampling is a bit of a game changer. I've just shuttled thru the various filters, and A (5th order) seems to give me the balance of detail and bass that I like. I can't see myself not upsampling with this little DAC.

Still need to fiddle a bit, but I'm mostly there. I'm thinking the issue of native DSD playback may be more to do with El Capitan than the DAC. Will continue to look into that. And on the 'net there's video of a fellow using Volumio2 on an RPI3 playing back DSD files. Didn't think the Pi would be a suitable choice for a bus powered DAC, but there you go. Fun and games.
 
Sounds like fun. :)

Do you have Tidal? Wondering what happens if you run Tidal straight in without going through AudirvanaPlus?
 
Cool stuff, I'm still on Audirvana v1.5 on the Mac, largely because I migrated to JRiver as my primary music player some years ago. That said, maybe I should revisit Audirvana in it's most recent incarnation?

The Mac platform via Core Audio only supports DoP as far as I know. Several years ago exaSound wrote a Mac extension for Steinberg's superb ASIO driver, and that does allow native DSD on a Mac when used with an exaSound DAC, however I don't think anyone else has followed suit.

I know at the time Decibel was the only compatible software music player for that, but maybe others now also work, however you'd need an exaSound DAC and their ASIO driver too.

Yes Moode also supports DSD via the RPi3's USB output and has for years, it works fine except not (yet) with I²S DACs, currently it is limited to USB only on RPi3, it gets downsampled to PCM automatically when using an I²S connected DAC.
 
Still need to fiddle a bit, but I'm mostly there. I'm thinking the issue of native DSD playback may be more to do with El Capitan than the DAC.

Unfortunately it's any version of OSX, not just El Capitan. This has been true since the dawn of OSX (2001), when Macs dropped all ASIO support in favor of Core Audio. The DoP workaround was developed sometime thereafter as a result (2011 or so).
 
Sounds like fun. :)

Do you have Tidal? Wondering what happens if you run Tidal straight in without going through AudirvanaPlus?
I had Tidal in the past and cancelled. I prefer Spotify, even without redbook streaming. I did use Tidal's desktop app for a time, and when I was able to try the A+ Tidal integration that won me over.

@MikeyFresh it's my understanding that by enabling Direct Mode in A+ it bypasses Core Audio. Also I'm using the iZotope 64 bit SRC. A+ has changed over the years, current version is 3.2.8 and late Summer or Fall will see a Windows version released and a new UI. At any rate, DoP isn't a deal breaker for me.

In the process of getting some hours on this little DAC. The display you see above will dim after several seconds of playback, so it is unobtrusive and a welcome change from the bright white LED of the Bifrost.
 
Will the DAC upsample Spotify without AudirvanaPlus?
Spotify doesn't integrate with A+ so you need to use it's desktop app or web interface (I use the former). It appears to take whatever Audio Midi is set to - in my case it had defaulted to the top PCM rate of 384 khz and that's what was displayed during playback. I imagine I could adjust that. Just played Scofield's "Country For Old Men" on Spotify and it was sounding better than it had any right to.... :)
 
At any rate, DoP isn't a deal breaker for me.

It shouldn't be, though there are additional CPU overhead burdens imposed by DoP, it's never really been proven to sound worse than native DSD, that would likely differ system to system depending on a lot of variables. DoP sounds fine to me.

Audirvana Plus in Direct Mode probably doesn't fully bypass Core Audio, just most of it. Either that or it still needs something additional, similar to that ASIO driver extension utilized by exaSound which I believe is coupled with Exclusive Mode and Integer Mode. That definitely is in part dependent on the specific version of OSX in use, so maybe Direct Mode is similar, perhaps you are right and El Capitan isn't 100% compatible with Direct Mode.

Either way, enjoy the sound! Some years ago when I switched to JRiver, I did so for features such as DLNA/UPnP, not for sound quality. Although JRiver does sound very good, I did feel at the time I stopped using A+ that it did result in a slight drop in sound quality as compared to A+, which was just a tad more natural, full, and organic sounding.

Others posted that very same thing on the JRiver forum back then, and it prompted a very nasty response from the JRiver developer along the lines of "don't you morons know bits are bits and JRiver is bit perfect?". I steered clear of that entire discussion!
 
I'm thinking this could be Gateway material, if USB connection isn't an issue. Massdrop has them on for one more day, $75 for US customers. We'd get dinged for exchange, shipping, and an extra $10 for Customs. Doesn't ship till August though. Mine was thru an Amazon vendor.
 
Just a quick follow-up on the Topping D10, not entirely off-topic but no upsampling trial here, just that Massdrop has had these as low as $69 periodically and so I picked one up for use with one of my Raspberry Pi based streamer endpoints running Moode Audio Player.

D10-RPi3.jpg

Still not too many hours of use on it just yet and it's not hooked up to a super critical listening system, however it certainly sounds really good and is kind of unbelievably well made for $69.

About the only corner they seem to have cut is the warranty, or maybe I should say lack thereof. There is broken English on the tiny warranty card suggesting you need to contact the seller if there is a problem, and only failing that contact Topping (in China, no N. American representative). You are then on the hook for shipping cost to China, with a "1 month free replacement guarantee". After that 1 month, "1 year of free maintenance" (I guess labor?).

So pretty much no warranty, but I'm not sweating it at $69 and I don't expect any issues given the apparent nice build quality. You can even swap out the op-amp output stage, it is DIP8 socketed, allowing sort of a solid-state version of tube rolling.

All in all very high value at $69, but if the non-existent warranty is bothersome, then maybe an iFi nano iDSD LE at $129 is a better choice with the added bonus of onboard battery power/portable use.

Thanks to @airdronian for pointing out the Massdrop availability on these, you really can't miss at $69, most stuff at that price point is complete garbage. This is a very nice unit, definite doorway material if the USB input is sufficient.

With coax and Toslink SPDIF output for connection to "legacy" DACs, this is also a $69 USB -> SPDIF converter.
 
Yup, no complaints at all about the build quality. I've run this from my Mac Mini as well as a Pi. I really like the way the display dims automatically, it's less obtrusive that way.

Day to day in the main rig I'm continuing to use the Bifrost 4490. This has the Gen 2 USB and I find I prefer to use the D10 for USB/SPDIF conversion. I do change integer mode in Audirvana to adjust the sound - it's more analytical than the Bifrost USB connection. Works well though, no real complaints.
 
I really like the way the display dims automatically, it's less obtrusive that way.
Yes and also cool how it briefly lights up again on sample rate changes.

So if you make a playlist of tracks that are mix of Redbook, DSD, and hi-rez PCM, each time a track with a different sample rate starts, the display briefly brightens again on that transition, then dims again. If nothing else that indicates mature firmware, something many of these cheap DACs don't have, and some of which never will.
 
I did encounter a hiccup with DSD upsampling now that I think of it. On some tracks it would pause and others would work.

After watching the Audirvana interface while in playback, you could see the progress bar stop as it was preloading the file, and at this point it would pause. Since the software is upsampling the file to DSD I am thinking the resultant file size may have exhausted the preload limit that was set, of around 2 GB. There's 4 GB memory installed total, and that's not going to change.

With all of 8 tracks of DSD music on hand it was not an issue worth pursuing.
 
Revisiting this after a while of mixing and matching between the D10 and the Bifrost, with output from Audirvana +. Using either alone, or with the D10 as usb converter. The DSD upsampling was a bit problematic. On PCM the D10 had more precision than the Bifrost, but this came with a bit of a penalty - a clinical nature, perhaps a little "sterility". This was really noticeable when using Integer Mode 1 in A+, which is described as more transparent, as opposed to the warmer Mode 2.

Added an iFi iSilencer 3.0 to the USB chain. The little gadget seems to help both DACs.

Reading other's experiences that were similar to mine with DSD upsampling, one comment was to avoid upsampling to DSD and just stay in PCM. So today I set upsampling for the 44.1 khz family to 352.8 and the 48 khz family to 384. Now this is more like it. More natural, less clinical, even in Mode 1. Zero issues in playback.

I'm going to live with the D10 in this configuration for a while and see how it treats me.
 
I'm going to live with the D10 in this configuration for a while and see how it treats me.

I'll have to give that a go, I did eventually upgrade my older v1.5 instance of Audirvana+ to current status (now 3.2.14 as of today's update) after seeing they offered a big discount to previous buyers.

Today after checking the mail box, I found the SparkoS Labs SS3602 Dual Discrete Op Amp had been delivered. This is a drop in replacement for the Topping D10's stock Burr Brown OPA2134, same DIP8 mount, no soldering required:

SS3602.jpg

While both the D10 in general and the OPA2134 are considered to be very good sounding, I couldn't leave well enough alone and ordered a discrete dual op amp for a rolling experiment, all in the name of science.

This is especially questionable thinking, given the output stage in the D10 doesn't even use the OPA2134 for gain, it is a passive buffer. So the SS3602's 20dB greater gain capability compared to the OPA2134 is in this case moot. Perhaps the 3 nV/√Hz noise spec of the SS3602 will actually result in a sonic difference as compared to the OPA2134's noise at 8 nV/√Hz.

Toss in the fact the SS3602 costs $79 while the entire Topping D10 goes for $69 on Massdrop, and I must be nuts. I'm not at all jazzed by the specs in this or many other cases, I just like messing around with stuff and the DIP8 socketed op amp in the D10 is just too easy.

If there is an expectation bias here, it's for a small but noticeable difference in sound. And different may not be better, one of the generally accepted traits of the stock OPA2134 is a rather warm/musical sound quality, not at all etched, analytical, or fatiguing. We shall see, if I were to fault the D10 in any way it is perhaps just a bit too mellow and forgiving, but not necessarily on a detail retrieval basis, more so dynamically speaking. Maybe a discrete buffer stage will jazz things up a touch.
 
I haven't thought about opamp rolling. I'm still upsampling as above, seems to work quite well. In fact I may send the Bifrost down the road as it can't manage what I'm doing with the D10. And I think the D10 good enough to tide me over until I find something else. It will have to be another budget piece, no funds for something like a Holo Spring.
 
I haven't thought about opamp rolling.

From my investigation, there aren't many discrete designs that will actually fit under the D10's hood.

The Bursons are far too tall, they won't come close to fitting. Even the SS3602 I fitted doesn't leave a lot of room to spare in terms of height, and it is billed as the world's smallest discrete op amp, 33% smaller than any other discrete design.

Here is the D10 with the lid removed, the socketed OPA2134 slightly left of middle:

opa2134.jpg

And the somewhat taller SS3602:

SS3602 installed.jpg

I thought it might sound a little hard or congested right out of the box, but that wasn't the case, maybe these don't really need burn in or they get some burn in from the manufacturer as part of the QC, but it sounded quite nice straight off.

Seems like a little more extension at each frequency extreme, and better bass definition, a touch more clarity/detail and a touch less mid bass warmth. Overall slightly more neutral and dynamic, but not a huge difference by any stretch.

We'll see if maybe it does actually settle in more or not, but it didn't super alter the character, and it didn't add any glare, hardness, or tizz, same smooth character as with the OPA2134.

EDIT: The Topping D10 is currently $64.99 on Massdrop, the lowest listing yet. This is a steal for anyone needing a DAC with a USB input.
 
Last edited:
Hey @MikeyFresh have you tried the D10 with any USB conditioning like a Jitterbug or similar ?

I guess I have committed to another DAC, my Bifrost shipped out today.
 
Hey @MikeyFresh have you tried the D10 with any USB conditioning like a Jitterbug or similar ?

I guess I have committed to another DAC, my Bifrost shipped out today.

Yes I use the Wired For Sound Recovery, powered by an iPower (with JS ground shunt).

This re-clocks the output of the Raspberry Pi, and provides clean power to the D10 in one shot. Also using an UpTone Audio USPCB there as it seems the best USB cable is no USB cable. That actually works better on the Recovery's input side because you can then cut the USB bus power and ground, providing electrical isolation between the RPi3 and the Recovery which has it's own power supply and doesn't need any bus power from the RPi3.
 
It may not matter to the audio subjectivists but Audio Science Review measured the D10 with a variety of opamps and found the OPA2134 to have the best specs of the six or so that they tried. Just sayin'.
Also, the replaceable op amp is the low-pass filter and output buffer so it may not really have an effect on the sound beyond what the difference between brands of resistors would be as a comparison. Obviously it's an op amp and not a resistor but it may have similar effects on the sound.
 
Back
Top