The Mola Mola Tambaqui DAC

[Miska]

On 1/2/2019 at 9:26 AM, firedog said:

It doesn't mean all units sound the same, or that expensive DACs are for idiots, but I think it is useful data to see that sometimes more modestly priced components measure as well or better than more expensive ones, and that some of the tweaks that audiophiles think are so important actually don't seem to make a difference at all at the output.

 

I'm most worried about the expensive gear that measures badly... I sometimes wonder if the manufacturer ever measured their creation...

 

[Miska]

1 hour ago, Em2016 said:

We all remember the Stereophile Playback Designs MPS-5 measurements ... and the manufacturers response...

 

https://www.stereophile.com/content/playback-designs-mps-5-sacdcd-player-manufacturers-comment

 

“As we are always researching new ideas, the next-generation algorithms may very well make these kind of measurements even worse”. 

 

Full response at link above.

 

Well, I don't think it measures badly... SNR could be better, but I don't see big faults. Would be nice to get my hands on Playback Designs Merlot DAC. I've heard the Nagra Classic DAC running at DSD128 (from HQPlayer) and it sounded really good, and Andreas Koch is one of the designers behind it. Hoping to hear the HD DAC X at Munich, but unfortunately probably not through HQPlayer.

 

[Miska]

4 minutes ago, Em2016 said:

I’ve also heard the MPS-5 and Merlot DAC (latter directly driving my headphones) and it was incredible.. in fact I can still remember how the music sounded and the feeling. And that was even after reading those measurements. Whether I enjoyed more distortions or whatever, there was something quite special.

 

Only "fault" JA complains about is random background noise. If you can hear it, it is just like tape hiss or radio background noise. Important thing is that it is not correlated with the music. But the distortion levels are low.

 

Sometimes I tend to disagree with JA's conclusions. For example if you look at this:

https://www.stereophile.com/content/kalista-dreamplay-one-cd-player-measurements

 

[Miska]

10 hours ago, guiltyboxswapper said:

Have any Tambaqui or Makua DAC owners tried HQPlayer with this DAC?  Given the design does it reap much benefits?

 

Based on their documentation it converts DSD to PCM first before converting it again to 1-bit stream. So it doesn't play DSD natively. You likely get better results by sending highest supported rate PCM from HQPlayer instead. Of course that way, all the various upsampling filter options, convolution engine for digital room correction, etc of HQPlayer is still available and fully usable.

 

[Miska]

5 hours ago, Davidny said:

In the Tambaqui  “...all incoming digital audio is upsampled to 3.125MHz/32 bits and converted  to noise shaped PWM”  Given this internal processing, which seems unique and to my ears makes the Tambaqui sound very special, I doubt that pre processing the digital signal with add anything, and more likely will subtract from the end result.  

But if you try HQP let us know what you hear, one never knows unless you try it for yourself.

 

Generally, the first rate conversion step up from RedBook is the most critical, then the following things are less, until you hit the modulator. Modulator is something you cannot replace or change in the Mola-Mola, you always get one fixed choice and that's it.

 

4 hours ago, guiltyboxswapper said:

I agree with your comment - its likely the best of HQPlayer cannot be achieved with this DAC.  Which leaves me at an odd cross-road - can a good direct (no decimation, tampering of DSD) 1-bit convertor with a streamer endpoint good enough to keep up, match the Tambaqui?

 

You could compare it to the T+A SDV3100 HV running at DSD1024. Would be interesting to hear how people think the two compare. Unfortunately I have neither one, so I don't know...

 

[Miska]

On 12/24/2020 at 6:54 PM, barrows said:

I won't speculate on the audibility of this phenomenon but anything that is measurable is fair game for me.

 

Nice to see that someone is along the same lines on this with me. So many people screaming at ASR that "yes you can measure a lot of differences, but you won't hear any of it".

 

On 12/24/2020 at 6:54 PM, barrows said:

Perhaps the idle tones Bruno speaks of are responsible for the relatively obvious (to me) sound quality advantage of the Tambaqui vs. common SDM chip based DACs, perhaps it is something else.

 

I personally don't get stuck to just single aspect, because there's so much more going on, all measurable.

 

On 12/24/2020 at 6:54 PM, barrows said:

What is clear from the above is that all aspects of DAC performance cannot be expressed by a few common measurements: if one is looking to produce a superior product, one has to explore beyond the standard single tone responses.

 

I completely agree with this too.

 

[Miska]

On 3/31/2021 at 6:33 PM, ted_b said:

Same goes for DSD files....thx

 

Mola-Mola DACs convert DSD inputs to PCM. So if you use HQPlayer with those, it is best to always send highest possible rate PCM there and never DSD. (similar way as with Chord DACs)

[Miska]

3 hours ago, skatbelt said:

I agree with never DSD on Chord DAC's but my DAVE likes PCM at native rates the best. So no upsampling in Roon or HQPlayer for me.

 

Well, you like the native rates best. ;)

 

IOW, Chord's upsampling to 705.6/768k PCM.

[Miska]

13 hours ago, Nenon said:

It seems like heavy upsampling on the Mola-mola is a major part of its design and cannot be bypassed completely. There is a chance that sending the max 384kHz rate would reduce the processing by a fraction. But in that case we would be trading a fraction of less processing in the DAC for more real time processing on the server side. In other words, hard to predict what would sound better. 

 

First step up is the one that makes biggest audible difference. So if you get started with 44.1k and output 384k, then the next step from there on has much less audible impact on the sound. One big benefit in the first step comes from use of apodizing filters which allows to fix up some problems caused by downsampling in the original source material.

 

Of course this is still different than doing let's say upsampling from 44.1 kHz to 11.2896 MHz like you would typically do in HQPlayer. Not to mention effect of different modulators. But it is not black and white.

 

[Miska]

14 hours ago, barrows said:

Given how good it sounds with native rate files, it makes me wonder if this might be one of a very few DACs which might not benefit from external oversampling via HQPlayer

 

It depends if you like the digital filter it has or not, I don't know how many different filter options MM DACs offer. For example lot of Chord users still use HQPlayer upsampling.

 

I'm usually changing filter based on type of source material, because I know certain material sounds better with certain filters.

 

[Miska]

41 minutes ago, skatbelt said:

I am not one of them. Everything I tried (including HQPlayer and even with redbook material) led to suboptimal results. My suspicion is that if Chord DAC users prefer upsampling before their DAC, it is because they have some kind of phase problem elsewhere in their system that gets somewhat masked with upsampling. Or that they simply prefer a more smeared and rounded sound signature.

 

Suboptimal in objective or subjective sense? I mean by measurement results or by listening?

 

I don't think there's a technical way for upsampling to mask phase problems. Most of them use sinc-M filter which is apodizing, but technically takes you beyond the M-Scaler.

 

Or they don't like the trashy messy exaggerated highs that you get with most modern RedBook content without apodizing filters...

 

[Miska]

11 minutes ago, skatbelt said:

What I can say objectively is that you have a double hat on in these kinds of discussions.... :)

 

Sure, it is not a secret and clear to most people reading. I just want to make things clear objective vs subjective. 

 

Digital filters can be categorized in various ways. From objective point of view each have their own properties. There are certain categories of technical problems in lot of RedBook source content that can be fixed with apodizing filters. Digital filter in Chord DAVE is non-apodizing and thus it reproduces the errors/problems in the source content.

 

15 minutes ago, skatbelt said:

In the sense that spatial clues, perceived depth or very small details (for example on stringed instruments) disappear or become less apparent.

 

Now the next question is which musical genres you listen. With classical music you are less exposed to the digital filter effects I'm talking about than with rock/pop or even jazz/blues. Musical genres you listen influence the digital filter preferences, because same filter is not necessarily optimal for all genres.

 

In the sense how clean is for example a cymbal snap, does it have artificial "sheen" around it that makes it sound brighter but more messy? Someone could mistake such for "more detail". These are caused by error components in the source data that are caused by poor decimation filters at the production side and shouldn't be there since they are not in the original analog signal, but instead digital artefacts. Once you learn how the errors sound, you can detect those pretty easily with suitable material.

 

Question is also not so much how good some good recording can sound. But instead how good you can make a bad recording sound. Because latter one is more difficult case.

 

 

Of course everyone has their own preferences regarding digital filters. Different people are sensitive to different aspects of the sound, etc. Nothing wrong with that. Some may even like MQA's upsampling filters... ;)

 

It doesn't mean though that we couldn't also discuss technical aspects.

 

[Miska]

40 minutes ago, ted_b said:

I can confirm that the DXD file list Craig supplied does indeed have issues when playing via the Tambaqui, and yet plays fine via my Holo May.  Mola Mola techs have confirmed it, too, and supposedly were creating a firmware update to solve it (seems some noise on the files that other dacs filter out the Mola Mola does not.....which leads me to want to investigate the noise source...and many of these we sold him).  This is DXD only, and a few selections (same label, same artist, same cycle...rest are fine).  Weird.

 

It's not about some DACs filtering something out, but more about the modulator design. Likely it trips up the Mola Mola modulator...

 

2 hours ago, ted_b said:

I'll also spend more time on DSD sources to see if these are rendered with the same care and precision of the May at DSD256 EC7V2.  Most of my DSD listening will be multichannel (although I have a lot of 2 channel DSD) so Tambaqui's resampling of DSD is not a deal breaker, assuming it is not the giant leap that PCM is.

 

Too bad it behaves like Chord, converting DSD to lower rate PCM first and then back up. It could as well feed it straight to the conversion stage, but for whatever reason is not doing it.

 

[Miska]

On 2/13/2022 at 6:02 PM, ecwl said:

T&A and most other brands are using multi-element multi-bit delta-sigma DAC chips with their own upsampling algorithms but the DAC chip still uses thermometer code to dynamically select the elements (DEM).

 

T+A, Holo and some others have multi-element DSD DAC that you can run from external oversampling and modulators... 😉

 

On 2/13/2022 at 6:02 PM, ecwl said:

I believe the advantage of the analog moving average filter is that you do get less glitch energy with each flip-flop switching compared to a DAC chip with dynamically selected elements

 

That is also what I did in my DSC1 DAC 8+ years ago (relesed March 2014, but I started working on it in spring 2013).