Schiit Modi 3+ Review (Stereo DAC)

[TheMarshal]

Don’t just rely on measurements. I have 2 good measuring dacs: Topping D50s and Auralic Vega G2. I bought the Topping after reading the review on this site to save some money; the Topping should be comparable to the Auralic. But to my ears that is not the case. The Topping sounds very detailed but lifeless and the ‘expensive‘ Auralic Vega G2 outperforms the Topping completely on musicality (I know some people hate that word). This is a conclusion after weeks of trying and comparing. I really wanted the Topping to sound good.

As you already know, this is not that interesting for ASR standards. You cannot just say, "don't rely on measurements" only based on "to my ears", and so son. You see it coming, so... Blind test, level-match, blablabla...

Honestly, both of you are right in some way. My opinion is that you shouldn't take only the SINAD score as the main motivator, you should go through the other measurements too before making a decision.

The word "musicality" is very subjective though. For some of us, it can mean that it should have a flat studio sound, for others it could be the sound of a tube amplifier. There is no right and there is no wrong. At the end of the day, both of us objectivists and subjectivists should discuss, listen to music, and enjoy our hobby.

Oh, and personally, I swear that I can't notice a difference between Fiio K3, Topping E30, Topping DX7 Pro (DAC section), and some high-end NADs that I have heard. However, I clearly noticed a difference between my motherboard's DAC and Fiio K3.

For headphone amps, it's a very different thing, having more power and lower distortion does impact sound quality. For example, the Sennheiser 6xx sound absolutely rubbish on the integrated motherboard headphone out (they are very quiet, there is some clipping and etc), they did sound okay-ish od Fiio K3 (just they need more power), the did sound passable on the Topping DX7 Pro's SE headphone out, they sound awesome on JDS Atom - and they sound amazing on DX7 Pro's balanced out.

 

[VintageFlanker]

My opinion is that you shouldn't take only the SINAD score as the main motivator, you should go through the other measurements too before making a decision.

Haem... Yes. I mean: obviously. What would be the all point of others tests, then?

 

[DuncanTodd]

What are your "transducers" (speakers, headphones)? Aplologies if you're already aware of it, but almost always, its those air moving things that create the most distortion in a system.

And, depending on what you have now, you might be able to make a large difference for just a few hundred, rather than many thousands of dollars, by some informed smart shopping.

I'll agree on it maybe still being a maybe, ;-) but that would be the most likely place/best bang for buck area of resolution improvement.

To be clear, it wasn't like I felt the sound is lacking. To my non-audiophile ears it's sounds perfectly fine. I wouldn't say excellent just because I haven't tried anything else.
I was mostly curious to try out a DAC seeing people are willing to spend hundreds-thousands of bucks on them, and to see if I would notice any difference.

I have "vintage" B&Ws towers (not some fancy model I think) and a "vintage" Nakamichi amp. Both going back 25-30 years. I have a "vintage" Marantz CD player from the same era that I pulled out of storage. It's still mostly working fine but had trouble loading some CDs (mostly worked after a few tries, and a couple wouldn't load). I always disliked using headphones, I have some cheap Chinese ones for emergencies.

Options were to try to service the CD player (maybe like $100? If I can even find someone to do it and do it well) or buy some decent but not too expensive CD player ($300?). Neither appealed to me, so I bought a barely used 2nd hand BD player for $30 (original MSRP was like $500 in 2010). Then E30 for $90-100 (like new) to see if it would improve on what already sounded good enough to me. As you can see frugality is a running theme here

I'm curious to hear how would you improve with a few hundred bucks. Replacing the speakers?

 

[Helicopter]

To be clear, it wasn't like I felt the sound is lacking. To my non-audiophile ears it's sounds perfectly fine. I wouldn't say excellent just because I haven't tried anything else.
I was mostly curious to try out a DAC seeing people are willing to spend hundreds-thousands of bucks on them, and to see if I would notice any difference.

I have "vintage" B&Ws towers (not some fancy model I think) and a "vintage" Nakamichi amp. Both going back 25-30 years. I have a "vintage" Marantz CD player from the same era that I pulled out of storage. It's still mostly working fine but had trouble loading some CDs (mostly worked after a few tries, and a couple wouldn't load). I always disliked using headphones, I have some cheap Chinese ones for emergencies.

Options were to try to service the CD player (maybe like $100? If I can even find someone to do it and do it well) or buy some decent but not too expensive CD player ($300?). Neither appealed to me, so I bought a barely used 2nd hand BD player for $30 (original MSRP was like $500 in 2010). Then E30 for $90-100 (like new) to see if it would improve on what already sounded good enough to me. As you can see frugality is a running theme here

I'm curious to hear how would you improve with a few hundred bucks. Replacing the speakers?

Speakers (and how they interact with your room) and source are by far the most important parts of the chain in terms of getting good reproduction. E30 is good enough, that nothing will actually sound better, so yes, you are on the right path with speakers. Amir has reviewed lots of them, and there are several good choices that stick out.

 

[testp]

I think a kettle is a good analogy for some of this stuff. Some kettles suck; they don't work, or switch off at 98 degrees C, or take ages to reach 100 degrees C. Some clown is going to pay £200 for a kettle because it's the colour as their sports car, or what have you. Amongst the ones which reach 100 degrees C some get there 5 seconds faster. The best 10 reach 100 degrees within .6 of a second. From the outside, if you ask a normal person, they'll treat them all (other than the broken ones) as identical. But we're not normal, and (typically) want the one which gets to 100 degrees C the fastest. I think there's an argument for having the SINAD ranking and other charts show the point where there is no audible difference, so you're purely picking for reason of style (I like silver), or racism (can't be buying Chinese) etc. I have the Topping E30 and the Schiit Asgard 3 with the AK4490 DAC; one DAC measures a lot better than the other but (worthless subjective opinion alert!!!) there's no obvious audible difference to me, nor would I imagine would a test reveal one either. A lot of people are going to prefer the look of the modi 3 over the Topping alternative. But at this point, that's pretty much the extent of the difference. Or features - the E30 has a remote which my Asgard 3 doesn't.

i like kettles that don't think 30 seconds for something (maybe a joke it cant output..) then starts to heat water... and stay at 100 C for another 30 seconds for god knows why.. like my latest philips does, that cost like 70 euros..
yet my mom just got a 24eur Philips without buttons, that just starts when you push the leveler..

 

[Ron Texas]

They finally got their Schiit together. Thank you @amirm

I think we have to step back occasionally and realize a SINAD of 110 is more than enough. Isn't the limit of human hearing 96 db, and that's for young people.

 

[amirm]

1. What is the minimum +/- delta in the volume of that tone for people to reliably detect a difference between the two? How does this delta vary with frequency and does it have any correlation with the minimum audibility threshold at that frequency? How does this delta vary with the volume you are initially listening at?

I know of a rough published study by Clark in an AES paper:

It will take me a while to find the paper if you need it (had saved the above years back for another argument). Note that the study was not rigorous.

 

[boselover61]

They finally got their Schiit together. Thank you @amirm

I think we have to step back occasionally and realize a SINAD of 110 is more than enough. Isn't the limit of human hearing 96 db, and that's for young people.

I'd be more concerned about the multi tone test. Topping E30 has much better results. Imo the only performance-impacting stat I look after sinad is multi tone. This ones quite indecent yeah?

 

[andreasmaaan]

I'd be more concerned about the multi tone test. Topping E30 has much better results. Imo the only performance-impacting stat I look after sinad is multi tone. This ones quite indecent yeah?

The spuriae in the multitone test are 100dB below the signal level. If you use software to play a signal at -100dBfs with your system set to the loudest level at which you normally listen, this will illustrate how loud -100dB is in your listening environment.

 

[Vasr]

I know of a rough published study by Clark in an AES paper:

View attachment 99882

It will take me a while to find the paper if you need it (had saved the above years back for another argument). Note that the study was not rigorous.

Thanks. I am not sure I understand what the graph is saying. Is it about minimum change in volume of the same tone (frequency) to detect the change (my question) or is it about detecting other related tones in comparison or together (which is more of a masking issue)? Not sure what those Octaves mean in this context.

 

[Vasr]

Just had a quick re-read of the relevant chapter of Zwicker and Fastl. Your second point is partially addressed Ch. 7.1.1 Threshold of amplitude variation.

In this section, a study is described in which increasingly narrow bands of white noise are amplitude modulated, with subjects tested to determine the threshold at which amplitude modulation becomes just-noticeable, and the dependence of this (if any) on the bandwidth of the white noise.

The authors note that, for spectral widths of band-passed noise larger than a critical band, the just-noticeable degree of amplitude modulation decreases. In other words, the wider the bandwidth of the signal, the lower the threshold of detection for amplitude modulation.

They go on to say:

Thanks. Looks like I can get that book as a rental on Kindle. Will take a look when I get the time.

What I am trying to explore purely from intellectual curiosity (not getting into the ideological battles of hearing differences) is the sensitivity to human hearing in both static and dynamic contexts. Trying to figure out which studies have studied which.

There are three differentiating contexts I can think of:
Taking the simplest example of volume
1. The ability to hear a signal at constant volume. Thresholds apply here. Studies seem to have established this fairly rigorously.
2. The ability to differentiate between two volume levels in audible range for the same tone. How much should the volume difference be before people can differentiate them as different vs the same. The testing here is a discrete switching test between the two levels.
3. The ability to detect movement when it changes from one to the other smoothly. We know from visual acuity that people are less sensitive to differences between slightly different pictures than to the movement between the two. Perhaps similar things apply to audio as well.

A thesis is that the differentiating capability of humans relies on cues from all those three dimensions (which are in play while say listening to music) and therefore two devices must be equivalent in all those dimensions for people not to perceive a difference between the two.

Before trying to prove or disprove that thesis, I am trying to classify studies and measurements as to which of those dimension(s) they fall under and whether they cover all of those dimensions to make definitive statements about the ability to differentiate from the current set of measurements and current set of studies.

You can then construct similar dimensions for phase differences, harmonic content, channel balance, etc.

This is off-topic for this thread, so I will follow up in a separate thread when I have more info. If you have any other pointers in addition to the book that may be relevant to the above, do let me know.

 

[amirm]

Thanks. I am not sure I understand what the graph is saying. Is it about minimum change in volume of the same tone (frequency) to detect the change (my question) or is it about detecting other related tones in comparison or together (which is more of a masking issue)? Not sure what those Octaves mean in this context.

It is required level match vs frequency for it to not be detectable. The different curves are for filtering music to that level of bandwidth for the level measurement. The different curves are different levels of filtering. It is saying that if the source has more varied amplitude with a window, then it matters less than if the whole segment was at higher value.

I found the paper it is from:
High-Resolution Subjective Testing Using
a Double-Blind Comparator*
DAVID CLARK
ABX Company, Troy, MI 48099, USA

There is just one a short one-paragraph mention and no more detail about it.

 

[andreasmaaan]

Thanks. Looks like I can get that book as a rental on Kindle. Will take a look when I get the time.

What I am trying to explore purely from intellectual curiosity (not getting into the ideological battles of hearing differences) is the sensitivity to human hearing in both static and dynamic contexts. Trying to figure out which studies have studied which.

There are three differentiating contexts I can think of:
Taking the simplest example of volume
1. The ability to hear a signal at constant volume. Thresholds apply here. Studies seem to have established this fairly rigorously.
2. The ability to differentiate between two volume levels in audible range for the same tone. How much should the volume difference be before people can differentiate them as different vs the same. The testing here is a discrete switching test between the two levels.
3. The ability to detect movement when it changes from one to the other smoothly. We know from visual acuity that people are less sensitive to differences between slightly different pictures than to the movement between the two. Perhaps similar things apply to audio as well.

A thesis is that the differentiating capability of humans relies on cues from all those three dimensions (which are in play while say listening to music) and therefore two devices must be equivalent in all those dimensions for people not to perceive a difference between the two.

Before trying to prove or disprove that thesis, I am trying to classify studies and measurements as to which of those dimension(s) they fall under and whether they cover all of those dimensions to make definitive statements about the ability to differentiate from the current set of measurements and current set of studies.

You can then construct similar dimensions for phase differences, harmonic content, channel balance, etc.

This is off-topic for this thread, so I will follow up in a separate thread when I have more info. If you have any other pointers in addition to the book that may be relevant to the above, do let me know.

I'll have a think about whether there's anything else I can recommend. The book though is a more than excellent start.

Off the top of my head, the studies discussed in there that will address some of the areas you've raised are those that look at temporal masking (masking of a maskee by a masker that comes either before or after it in time), influence of the presence of partial maskers on just-noticeable difference thresholds, temporal loudness effects, and so on. The basic physiological models of the auditory system are also described and sections of the book seek to connect the observed psychoacoustical effects with the physiological models.

The final chapter (which actually I haven't read) also attempts to link the research to various real-world applications. I guess it might discuss some of the ideas you mention towards the end of your post, although the focus is not particularly on music reproduction AFAIK.

 

[PeteL]

USB / SPDIF - If Schiit can clean up the USB signal using their proprietary design, shouldn't they be able to do the same for the SPDIF?

The challenges are quite different, and the work they did on USB is likely not applicable to the other. SPDIF is a real time protocol and UAC2 is an asynchronous protocol. At least theoretically USB is imune to input Jitter, all timing/clocking issue are intrinsic. It is however not fully imune to noise and data integrity is not guaranteed by any error correction provision. SPDIF, again theoretically, is not imune to jitter from the source or during transmission. Good SPDIF receivers do however perform reclocking, to a limit, they can't magically fix a significant clock drift. This is a couple exemples where the fundamentals of data transfer differs, it's more complex than that but a spdif receiver works quite differently than a USB receiver.

 

[JohnYang1997]

I registered just to respond to this.
Very well said; I still have my two original modi 3s and they've been rock solid reliable and sound great. I've seen little reason to "upgrade" to a same-tier dac.

However, I did want to address the bolded. Wanting to support one's home country, or not wanting to support another is absolutely not necessarily racism. It's actually pretty irritating that is the assumption. Personally I don't care where a product comes from, but the company that manufacturers that product needs to have a presence in my country. Otherwise if something goes wrong, you have to ship it off to the product's home country, which isn't always inexpensive or fast.

Case in point for me, topping. They make some great products (though first batch QC is a big gamble with them), but I can't bring myself to spend money on something topping makes because of the reason I mentioned above. If I buy an D10S, and something goes wrong, it's a minimum of $37 to ship it off to china (from the US), and it could be well over a month before I get it back. Whereas, if one of my modi 3s gives up the ghost, $9 to cali and I'll have it back in a week or two.

So again, well-said, but please don't assume racism because someone chooses to support their own country, or not support another.

That's why distributors exist.

 

[JohnYang1997]

Not trying to bash either side of the arguement, but I quickly checked the reviews here, it seems that the ESS hump was there with the D50 and not with the Vega G2, and that hump was around -30 to -50db range, which I suppose is more likely a use case for the dacs if one is using digital volume control? if so then maybe that IMD is the un-natural or not musical part you've heard?

Not with D50s D10s.

 

[YSC]

Not with D50s D10s.

um~ I missed out the s version in the original post, did a quick read and the D50s have slight roll off from 10k-20khz and maybe that contribute to the detail retrieval? and one more thing I am thinking in a noob wild guess is about the amount of negative feedback used, will a lot of negative feedback results in some time domain problems which can make apparent sound difference? or ultrasonic noise or such... just wild guess

 

[Ron Texas]

I'd be more concerned about the multi tone test. Topping E30 has much better results. Imo the only performance-impacting stat I look after sinad is multi tone. This ones quite indecent yeah?

Indecent, like full frontal nudity?

 

[trl]

Sure, there are things ahead of it on the chart, but how many are ahead of it per dollar?

Still missing a reply here, so you might want to start counting items from the chart and compare prices. I find the Modi 3+ being the cheapest among items having the same small desktop-size, given that Loxjie D10 and Topping E30 (direct competitors) are having a starting price of 129 USD (not mentioning that these need to be sent in ASIA for repairs).

 

[trl]

Some kettles suck; they don't work, or switch off at 98 degrees C

Off-Topic: This is the perfect protection and it's a valid setting for most digital kettles (it's setup by manufacturers from the built-in firmware), so shutting down a kettle at 98 C instead of 100 C it's a highlight, not a lowlight. Expecting a kettle to shut off at 100 C means that in few seconds after shut down it will overheat and bubble off water from the kettle if overfilled above the max. sign (injuries may occur).

I have a digital kettle that shuts off at 98 C and this is absolutely perfect operation, giving heating resistance latency (it gets to 100 C anyway after few seconds from shut down).