Schiit Magni Mesh DAC Review

[Joe Smith]

They do have an APx555b in house, how can they still release it like this? I don't understand.

And, on a side note, looks like they did not update the report with the MESH card in place, as the report linked from the Magni Unity specs page has a 12/14/2023 date.

 

[solderdude]

They do have an APx555b in house, how can they still release it like this? I don't understand.

Because you won't catch this with an AP555 when there is no ground loop. You will need to inject common mode currents which seems to happen in Amir's case.
They may be using/wiring the DUT (Device Under Test) differently.
The correlating noise (the injected noise that seems to have a spectrum) is 100dB below the 0dBFS and mostly low frequency and thus inaudible.
Besides ... it might well remain below audible thresholds in most cases It is highly dependent on the amount and type of injected common mode/leakage currents.
The measurements are good enough for home audio when used as intended when not plagued by common mode/leakage currents.
How much uncorrelated noise there is that could affect the sound will differ from situation to situation.

I know they're currently printing boards in house and the Mesh was one of the first things they learned to do it with, wonder how much that resulted in a gimped performance.

It is about the practical design of the circuit board. How to route the ground path and how to 'separate' yet connect the digital and analog grounds (which have to be connected).
This is difficult whenever more than 1 ground (digital and audio) convert and 2 ground paths exist.
In this case there is (most likely) a single ground for the audio and power supply feeding both the digital and analog parts in the DAC.
I expect this to be a possible problem.
Could be a combination of an inexperienced PCB designer and lack of EMC testing (just stabbing at it).
It is just my suspicion.

 

[HalSF]

I appreciate independent measurements and evaluation that holds brands to high engineering standards of performance.

But after reading this review and subsequent comments that seem to cast doubt the audibility of the USB noise that seems to be the crux of the not-recommended conclusion of the review, I wonder if someone could explain to me what I’ll actually be hearing when using this product that makes it bad.

 

[solderdude]

I appreciate independent measurements and evaluation that holds brands to high engineering standards of performance.

But after reading this review and subsequent comments that seem to cast doubt the audibility of the USB noise that seems to be the crux of the not-recommended conclusion of the review, I wonder if someone could explain to me what I’ll actually be hearing when using this product that makes it bad.

In case the common mode currents/leakage currents reach audible levels you might hear soft unwanted sounds or noises when there is no music playing. The noises will not be music related. The most likely scenario is that the Magni + DAC card will sound just fine.
It just is not SOTA performance what concerns measurements.

 

[Joe Smith]

In case the common mode currents/leakage currents reach audible levels you might hear soft unwanted sounds or noises when there is no music playing. The noises will not be music related. The most likely scenario is that the Magni + DAC card will sound just fine.
It just is not SOTA performance what concerns measurements.

And, as mentioned above, a $30-$50 galvanic isolator between the streamer/computer and the dac should eliminate most if not all of that.

 

[VientoB]

I appreciate independent measurements and evaluation that holds brands to high engineering standards of performance.

But after reading this review and subsequent comments that seem to cast doubt the audibility of the USB noise that seems to be the crux of the not-recommended conclusion of the review, I wonder if someone could explain to me what I’ll actually be hearing when using this product that makes it bad.

I think it's more to do with the engineering oversight that has led to poorer than usual measurements.

 

[shuppatsu]

They do have an APx555b in house, how can they still release it like this? I don't understand.

@solderdude's response to this is more definitive, but here's my "yes and..." Forgive me while I repeat myself, both in this thread and more broadly across multiple threads.

The most direct response to your question is that even if they got the same results from their AP that Amir did, those results show them that the product is ready to push out the door.

1. You can't get more transparent than transparent. Engineering for engineering's sake is fine, so long as that's what their customers want. But if we care about high fidelity this gets the job done as well as anything Topping ever did. Why add to the price? This is literally the cheapest product in their range. I defend ASR from outsiders when they say we cargo cult the measurements, but the flip side of that is that I need to bring it up when we start to cargo cult the measurements. Yes, I believe measurements capture the whole story from a sonic standpoint, and evaluation based on measurements is a thousand times better than relying on Jason Victor Serinus' or some rando from Head-Fi's subjective opinion on something. But transparency is a bar, and once you meet it, there's little value in going further.

2. I've read complaints over the years about Schitt pots being prone to getting scratchy or having channel imbalance as they age. But we don't evaluate on that basis, because it's purely anecdotal. It's bad data, garbage-in-garbage-out and all that.

And yet if there is any semblance of truth to this, it's 1,000x more damning than any of the measurements we see here. Which is to say, these reviews should not be considered to be comprehensive as to whether the device is any good. The measurements are comprehensive, but that doesn't tell the whole story. From a sonic standpoint, it tells us that we have the green light to use it. It won't hold our systems back. I think we tend to ignore other, more important aspects of ownership because there’s no good way of getting reliable information about it. I.e., looking for our lost car keys under the lamp post not because that’s where we dropped it, but because that’s where the light is.

 

[amirm]

You can't get more transparent than transparent. Engineering for engineering's sake is fine, so long as that's what their customers want. But if we care about high fidelity this gets the job done as well as anything Topping ever did.

Are you just running with a talking point or what I measured in the review? I clearly showed that USB noise bleeds into the analog output of the DAC. Even if you are not using the DAC itself! Such noise is frequently seen by users as they hear it when there is no music playing, or during quiet parts. I have never seen this in any Topping product. Or SMSL. Or plenty of other competitors to this product.

I also showed distortion in IMD at low level playback. Crank up the volume on some softly recorded music and you may also hear that as well.

Finally, the 50 mv test shows that the whole unit is not quiet enough for very sensitive IEMs.

Now, if this was a much cheaper product that would be something. Or from the company that takes review feedback seriously, we could hang our hat on that. But neither is true. The only benefit you get, if you are in US, is buying local. And for that, you get an inferior product with certain potential for noise audibility.

But transparency is a bar, and once you meet it, there's little value in going further.

Transparency per peer reviewed research requires performance well above what the Magni produces. I am talking about real psychoacoustic analysis of such things as noise relative to payback level. Not hand waiving that everything is transparent.

2. I've read complaints over the years about Schitt pots being prone to getting scratchy or having channel imbalance as they age. But we don't evaluate on that basis, because it's purely anecdotal. It's bad data, garbage-in-garbage-out and all that.

I have found Schiit products which readily produce hum due to improper grounding. I have documented how to fix it. See this: https://www.audiosciencereview.com/...otunheim-headphone-amplifier.3733/#post-89406

You would hand waive this as also being transparent, yes?

Yet, it is an absolutely audible problem fixed by the procedure I outlined in that link. Schiit of course denied there is such a problem in their manufacturing but quietly fixed it. At least I think they did.

Bottom line is that Schiit has a loose relationship with engineering excellence. They came kicking and screaming into our domain seeing the impact we have in their customer base. For a bit, they were doing well, paying so much attention to measurements that they were volunteering to send me new products for review. I thought all was well so didn't seek out to test their stuff anymore. Seems they took advantage of this and stopped caring as much as they should.

So for heaven's sake, don't give them permission to do much worse than they could be doing. It comes at no expense to you to get Topping class performance from them. And that is what they should produce, or admit they don't have the engineering ability or the will to do it.

 

[Joe Smith]

I feel like half of the problem at Schiit is the pace/number of new product releases. Better to focus down the product line and get everything as good as it can be, for the price point.

 

[aagstn]

I feel like half of the problem at Schiit is the pace/number of new product releases. Better to focus down the product line and get everything as good as it can be, for the price point.

The previous Magni and Modi were around for several years and both were very good. The Heretic was exceptional. This could be price pressure with all the increases in cost after Covid and Tariffs. It has to be tough to make a combo amp and Dac for $200 in the US. They went with good enough performance to hit that price point seems the most likely. Most people wouldn't notice the noise and it would depend on the system that USB was connected to.

All of these companies have their issues making budget gear. Yes, Topping and SMSL make DACs and Amps that measure well. However, one thing that is never covered in the reviews is all the bugs and issues these devices can have. I had popping and clicking issues, buggy performance, and weird shut downs with all the SMSL and Topping products I owned. I returned my Wiim because the left and right channels leaked audio into each other. Some people had speakers destroyed with bugs in devices like the SMSL MDA 1 that was well reviewed for objective performance.There are compromises all of these companies make to hit a price point. I agree Schitt could be doing better and hopefully they will take this feedback seriously and fix issues like poor grounding.

 

[solderdude]

The Heretic was merely created to 'show' how easy it is to make a well measuring cheap little amplifier using a few simple op-amps.
It was not named Heretic for nothing as it strayed from their design philosophy.
They had no intention to go further on its path as that is not what Schiit fans are looking for.
They want the stories, the myths, the subjective reviews.

Their aim is to create gear with good sound, preferably with as few as possible discrete components in the signal path.
Sometimes they use op-amps but rarely in the signal path unless they can't get around it.
This is what the Magni and more expensive models are (their continuity design).
And ... for just a handful of discrete parts it is doing rather well.
Could that easily be improved a bit measurement wise ? sure ... Is that needed ? probably not.

The same with their DACs... they prefer R2R but to go cheap they must use the usual parts, they want their 'promised to be better sounding' megacomboburritofilter.
They want the 'better sounding' own USB receiver as they believe USB sounds poor (that was the understanding in those circles back then and even now).
And it is designed and assembled in the USA (OK.. OK. they also have to use parts from the Orient)

Schiit are doing well with only their fan-base which loves the stories (Jason is also a gifted audio writer)
They are not in the SINAD race camp, they just thought it was funny to show how easy it was to go down that route.

Do they make mistakes now and then, do parts like volpots and switches go bad and/or components break ? Sure but this is equally true for other brands.

As long as their customers keep buying the Schiit they are happy and can keep the chimney smoking.

disclaimer: To make things clear ... I am not a Schiit fanboy nor have I ever owned nor tried any of their gear and I don't have to either to form an opinion.

 

[AndreaT]

@solderdude's response to this is more definitive, but here's my "yes and..." Forgive me while I repeat myself, both in this thread and more broadly across multiple threads.

The most direct response to your question is that even if they got the same results from their AP that Amir did, those results show them that the product is ready to push out the door.

1. You can't get more transparent than transparent. Engineering for engineering's sake is fine, so long as that's what their customers want. But if we care about high fidelity this gets the job done as well as anything Topping ever did. Why add to the price? This is literally the cheapest product in their range. I defend ASR from outsiders when they say we cargo cult the measurements, but the flip side of that is that I need to bring it up when we start to cargo cult the measurements. Yes, I believe measurements capture the whole story from a sonic standpoint, and evaluation based on measurements is a thousand times better than relying on Jason Victor Serinus' or some rando from Head-Fi's subjective opinion on something. But transparency is a bar, and once you meet it, there's little value in going further.

2. I've read complaints over the years about Schitt pots being prone to getting scratchy or having channel imbalance as they age. But we don't evaluate on that basis, because it's purely anecdotal. It's bad data, garbage-in-garbage-out and all that.

And yet if there is any semblance of truth to this, it's 1,000x more damning than any of the measurements we see here. Which is to say, these reviews should not be considered to be comprehensive as to whether the device is any good. The measurements are comprehensive, but that doesn't tell the whole story. From a sonic standpoint, it tells us that we have the green light to use it. It won't hold our systems back. I think we tend to ignore other, more important aspects of ownership because there’s no good way of getting reliable information about it. I.e., looking for our lost car keys under the lamp post not because that’s where we dropped it, but because that’s where the light is.

Obviously your transparency is fog to me. Why should I choose a lowest bar engineering product when, for a similar cost to me, I can have SOTA?

 

[Sokel]

Why should I choose a lowest bar engineering product when, for a similar cost to me, I can have SOTA?

Aside from this thing, engineering is not only about THD+N.

In my book, is first safety and conformity to regulations, reliability, repair- ability (along with schematics and all) , of course decent performance, efficiency, intuitive UI, premium feeling (yes, that's part of engineering as well) and many more.

And as we talk about the lowest budget gear and the frantic rhythm they ditch gear in the market , iron law of production is applied as well:
Cheap - fast - safe , pick two.

 

[VientoB]

Aside from this thing, engineering is not only about THD+N.

In my book, is first safety and conformity to regulations, reliability, repair- ability (along with schematics and all) , of course decent performance, efficiency, intuitive UI, premium feeling (yes, that's part of engineering as well) and many more.

And as we talk about the lowest budget gear and the frantic rhythm they ditch gear in the market , iron law of production is applied as well:
Cheap - fast - safe , pick two.

Measurements are important here.

 

[artismo]

FYI this Magni MESH DAC card still uses ESS dac chip, so it is NOT an all new in-house developed DAC.
They describe it as a "DAC topology that fuses our unique time- and frequency-domain optimized digital filter with a standard delta-sigma modulator".

View attachment 514177

My old man eyes spot NE5532 and OPA1656? I wonder if the NE is handling the reference buffer? TBF it's cramped in there but I wouldn't have put the daughterboard near the USB C port and probably went with an OPA1612 for higher CMRR but it's picky about circuit layouts and traces to prevent oscillation. Bet there's current feedback noise from that connector too. Looks like there might be some ground bounce based on the jitter. Well, NFJ does something similar with its engineering on some devices unfortunately, it leaves the problem solving to the end user and caters to the tinkerers.

 

[Matias]

To me no excuse of cramped space of low cost justifies when a JM20 dongle measures better in every test while being way smaller and a fraction of the price.

 

[artismo]

To me no excuse of cramped space of low cost justifies when a JM20 dongle measures better in every test while being way smaller and a fraction of the price.

Fair point on the JM20 - it’s an incredible feat of modern SoC integration. I think comparing a 30mW dongle to a 3W desktop amp has its challenges.

The hurdle here isn't just making a clean DAC; it’s making a clean DAC that shares a ground plane and a cramped modular header with a high-current, ±17V discrete power stage. That heavy lifting is exactly what creates the ground bounce and jitter skirts we see in the measurements.

A dongle measures better because it lives in a low-voltage, low-current vacuum. Once you introduce the power requirements of a desktop amp into that same small chassis, layout topology becomes the real bottleneck. That’s why I’m a fan of modularity/separates - it lets the DAC be a DAC in its own quiet neighborhood (like the JM20) while letting the amp handle the current elsewhere. The Magni Mesh tried to have its cake and eat it too, and the layout just couldn't support both perfectly.

Using a JM20 as a standalone DAC to feed the Magni's RCA inputs would likely measure better because you're getting physical galvanic isolation - the DAC's sensitive clock and reference buffer aren't sharing a ground plane with a 3W power stage. But I'm probably preaching to the choir - ramblings of an old man burning the midnight oil.

It's tough too because even though the amp is SET, using a 3.5mm before the amplification already introduces a common ground and would possibly kill separation.

 

[ButHowDoesItSound?]

Their PCBs are red, so they sound warmer.

 

[artismo]

Their PCBs are red, so they sound warmer.

 

[solderdude]

Schiit buyers usually don't care about SINAD as much and believe in and want the 'Unison 384 USB' part, the 'Mega Combo Burrito filter' and 'high-precision current-feedback discrete topology ' which is an exotic, bespoke amplifier stage unlike anything available from the competition and is an American made product as that is more important to those buyers than the latest and greatest SINAD beating amp made in the far East.

ASR buyers might care more about other aspects.