Review and Measurements of Schiit Modi 3 DAC

[amirm]

Putting aside the unprofessional language and focusing on technical detail, I don't know what he is saying about the power supply. Here is the Modi 3 with USB power and no external power supply:

As you see, the output remains the same at essentially 2 volts. No external power supply is needed which makes sense given the fact that USB provides 5 volts.

 

[garbulky]

You were right so I did that yesterday after getting their answer:

View attachment 16208

At least they are confirming that there are no ASIO drivers.

I think Laura Z misunderstood your question. You are asking what ASIO drivers they used when they published their AP test. It looks like she is thinking that you are asking for ASIO drivers and telling you that the device doesn't need drivers with Win 10.

 

[garbulky]

Also what is this PTSD comments? What is this referred to?

 

[amirm]

Amir, please take a look at my post #130. There are two other ASIO wrappers freely available: ASIO2WASAPI and ASIO2KS. Maybe you can try to use them with your AP, either with Microsoft's or Schiit's USB driver. Just in case this is an ASIO4ALL issue.

Thanks. It is resolved now. With Schiit drivers ASIO4ALL works correctly.

 

[derp1n]

Putting aside the unprofessional language and focusing on technical detail, I don't know what he is saying about the power supply. Here is the Modi 3 with USB power and no external power supply:

View attachment 16210

As you see, the output remains the same at essentially 2 volts. No external power supply is needed which makes sense given the fact that USB provides 5 volts.

I suspect he's trying to address the cognitive dissonance created by the fact that Schiit heavily push linear PSUs as a feature in their gear, yet here they are with their best performing product and no LPS in sight.

I think they create a lot of problems for themselves using strong arrogantly worded attacks on various design approaches, then end up finding themselves unable to use them later when it's obvious the design is superior or the trade offs are necessary. An example being the Vidar which would've been better in class D in that packaging. Or pushing Class A heavily and then trying to back out of it with "Continuity" in Jotunheim.
You can see the same attempt at walking back the LPS doctrine around the Fulla 2 forum post marketing too.

 

[gvl]

Yes, she writes "supports our products with no additional driver" and really means "works only in 16 bits". And with "optimum performance" she really means "works also in 24 bits". They should use that decoded wording on their website and in the manual.

It seems to be an ASIO4ALL peculiarity. Roon over WASAPI could do 24 bits using the Windows stock driver without issues.

 

[garbulky]

It means posttraumatic stress disorder (I asked the same question). But I still don't know what that has got to do with Amir.

Lol I know what PTSD means. I just didn't know what they were referring to as I don't see the original post.

 

[gvl]

He is downplaying the issues with their TOTL DAC that Amir found.

 

[garbulky]

He is downplaying the issues with their TOTL DAC that Amir found.

who?

 

[gvl]

who?

Mike of Schitt Audio a.k.a. Baldr.

 

[JJB70]

Lol I know what PTSD means. I just didn't know what they were referring to as I don't see the original post.

It is just an abusive comment, an ad hominem attack which demeans Amir, demeans those who suffer from PTSD and brings nothing but discredit on Schiit.

 

[JJB70]

I suspect he's trying to address the cognitive dissonance created by the fact that Schiit heavily push linear PSUs as a feature in their gear, yet here they are with their best performing product and no LPS in sight.

I think they create a lot of problems for themselves using strong arrogantly worded attacks on various design approaches, then end up finding themselves unable to use them later when it's obvious the design is superior or the trade offs are necessary. An example being the Vidar which would've been better in class D in that packaging. Or pushing Class A heavily and then trying to back out of it with "Continuity" in Jotunheim.
You can see the same attempt at walking back the LPS doctrine around the Fulla 2 forum post marketing too.

I agree, they are not the only audio equipment manufacturer (or indeed the only manufacturer of any sort) to find themselves going back on previously held positions and I think that it is entirely sensible in that if you find you were wrong or technology changes then the only sensible thing to do is to amend your position. However because of the way they've sold some of their opinions and derided the alternatives and convinced a lot of fanboys and shills they seem to be regularly getting themselves into this awkward position. Couldn't happen to nicer people!

 

[solderdude]

But .... I got the impression only Schiit DAC's have the famous 'Moffat-bass'.... so Mike seems essential for the marketting hype and ensuring the 'Moffat bass' is inserted in each device manually.

 

[March Audio]

Do you use the latest ASIO4ALL version 2.14? This is an updated version that includes a "Workaround for a bug in Windows 10 Creators Update: USB audio capture would not work any more (red exclamation mark, distorted sound, ...)". Maybe this also applies for USB audio render.

Besides, could you show a picture of your ASIO4ALL configuration like this one? Does the "WDM Device List" state "24Bits" for "Out"?

If I am not mistaken you did not install any device specific WDM driver. So, which driver does Windows really use? Its own internal Microsoft/Thesycon USB 2.0 driver (usbaudio2.sys)? Or the correct device specific Schiit driver (which seems to be this one) that Windows may have installed from Windows Update? Or a previously installed (and now incorrect) driver? I think there is a chance that ASIO4ALL on your system somehow does not use the driver that WASAPI uses.

It may also be that you have unveiled a yet undiscovered bug in ASIO4ALL. In that case the developer Michael Tippach would certainly help.

Also, in the introduction of Schiit's test report they write "If you still have significantly different results, please contact [email protected] with a copy of your results so we can bring back your product and check it against our standard." It would be interesting to know which ASIO driver they actually used. I think they would tell you if you showed them your results with the same 28k$ audio analyzer.

Overall I think it is worth seeking for the reason of this discrepancy (and finding a solution for it) as this device seems to be the first seriously engineered DAC from Schiit .

By the way, the 100 Hz spikes and its overtones are certainly the result of truncation from 24 to 16 bits (as you already mentioned). This is the spectrum of an undithered 24-bit full scale 1 kHz sine:
View attachment 16066
Here is the result of converting that signal to 16 bits with simple truncation, that is, no dither and no noise shaping:
View attachment 16067
This looks like the spectrum you recorded with your APx.

Got to ask a question, the first graph is indicating at lower frequencies a noise floor of around -190dB. Can you talk about what equipment was used to obtain the graph? The process gain of a 131k FFT seems unlikely to get the floor down that far, so I am curious what measured it and what the Schitt is doing to be able to achieve that outstanding (read unbelievable) figure.

 

[restorer-john]

Putting aside the unprofessional language and focusing on technical detail, I don't know what he is saying about the power supply. Here is the Modi 3 with USB power and no external power supply:

I would say he means something like this:

To get 2V output in a cheap USB (5V) powered D/A converter, you would need a switcher or buck converter to step up the rails internally unless you used a dual amplifier/buffer chip running in bridge tied load/balanced off a single 5V rail.

2V output is 2.83V peak and needs an available swing (PP) of 5.66V even if there were zero junction losses. Junction losses are normally around 1.6 to 2.0V minimum on typical split rail opamps/buffers. Even a so-called rail-to-rail on a single 5V supply couldn't give you more than around 1.7V. I think they used a AD8616 before, which is a single supply, rail to rail opamp on 5V and got ~1.5V RMS out and you'd need a coupling capacitor to block DC (half of 5V)

As he wants to eliminate capacitors to the load, split rail (+/-) supplies are required and therefore, switching internal supplies must be implemented.

This is how I would interpret the technical part of what he said. His other comments are simply out of order.

 

[Wombat]

I agree, they are not the only audio equipment manufacturer (or indeed the only manufacturer of any sort) to find themselves going back on previously held positions and I think that it is entirely sensible in that if you find you were wrong or technology changes then the only sensible thing to do is to amend your position. However because of the way they've sold some of their opinions and derided the alternatives and convinced a lot of fanboys and shills they seem to be regularly getting themselves into this awkward position. Couldn't happen to nicer people!

Schiit have previously strongly put forward their design creds to rebuff criticism. Back-pedalling is contradictory.

 

[March Audio]

I did not use any hardware equipment. I used iZotope's RX 4 and its internal spectrum analyzer. The idea with those two graphs was to just confirm Amir's findings that his 24-bit test file was indeed truncated to 16 bits. The resulting spectrum of this conversion is very similar to what he recorded with the APx and thus a good proof.
I used RX's internal Dither module to convert the original 24-bit file to 16 bits (with no dither and no noise shaping, that is, simple truncation).
Sorry if those graphs may be confusing, that was certainly not my intention.

Aaahh I see, my misunderstanding, thought it was too good to be true

 

[Thomas savage]

Because amir is a meaningless distraction we are plastering AP graphs all over our ad copy and shilling our schiit using ‘measurements’ , we are busy redesigning our products in order to do this...

Amir , meaninglessly keeping schiit peddlers honest and all totally free of charge.

I wonder if they will pop by and drop us a donation .

 

[Wombat]

Because amir is a meaningless distraction we are plastering AP graphs all over our ad copy and shilling our schiit using ‘measurements’ , we are busy redesigning our products in order to do this...

Amir , meaninglessly keeping schiit peddlers honest and all totally free of charge.

I wonder if they will pop by and drop us a donation .

I think you meant 'Poop by".

 

[garbulky]

Mike of Schitt Audio a.k.a. Baldr.

I must have missed all this. Is this a recent thing he said? What did he say?

I would say he means something like this:

To get 2V output in a cheap USB (5V) powered D/A converter, you would need a switcher or buck converter to step up the rails internally unless you used a dual amplifier/buffer chip running in bridge tied load/balanced off a single 5V rail.

2V output is 2.83V peak and needs an available swing (PP) of 5.66V even if there were zero junction losses. Junction losses are normally around 1.6 to 2.0V minimum on typical split rail opamps/buffers. Even a so-called rail-to-rail on a single 5V supply couldn't give you more than around 1.7V. I think they used a AD8616 before, which is a single supply, rail to rail opamp on 5V and got ~1.5V RMS out and you'd need a coupling capacitor to block DC (half of 5V)

As he wants to eliminate capacitors to the load, split rail (+/-) supplies are required and therefore, switching internal supplies must be implemented.

This is how I would interpret the technical part of what he said. His other comments are simply out of order.

Ah that's a post from SBAF. I thought Baldr's post was either here (by some miracle) or on his head-fi thread.
I'm not sure what he means by assy. Bassy? Anyway, I too don't worry about things below 100. I don't worry about them below the 80's as I really can't hear down that much. But his word usage is wrong about the PTSD. PTSD is where people suffer from trauma and then deal with the resulting mental consequences of it. Obsessional thoughts, an actual disorder that would be closer, more correct usage. Though that still would be making light of an actual disorder which I hope nobody reading is suffering from. Obsessional thoughts is what the name implies but it also creates in the person great anxiety when thinking about the recurring thoughts. Interestingly obsessional thoughts, OCD and PTSD are linked. They are all anxiety disorders.