[JTF]

This is more of a quick take than a detailed review, so I apologize if I'm posting in the wrong circle. I wanted to share something I found interesting. I decided to try some of Schiit's DAC offerings. The new Modius, which is a delta sigma dac with the AKM 4493 chip, and the updated Bifrost 2, a multibit dac. Both have multiple inputs and balanced outputs, the latter a feature I sought out specifically.

I spent a few days with each dac and noticed some differences in the sound, so I decided to measure the response of my system with each dac, just for fun. As for the dacs, I found the Modius to be really good, especially for the price, but I think I'm going to return it. The Bifrost is smoother and has more body, which I prefer. I've only spent a few days with each at this point. I'll update this post or start a new one with a full review after I've spent more time.

As for the measurements, you can see below that they are different above 2k. The red line is the Modius, the blue is the Bifrost.

I took measurements using REW, at the listening position. I repeated it multiple times to make sure the results were the same. The only difference was the dac in use. My room is 13 x 22 and the chain for this was mbp > Schiit DAC > Jotunheim as pre > Cherry monoblocks > Spatial X5s. The Spatials are ~4.5ft from the front wall and the listening position is 10ft from the speakers. The measurements should be taken with a few grains as I'm a REW noob.




Cheers

[poseidonsvoice]

And the fact that your vertical range is 130dB and 1/6th octave smoothing on the horizontal which makes the graph look overly smooth. I wouldn’t use in room measurements generally to illustrate differences between dacs, as there are too many variables. Sometimes small disturbances in how room reflections are captured by the mic can make things look worse than they are. But your measurements do make me pause a little. If you changed the vertical range to just 50dB (ie 40-90dB) the differences between the dacs would look more alarming not less (in the area above 2khz). *The other thing is if you measured BOTH speakers at the same time then your measurement is distorted due to comb filtering issues between each speaker, and between each speaker and the respective boundaries that particular speaker is surrounded by*.

Unless there is something functionally wrong with one of the dacs there shouldn’t be an FR difference between the two whatsoever (especially in that frequency range).

Schiit has been known to not have the best QC, and units have been sent back only to find out it was defective in the first place. See the Schiit Freya review on ASR to see an example. I am not making this up...Once fixed, the unit performed as spec’d.

Actually, if I were you, I would ship both back (or at least email/call them) and ask them if they truly do have identical or nearly identical FR curves. There shouldn’t even be a 1dB difference in the range of 20-20khz. They should have measurements of your particular dac units on file. They do have an Audio Precision analyzer.

If they do have identical FR measurements per Schiit then I would think small disturbances and comb filtering issues are the reason for the differences in the in room measurements particularly above 1khz. The way to mitigate that is to measure each speaker individually and then compare the dacs. Like Modius/RCH, Modius/LCH, compared to Bifrost/RCH and Bifrost/LCH.

Best,
Anand.

[JTF]

Thanks for your input Anand. Yes, it appears that measuring both speakers at the same time was the problem. I measured again just now, L and R separately. I measured each multiple times and threw out the duplicates.

All together



Right channel, Bifrost orange, Modius blue




Left channel, Bifrost red, Modius green

[poseidonsvoice]

Presto!

Awesome. Glad I could help.

Best,
Anand.

[JTF]

So... Maybe someone can explain why, when measuring both channels together in room, there's a noticeable difference between dacs? Clearly there's not much difference when looking at just one channel, either L or R. I also took those initial measurements a few times to make sure the result was repeatable, and it was.

[charmerci]

So... Maybe someone can explain why, when measuring both channels together in room, there's a noticeable difference between dacs?

Read Anand's last paragraph in reply #1.

[poseidonsvoice]

Read Anand’s last paragraph in reply #1.

With regards to the OP's query, I think it's important to understand that what a microphone 'sees' is a conglomerate of not only both speakers but associated reflections as well. There are a multitude of them (starting from the baffle to every boundary in the room associated with that speaker in that room). Also understand that when you do 1/6th octave smoothing, that isn't a perfect method, it's an approximation. Without it, you just see tons of reflections and one cannot discern with the human eye which ones are of importance.

When frequencies arrive at the microphone in phase, they add up and you get a slight boost in that area of the response curve. Consequently, if they are in opposite phase, they subtract. This is a simplification. In addition, there are  all sorts of phase differences with additional frequency response additions/subtractions in between that can cause the aberrations that you see.

Now to the question as to why two different dacs can cause this, (and have **in room response curves** that look different) as long as you understand that this isn't a DAC anomaly but an acoustical anomaly (because of user error), one reason is because of timing differences. The right speaker’s response can arrive a little sooner or later than the left, and the microphone discerns that very differently than the human ear. It discerns it in a very mathematical, 'literal' way if you will, while the human ear which lies on either side of a head and spread 6 inches apart discerns it differently (and better I think!). Look into binaural recording techniques and the like, you will learn more.

Assuming that the wave of frequencies projects from each speaker *at the same time* is just that...an assumption. The microphone disagrees, but that's okay. It wasn't meant to evaluate STEREO at least at frequencies greater than 300Hz. The microphone as a tool needs to be used properly.

I bet if you get 3 more dacs and run REW again with both speakers playing, you might see even more interesting aberrations. But if you then measure each speaker separately as you have done, they will be very similar or the same.

In any case, I suggest you trust your ears and keep the dac you like more.

Best,
Anand.

[charmerci]

I wrote it on my phone and it autocorrected... Changed! 

[JTF]

I did some googling yesterday and figured it out, but that's what I was getting at with my follow up question. Thank you for the more detailed reply Anand.

[Mike-48]

Thanks for your input Anand. Yes, it appears that measuring both speakers at the same time was the problem. I measured again just now, L and R separately. I measured each multiple times and threw out the duplicates.

On my screen, I can't make out the vertical scale. What is the increment between horizontal reference lines?

[poseidonsvoice]

On my screen, I can't make out the vertical scale. What is the increment between horizontal reference lines?

10 dB on the 1st graph. Goes from 0 to 130dB.

Subsequent graphs span 40dB (57dB to 97dB) in 1dB increments.

1/6th (?) octave smoothing over the horizontal axis.

Best,
Anand.

[JTF]

10 dB on the 1st graph. Goes from 0 to 130dB.

Subsequent graphs span 40dB (57dB to 97dB) in 1dB increments.

1/6th octave smoothing over the horizontal axis.

Best,
Anand.

All the graphs have 1/3 smoothing applied.

[poseidonsvoice]

Thanks. Very difficult to tell on my mobile device!

Best,
Anand