Measurement and Review of Schiit BiFrost Multibit DAC

[MingChops]

Sure, if you ask them to really listen for it, they'll hear it. Psychoacoustics.

I would tend to agree. But every test prior I have done using multiple DS DACs/streamers, comparing different digital connections (coax/optical), different analogue connections (RCA/XLR) from the same kit etc - always the same, I have never found there to be anything remotely like a tangible A/B difference at all.
However - this Modius to BF2 test, is the only test that's ever presented something that's identifiable. And in conjunction with the REW sweep difference, is the only reason I bothered to post back on this particular time.

Maybe I should have cut the post down to a few words, said: "I think I can hear a difference, I've done this REW sweep - could it explain it? The filtering is meant to be better - could that be why?"

I was just rather excited that for the first time ever, I've noticed a difference in something DAC related!!

 

[BDWoody]

Look I've tried to be as objective as I can, and I laid out already that I understand it falls short of objective testing - so chalk it down to whatever you want

Right... When doing subjective tests, controls are needed. Without them, we aren't able to get much meaning.

It isn't convenient, that's for sure... But unfortunately, controls matter.

The filters can certainly impact sound for some, and that could explain something... And I'm not trying to be difficult just for fun.

You would measure the output of the DAC's, not the amp, and if you can mix up the cables and switch them without knowing what is on the other end, you've got a couple of minimal controls that could help. Is it perfectly valid and ready for peer review? No... But it will tell you more than what you're doing now.

 

[MingChops]

Right... When doing subjective tests, controls are needed. Without them, we aren't able to get much meaning.

It isn't convenient, that's for sure... But unfortunately, controls matter.

I've done the best I can do with the limited resources (and I won't deny knowledge too) I have, and have captured repeatable data. I think that qualifies this has objective (at the very least not the subjectivist tripe, I'm really trying to distance myself from that - overly enthusiastic first post aside!). Granted I am looking for an objective means to try and rationalise a subjective feeling, however to move forwards I think let's go with the notion for now that I hear nothing on an A/B test. And therefore for now there is no need for a blind test at all.

Rather lets instead focus on the fact there is a REW sweep showing data/differences, and why would this be when DAC's have a frequency responses as good as ruler flat, when given sine waves.

The filters can certainly impact sound for some, and that could explain something... And I'm not trying to be difficult just for fun.

You would measure the output of the DAC's, not the amp, and if you can mix up the cables and switch them without knowing what is on the other end, you've got a couple of minimal controls that could help. Is it perfectly valid and ready for peer review? No... But it will tell you more than what you're doing now.

Ok but afraid I will still not putting probes up my equipment. Also you still have not given a method for how I could actually control the voltage, and would that control itself impart any problems?
But as said above for now lay aside the voltage matched test, and put the emphasis on filtering for a moment. If it can have a repeatable effect on the sweeps - does it stand to reason that change is audible? Does it also imply the more dynamic nature of music compared to a sine waves, could be subject to more of this change than perhaps given credit for?

 

[Dana reed]

I've done the best I can do with the limited resources (and I won't deny knowledge too) I have, and have captured repeatable data. I think that qualifies this has objective (at the very least not the subjectivist tripe, I'm really trying to distance myself from that - overly enthusiastic first post aside!). Granted I am looking for an objective means to try and rationalise a subjective feeling, however to move forwards I think let's go with the notion for now that I hear nothing on an A/B test. And therefore for now there is no need for a blind test at all.

Rather lets instead focus on the fact there is a REW sweep showing data/differences, and why would this be when DAC's have a frequency responses as good as ruler flat, when given sine waves.



Ok but afraid I will still not putting probes up my equipment. Also you still have not given a method for how I could actually control the voltage, and would that control itself impart any problems?
But as said above for now lay aside the voltage matched test, and put the emphasis on filtering for a moment. If it can have a repeatable effect on the sweeps - does it stand to reason that change is audible? Does it also imply the more dynamic nature of music compared to a sine waves, could be subject to more of this change than perhaps given credit for?

What I'd likely do to level match without having an oscilloscope would be to set a 60 Hz test tone to the DAC, and measure the output as VAC using a voltmeter on the output of a connected headphone amp. Whichever one has a higher level, I'd attenuate to match either using digital attenuation or an analog pot. If you're using Roon, you can set up headroom adjustment on whichever DAC is higher output, then you could use something like a Schiit Sys as an A/B switcher, while Roon outputs the same songs to both DACs at the same time. Then you can have headphones on and just A/B it without disconnecting anything. The only not-blind part if you live alone is that you would know which DAC was on which input.

 

[MingChops]

What I'd likely do to level match without having an oscilloscope would be to set a 60 Hz test tone to the DAC, and measure the output as VAC using a voltmeter on the output of a connected headphone amp. Whichever one has a higher level, I'd attenuate to match either using digital attenuation or an analog pot. If you're using Roon, you can set up headroom adjustment on whichever DAC is higher output, then you could use something like a Schiit Sys as an A/B switcher, while Roon outputs the same songs to both DACs at the same time. Then you can have headphones on and just A/B it without disconnecting anything. The only not-blind part if you live alone is that you would know which DAC was on which input.

I don't have headphones, a headphone amp, or Roon. But otherwise sure!
The equipment used, Node 2i streamer - all of it's outputs are active, so the same stream is spit out to both the Coax and Optical SPIDF's. I took one to Modius, one to BF2 - and both in turn to my speaker amp. And I can toggle the respective inputs for A/B.
Is it perfectly level matched, of course not, is it in terms of the end result all but identical and closer than the threshold of hearing, yes I am satisfied it is. Just look at the graph, it's mirrored for 99% of it.

So I think we're still detracting back to this incessant need for level matching at the voltage level. I can't do that, but what I can do is a practical test using a calibrated mic in a controlled position, and to that end the levels are indeed functionally the same.

This test has turned up what I believe to be unexpected results in that there are differences in just a few subtle places with respect to the frequency response. It is a variance large enough in size that we can easily ignore the fact the voltage might not be spot on between them, because the variance in those places is clearly greater and would still exist if the levels were matched.
And that is the question at large here, nothing else. Voltage matched blind test is not going to change that, there is a fundamental variance in response here that doesn't seem to fit with the notion that otherwise ruler flat DACs produce the same output.

I believe (as a layman) the explanation could be filtering related with the way it's handling and interpolating the data. I.e it seams to have retained a bit more signal than the Modius has done. And if 2 devices that handle a sine wave the same way, but not music, does this add any credibility to the subjective claims of others?

Certainly the end by-product of these 2 DACs has resulted in a subtle change in audible frequency response, and I think that implies I am indeed hearing the 2 DACs differently. I've presented relevant data (a level matched A/B test wouldn't actually produce any data for that matter, just give me the mother of all golden ear bragging rights no! ), jokes aside I'm not just throwing audiophile terms about here with zero evidence to substantiate. I'm also here to learn so if my assumptions are wrong please correct.

 

[Dana reed]

I don't have headphones, a headphone amp, or Roon. But otherwise sure!
The equipment used, Node 2i streamer - all of it's outputs are active, so the same stream is spit out to both the Coax and Optical SPIDF's. I took one to Modius, one to BF2 - and both in turn to my speaker amp. And I can toggle the respective inputs for A/B.
Is it perfectly level matched, of course not, is it in terms of the end result all but identical and closer than the threshold of hearing, yes I am satisfied it is. Just look at the graph, it's mirrored for 99% of it.

So I think we're still detracting back to this incessant need for level matching at the voltage level. I can't do that, but what I can do is a practical test using a calibrated mic in a controlled position, and to that end the levels are indeed functionally the same.

This test has turned up what I believe to be unexpected results in that there are differences in just a few subtle places with respect to the frequency response. It is a variance large enough in size that we can easily ignore the fact the voltage might not be spot on between them, because the variance in those places is clearly greater and would still exist if the levels were matched.
And that is the question at large here, nothing else. Voltage matched blind test is not going to change that, there is a fundamental variance in response here that doesn't seem to fit with the notion that otherwise ruler flat DACs produce the same output.

I believe (as a layman) the explanation could be filtering related with the way it's handling and interpolating the data. I.e it seams to have retained a bit more signal than the Modius has done. And if 2 devices that handle a sine wave the same way, but not music, does this add any credibility to the subjective claims of others?

Certainly the end by-product of these 2 DACs has resulted in a subtle change in audible frequency response, and I think that implies I am indeed hearing the 2 DACs differently. I've presented relevant data (a level matched A/B test wouldn't actually produce any data for that matter, just give me the mother of all golden ear bragging rights no! ), jokes aside I'm not just throwing audiophile terms about here with zero evidence to substantiate. I'm also here to learn so if my assumptions are wrong please correct.

I’m just pointing out one way to do it. I thought my Gungnir Multibit and Yggdrasil had obviously different sound, until I attenuated the Yggdrasil to match the Gungnir output. Then the differences were something I’d not be able to pick out reliably. In the AP reports on the schiit site, the output levels are different between those two, and also between modius and bifrost.

 

[MingChops]

I’m just pointing out one way to do it. I thought my Gungnir Multibit and Yggdrasil had obviously different sound, until I attenuated the Yggdrasil to match the Gungnir output. Then the differences were something I’d not be able to pick out reliably. In the AP reports on the schiit site, the output levels are different between those two, and also between modius and bifrost.

Yeah fair enough, I'm of the current opinion that the difference I have captured is likely down to the vastly different architectures. Gumby to Yggy of course share the same fundamental topology and filter design, so I would hypothesise they are likely to conform more than than they are deviate, much the same as I have found when comparing all D/S product's I have - nothing in it.