ifi iPurifier3 USB Filter Review

[Lambda]

Yes, it is. But if we want to have this properly tested, we need a proper test protocol, and that's what this should be about.

But. As i said there are already many common tests for this.

and if you have them, they are more annoying than any SINAD issue a DAC might have.

It’s not an "if" Question.
It is inevitable.

But with proper Design on the DAC side this Common mode Currents can be minimized.
But with proper Design on the AMP side also.
And what many ASR users don't want to belie. even the cables can make a difference.

And since Good DACs have no audible difference in in good conditions (almost no common mode currents/noise and a high end receiver with good CMRR) and can barely be differentiated with the AP let alone by ear...
it makes no sens to test under this "ideal" conditions.


There are dozens of cheap DACs that are audible perfect under "ideal" conditions.
It matters more waht DACs are audible perfect under nonideal conditions.

 

[tmtomh]

But. As i said there are already many common tests for this.


It’s not an "if" Question.
It is inevitable.

But with proper Design on the DAC side this Common mode Currents can be minimized.
But with proper Design on the AMP side also.
And what many ASR users don't want to belie. even the cables can make a difference.

And since Good DACs have no audible difference in in good conditions (almost no common mode currents/noise and a high end receiver with good CMRR) and can barely be differentiated with the AP let alone by ear...
it makes no sens to test under this "ideal" conditions.


There are dozens of cheap DACs that are audible perfect under "ideal" conditions.
It matters more waht DACs are audible perfect under nonideal conditions.

It's not inevitable that you will have a ground loop. And I would think it would be impossible, or at least impractical, to test gear's immunity to passing along ground-loop noise because the test condition cannot replicate or anticipate the range of variability in real-world conditions. And since a ground loop is not simply a product of the DAC or any other one piece of gear, as it is by definition a product of the interaction of two or more interconnected pieces of gear, I don't know how you'd even use the dB level of ground-loop noise coming out of the speakers as a measure of any one piece of gear.

 

[pkane]

Everyone makes fun about PS audio and there Power optimizes...
But Topping selling USB isolators as add-on instead of integrating them in DACs.

Topping HS02 = $99
PS Audio PowerPlant 3 (cheapest one) = $2999

Interesting comparison to make, considering that power plant 3 passes through most of the line noise/distortion in the audio band, just attenuated by about 10dB, while HS02 actually removes it. In case it matters at all.

 

[Lambda]

It's not inevitable that you will have a ground loop.

And how can it completely avoided? you will always have some stray capacitance even with battery powered devices.

And I would think it would be impossible, or at least impractical, to test gear's immunity to passing along ground-loop noise because the test condition cannot replicate or anticipate the range of variability in real-world conditions.

Whats why we don’t test with "real world" conditions but with synthetic tests.
DACs don’t get tested with music but with special test tones.
The same way you can measure CM impedance or rejection ration.

And since a ground loop is not simply a product of the DAC or any other one piece of gear, as it is by definition a product of the interaction of two or more interconnected pieces of gear,

And that's why with proper testing you would isolate how much the DAC alone is contributing or rejecting.

, I don't know how you'd even use the dB level of ground-loop noise

Whats your problem with this? what would you uses to express a ratio? is said "compared to listening level."

Interesting comparison to make,

Comparison is you shuld not need an external device for this. filtering shuld be build in.

 

[pkane]

Comparison is you shuld not need an external device for this. filtering shuld be build in.

Because you say so? I have 20+ DACs and a number of audio interfaces on hand. None suffer from line-induced noise in my normally polluted-AC environment. Not in listening, and not in measuring. Amir's measurements don't show such sensitivity for most modern DACs, also.

So, no, I don't need an external (or internal) device for this, except when I need to break a ground loop due to a complex grounds topology when doing measurements. And PS Audio power plant doesn't help me there, but HS02 does.

 

[Ian Wendt]

My issue primarily comes down to the oft-made statement here in this thread and elsewhere that as long as you "buy a good DAC", then you won't need to worry about USB noise and it's not a problem. And that's clearly not the case.

"One could stop here and think that this is a useful device. But such is not the case because any DAC with half decent design would isolate itself against USB vagaries. "

"It seems that the iPurifier does what it says it does: clean up the signal/power lines over USB and with it help devices that need it. Then again every DAC I recommend of which there are probably 50 to 100, already perform their own filtering and don't need any help. After all, I test them all on my desktop workstation without any such filters."

"The Mark III revision of M500 improves on both looks and performance of this balanced DAC. In that subsystem, everything you could ask for is provided as far as performance. The headphone amplifier matches that overall but is a step down from total power point of view compared to what you can get. For most applications involving headphone use though, you should still be all set.

It is my pleasure to recommend SMSL M500 MKIII."

I fully understand that not all configurations will result in audible noise, but it clearly is an issue that comes up fairly often which a quick Google search will readily demonstrate, which is what led me to suggest that there might be a gap in the testing protocols.

 

[Lambda]

Because you say so?

Not only i...

Amir and others have repeatedly stated that a well-designed DAC will already filter power/be galvanically isolated, and devices like the Purifier and others are not necessary.

Amir's measurements don't show such sensitivity for most modern DACs, also.

They don’t test for this.

 

[pkane]

Not only i...



They don’t test for this.

Who is ‘they’? I’m describing to you what I’ve tested, not ‘they’. And I have one of the original PS Audio regenerators on hand, which works a lot better at removing line noise than all the modern incarnations, as it is a true regenerator. So I can compare these directly.

 

[voodooless]

Who is ‘they’? I’m describing to you what I’ve tested, not ‘they’. And I have one of the original PS Audio regenerators on hand, which works a lot better at removing line noise than all the modern incarnations, as it is a true regenerator. So I can compare these directly.

I think with “test for this” means to specifically inject the right kinds of noise into the USB power to see how the DACs react to this. This is the only way to actually do this in a controlled and repeatable manor.

Given though that not that many people have issues with ground loops, this kind of testing is just not done. Isolation equipment is cheap, easy and works.

 

[pkane]

I think with “test for this” means to specifically inject the right kinds of noise into the USB power to see how the DACs react to this. This is the only way to actually do this in a controlled and repeatable manor.

Given though that not that many people have issues with ground loops, this kind of testing is just not done. Isolation equipment is cheap, easy and works.

I like the line from a Bob Katz' video interview, where he says (repeating from memory):

"If jumping to conclusions was an olympic event, audiophiles would take first, second and third place."

 

[Lambda]

Who is ‘they’?

Amir's measurements.
The measurements don’t show sensitivity for common mode current from USB because it’s not tested.

On the other hand:

"One could stop here and think that this is a useful device. But such is not the case because any DAC with half decent design would isolate itself against USB vagaries. "

There are many DACs that measure good but have no mitigation "against USB vagaries" at all and as re result pass them to the output. (if you have them)
And to have them is perfectly normal and in "within spec"

Many modern and good DACs have some sort of controlled impedance between there digital and analog GND with different impedance and effectiveness.
only very view have true isolation.
And there are still some with no filtering or isolation at all. and they can (and some do) still measure fine.


The same way some noise and distortion on the AC line is in spec and should not effect audio quality weather you have it or not.
the same way common mode currents and noise on the USB port and interconects are normal and within spec and should not effect audio quality.
because "because any DAC with half decent design would isolate itself against USB vagaries."

 

[tmtomh]

And how can it completely avoided? you will always have some stray capacitance even with battery powered devices.



Whats why we don’t test with "real world" conditions but with synthetic tests.
DACs don’t get tested with music but with special test tones.
The same way you can measure CM impedance or rejection ration.


And that's why with proper testing you would isolate how much the DAC alone is contributing or rejecting.


Whats your problem with this? what would you uses to express a ratio? is said "compared to listening level."


Comparison is you shuld not need an external device for this. filtering shuld be build in.

It doesn't matter if stray capacitance is present; what matters is whether or not ground-related hum or noise is present at audible levels. Whether you measure that distance from the listening position, a foot or two from the speakers, or with your ear pressed against the speaker driver, is a question reasonable people can disagree on, precisely because we're talking about an engineering and personal-preference question of practical sufficiency, rather than literal 100% perfection.

As for your "that's why DACs don't get tested with music but with special test tones," no, in fact, you're incorrect about that. With a musical signal it's difficult to impossible to measure noise and distortion with anything close to the precision necessary to ensure (a) repeatable results with the same gear, and (b) comparable results between different pieces of gear tested at different times by different people. That is why we don't test DACs with music. And as far as a DAC is concerned, a 32-tone test tone and actual music are essentially the same, with the only difference being that the 32-tone signal is more regularized and controlled so that results can be repeatable and comparable.

As for isolating the DAC's role in ground-loop noise, I think that's more difficult than you're making it out to be. A reviewer like Amir would have to create a setup with a known ground loop in it, and he'd have to more or less randomly pick one DAC to serve as his reference for the "baseline" ground-loop noise. And - not a trivial thing - he'd have to determine what components of the overall measured noise were coming from the ground loop, because a ground loop doesn't always generate just 50 or 60Hz hum, and it doesn't even just generate obvious low-order harmonics - a ground loop can also increase the amplitude of higher-level noise that sounds identical to the hiss we hear from virtually all gear's self-noise (or self-noise being passed through or amplified from upstream in the chain).

And forever into the future, Amir would have to test every single DAC with that exact setup. And we would have to assume that the amplitude and frequency content of that ground loop never changes over time - which in my understanding is far from a safe assumption.

And after all that, we'd still end up with measurements that were of extremely limited value, if any value at all, because from several of Amir's reviews and in the widely varied experience of countless members of this forum (including me), we know that it's almost always possible to significantly reduce or for practical purposes eliminate audible ground-loop noise - and that it's foolish to think you can depend on something like a DAC to make your ground loop problem go away.

I noticed in your prior comment that you claimed you were citing facts that "many at ASR don't want to believe." I almost commented on that in my prior post; I will do so now: I see this tired "ASR members are hypocrites because they don't want to believe scientific facts that are inconvenient" trope come up periodically here on the forums. And I've noticed that every time someone trots it out, the "hard facts" they'e banging on about are not in fact things that ASR members don't want to believe. Instead, they are facts that ASR members acknowledge but don't think matter. And if a ground loop or stray capacitance is always present, but you can set up your signal chain so that it does not produce audible noise from your speakers, it doesn't matter that it's present. You could say the exact same thing about distortion: always present, can't deny that. But if it's at -120dB, who cares? (And if it rises to -80dB at 20kHz, who cares either - our hearing acuity isn't good enough to detect that.) We know that you can run an unfiltered 20kHz square wave into a Class D amp and cause its measured performance to degrade significantly. But that's not "a fact many at ASR don't want to believe." It's just a fact that doesn't matter.

So if you want tests that say, "this DAC has noise at -140dB but what happens if we put it in a system with a terrible ground loop problem?" then you're free to run those tests yourself. Until you do, and until you can produce some evidence or reasoning for why such tests would matter, I don't see the point.

 

[Lambda]

Given though that not that many people have issues with ground loops

sad who
I would say way more people have issues with that.
then people have issues with a ~115dB SINAD DAC because it dose not sound as good or have more noise and distortion then a 123dB SINAD DAC.
And DACs with >115dB SIAND are also cheap and common.

present at audible levels.

That’s a completely different statement. of cause its perfectly possible to have this below audible levels.

because a ground loop doesn't always generate just 50 or 60Hz hum,

of cause not.. this is why you would have a plot with Rejection vs. frequency.
And/or Impedance vs. frequency.

d he'd have to more or less randomly pick one DAC to serve as his reference for the "baseline"

Or uses Absolute values instead of an arbitrary basline...

And forever into the future, Amir would have to test every single DAC with that exact setup. And we would have to assume that the amplitude and frequency content of that ground loop never changes over time - which in my understanding is far from a safe assumption.

This is why you would test input/output CM Impedance vs. frequency of the devices under test and not the wohle loop.
It’s easy and repeatable.

"And forever into the future, Amir would have to test every single DAC with that exact setup"
Why?
It’s better to have incomplete data then no data? The tests and processes change alle the time and new data points get added.

 

[voodooless]

sad who

Says the amount of topics with people having these issues. It’s not a perfect metric, but good enough for me.

I would say way more people have issues with that.
then people have issues with a ~115dB SINAD DAC because it dose not sound as good or have more noise and distortion then a 123dB SINAD DAC.
And DACs with >115dB SIAND are also cheap and common.

Not really, people claiming to hear differences in DACs rarely state noise as the determining factor.

 

[pkane]

Amir's measurements.
The measurements don’t show sensitivity for common mode current from USB because it’s not tested.

And this is somehow remedied by a PS Audio power plant? What are you even talking about?

 

[amirm]

Long time listener, first time caller.

@amirm I purchased the SMSL M500 MkIII after reading the review of it that you posted. But it very much had an issue with USB power noise. Notable only when the GPU was under load, IE during gaming. When gaming, the noise would shift in pitch based on mouse movements. When the GPU was not being stressed, the noise would essentially go away. The issue was fixed by using an iFi USB Defender+ between the PC and the DAC. The same issue with a Fiio K7 was also resolved with using the USB Defender+. Neither DAC exhibited any noise via their respective headphone outputs, the noise was only present with the line-out and was not affected by pre-amp volume level either.
Perhaps there is a gap in the testing protocol?

Hi there. Were you using balanced XLR or unbalanced out of those DACs? If the latter, then you are going to face ground loops which manifest themselves with a computer as you not. I too have had that happen and I know of others who are able to listen to not only their GPUs, but in extreme cases their mouse, etc.! If you use a complete balanced XLR path, this hugely reduces the chances of this happening. If it still happens, an easier solution if your computer has Toslink. That provides full optical isolation and is better than this device although it will limit your sample rate to 96 kHz.

RCA interconnects were invented to connect internal components of a stereo system and not for external interconnect. Their expansion and almost universal usage in audio system has been a mistake and is responsible for so much grief people for people who get subjected to its tendency to show ground differentials. This is why I am so in favor of balanced connections.

 

[amirm]

The DACs are not tested with common mode noise / current on the USB port.
But in the Real world it is inevitable to have some.

The Audio Precision Audio Analyzer has Extremely good CMRR and Functionally isolated inputs. so no Common mode currents.
AMPs on ASR don’t get tested for there input CMRR but in gneral CMRR will be lower then from the AP.

You are wrong on multiple fronts there. First, I test every DAC on a very noisy, workstation class desktop PC with GPU and such. Many people have more quiet source systems than I. To the extent I test that way, I am representing real world situation. You talk like I have used some kind of lab or isolated USB connection. I am not.

AP's unbalanced inputs do not have good let alone extremely good CMRR. The are floating which is sometimes good, sometimes bad. In many devices the unbalanced inputs on AP show ground loops which I then try to optimize with different grounding schemes and I note that in review. For high performance DACs that I test, this rarely if ever occurs. But it can happen when there are issues. Take this Chord DAVE DAC Review:

Notice how I caught variability of noise and it differing in each channel.

You talk about measuring CMRR. The input to those DACs in question is the USB. There is no such concept for USB in the way you are implying. That is a digital interface and I can't just ask the AP to measure its CMRR. So as far as DACs are concerned, your comment is non-sequitur.

You also have not shown any data to indicate common mode noise is an issue as opposed to plain ground loop caused by USB ground being shared with the RCA output ground.

For amps, AP actually represents a worse load as it is mains powered whereas your speakers are not. So the case is backward of what you say and is exaggerating tendency of an amplifier as far as ground loops. The floating inputs help though as does me optimizing for that situation.

As a practical point, we know that such problems are exceedingly rare or we would have hundreds of reports of issues. If people used proper DACs and XLR interconnects, the issue would be essentially non-existing, removing the need to do any testing anyway.

 

[amirm]

My issue primarily comes down to the oft-made statement here in this thread and elsewhere that as long as you "buy a good DAC", then you won't need to worry about USB noise and it's not a problem. And that's clearly not the case.

You are missing important context here. Vast majority of people buy these devices not because they have audible noise like you, but because they think these devices gets rid of inaudible noise over USB that messes up their soundstage, keeps their digital from sounding analog, etc. They imagine a "dirty" USB connection that does those things. For them, the statement we make is 100% valid: that no high performance DAC assumes a clean USB signal and passes it around to the analog output.

For people who use unbalanced connections, as I have explained before, due to broken system that unbalanced is, are going to potentially get hit with ground loops. A better solution is to use balanced connections. That would isolate the ground on the output of your DAC in acting as the reference for the signal. It may still be possible to get noise but the incident will become extremely rare.

I fully understand that not all configurations will result in audible noise, but it clearly is an issue that comes up fairly often which a quick Google search will readily demonstrate, which is what led me to suggest that there might be a gap in the testing protocols.

Countless people deploy these DACs with no audible issues. To the extent someone uses RCA connections and gets ground loops, then that is that. The solution is to heed the advice I give frequently to build systems using balanced connections. If you still get noise like this, then it is a very rare event and one that I cannot test for.

 

[Lambda]

For people who use unbalanced connections, as I have explained before, due to broken system that unbalanced is, are going to potentially get hit with ground loops. A better solution is to use balanced connections. That would isolate the ground on the output of your DAC in acting as the reference for the signal. It may still be possible to get noise but the incident will become extremely rare.

With balanced connections. (depending on the exact implementation) You will likely have exactly the same "ground loops" since they normally don't isolate the ground output of your DAC. they just don’t use the GND as Reference.
This is of cause better and helps to suppress the noise but only with the "common-mode rejection ratio (CMRR) " of your amp/preamp.

or them, the statement we make is 100% valid: that no high performance DAC assumes a clean USB signal and passes it around to the analog output.

but you said yourself.

I too have had that happen and I know of others who are able to listen to not only their GPUs, but in extreme cases their mouse, etc.! If

So if this problem can some times be so pronounced and strong that its clearly audible.... wouldn't this mean that it can also be less pronounced and strong so it’s not audible but still there.

You also have not shown any data to indicate common mode noise is an issue as opposed to plain ground loop caused by USB ground being shared with the RCA output ground.

Tomato tomato
The "ground loop" is a cases of common mode noise and common mode noise can go trough a "ground loop"
"caused by USB ground being shared with the RCA output ground."
Yes that is my point. IF USB ground being shared with the RCA output ground current/voltage can pass trug.
This is why many good DACs don’t do this.

How good they isolate USB and Analog ground (or how mush impedance they have) and subsequently how mush they suppress noise (in common mode) from the Digital port to the analog port can be measured.

And in combination with a defined cable (impedance) and receiver expressed as "common-mode rejection ratio"

 

[amirm]

So if this problem can some times be so pronounced and strong that its clearly audible.... wouldn't this mean that it can also be less pronounced and strong so it’s not audible but still there.

In every case I have experienced, switching to a higher performance DAC has solved the problem. The offending DACs I have measured have had serious problems with USB interference, visible in their measurements. Here is an example: https://www.audiosciencereview.com/forum/index.php?threads/schiit-modi-1-usb-dac-review.10273/

The problem is so acute with this DAC that I routinely use it to test ideas like USB cables mattering, power supply on USB mattering, etc.

So our tests are absolutely catching design issues like this.

What they won't measure are issues orthogonal to the device under test. Noise over USB ground will travel on the RCA output and potentially cause audible problems. Switch DACs is not going to help you except randomly.