Thanks Amir for sharing these measurements.
If I may offer an alternative interpretation of the results in conjunction with my own listening observations:
The fact that a DAC incorporates internal filtering mechanisms does not necessarily imply that filtering applied upstream of those stages is functionally equivalent or redundant.
A useful analogy would be the treatment of DC artefacts in the mains supply. While a galvanic isolation transformer cannot pass DC by definition, it does not follow that DC-related artefacts upstream are entirely irrelevant. In practice, introducing a dedicated DC-blocking stage ahead of the isolation transformer in my own system resulted in a clearly different—and not merely subtle—sonic outcome. This suggests that artefacts which are theoretically blocked can still influence operating conditions, magnetic behaviour, or downstream interactions in non-trivial ways. As always, system and room dependence applies.
With regard to the measurements discussed here, they show a distinct reduction of noise components above approximately 2 kHz. This frequency region is psychoacoustically relevant, for example with respect to second-order harmonic components and their influence on voice reproduction and transient clarity.
In my own listening, I perceived subtle yet fundamental reductions of high-frequency noise and masking effects across the spectrum, which is consistent with the observed changes and therefore not unexpected.
For those interested in system optimization, I would therefore encourage simply trying the device in their own setup. It may account for a small portion of what is often referred to as the “night effect”, i.e. the observation that music can sound subjectively better during night-time hours (at least here in Germany, including smaller cities).
It is important to clarify that devices such as the iPurifier do not and cannot address all underlying mechanisms associated with the night effect. Large-scale contributors—such as reduced mains distortion, lower grid load asymmetry, or generally quieter electromagnetic environments—are clearly outside the scope of USB-side conditioning.
However, a smaller but potentially relevant subset of the night effect can plausibly be addressed. In particular, the iPurifier targets USB-related high-frequency noise, common-mode interference, and signal-integrity issues at the physical layer. These mechanisms are known to be influenced indirectly by system activity, RF traffic, and switching noise, all of which tend to be reduced during night-time conditions.
From a conservative standpoint, it therefore seems reasonable to assume that such a device may account for a limited fraction of the perceived night effect—perhaps on the order of a few percent up to around ten percent—by reducing HF noise and masking artifacts that persist even in otherwise well-optimized systems. This does not replicate the night effect as a whole, but may slightly narrow the gap between typical daytime and night-time listening conditions.
Finally, it is worth noting why such effects may be more readily observable in laptop-based playback chains. Laptops integrate high-speed CPUs, switching regulators, wireless transceivers (Wi-Fi, Bluetooth), and densely packed digital subsystems operating within a shared ground and power distribution network. Even when running on battery power, these subsystems generate broadband high-frequency noise and transient current activity that can couple into the USB interface at the physical layer.
While modern asynchronous USB DACs are highly effective at isolating audio clocking from host timing, they are not fully isolated from host-generated common-mode noise, ground modulation, or PHY-level signal-integrity issues. As a result, residual high-frequency artifacts originating from the laptop environment can still influence reference stability, analogue stages, and noise shaping within the DAC.
In this context, a device that specifically targets USB-side high-frequency noise, impedance balance, and common-mode interference may have a more clearly audible effect than in systems where the source is electrically simpler or already externally isolated. This does not imply a deficiency of the DAC or the USB protocol itself, but rather reflects the inherent complexity of modern laptop platforms and their electromagnetic operating environment.
I hope this helps to add some additional nuance to the discussion. The measurements themselves appear sound; my main point is that the conclusions drawn from them may benefit from a slightly less binary interpretation.
