IFI Neo Stream
Contents
Why, your author does not know, but he often confuses IFI and FIIO. We now have the IFI Neo Stream in front of us…. A very complete streamer with ROON as well as UPnP and Airplay. On the front is a small display. Not readable from a distance, but it’s handy when you set things up.
The IFI has single ended analog outputs as well as digital outputs: AES, Coax, Optical and I2S. A funny bonus is the 4.4mm balanced headphone output. A bit of a crazy size, but there just wasn’t any more room on the back. It really is ram-packed!
Other connectivity includes a fiber connection for networking, usb A for audio, USB-C for system upgrades, Ethernet and a jack for an upgrade power supply.
The Sound
Prejudice is tricky and dangerous… Somehow, your author didn’t imagine as much. Completely unjustified, because this IFI just plays very nicely. We left the filter on standard for the test. And then the soundstage is complete and nice and robust. Separation of instruments is good and the imaging is precise. A little more “stage feel” than the Volumio and JBL…. but that’s a preference. In short: very little goes wrong with this IFI!
What is remarkable about the IFI is that the whole thing feels very ‘quiet’. How IFI did that is beyond us, but it’s something all three of us felt with this streamer. It could be due to the “custom dac” that IFI used?
Fiber?
Now IFI also included a fiber converter solution. In measurements, this adds little except some extra noise. But we are still curious to see how that translates to our listening test…
We can tell you that we strongly advise against it. It really does not sound any better in our case. The sound image becomes dull and dark and the stereo image falls completely backwards. We cannot imagine that anyone at IFI has tried this. We estimate that the phone charger is to blame…. power is everything in hi-fi, after all.
Measurements IFI Neo Stream
The IFI Neo Stream measures fine on its own. However, we find the ‘bitperfect’ mode rather crazy (very high noise) and the fiber module simply measures worse than the LAN cable directly in the streamer. In short: either improve the power supply of the fiber module or simply do not use it.
Furthermore, we see a series of measurements that are just fine in this field. Noise is reasonably low. Both over the network and the power supply.
Specifications IFI Neo Stream
| Model and type | IFI neo Stream |
| DAC Chip | Custom (Burr brown decoding / Xmos chip) |
| Outputs | Single ended / headphones I2S, AES, Coax, optical |
| Digital in | LAN / Optical / Wifi / Bluetooth |
| Analog in | No |
| Own app | Yes |
| Pre-amp | Yes, analog? |
| Price | 1299 |
Ifi Neo stream , the optical network connection generates more noise!!!! How is that possible?
The optical should disappear noise not introduce it? Why didn’t you try with a better power supply?
The power supply is the issue. And we test as-is. This is how you buy the streamer.
So the provided power x is for the neo stream only and the small usb ps causes all the problem?
Yes
Since the noise is generated in the eth to optical box shouldn’t be eliminated during the optical transfer to the streamer?
Apparently not… I measured multiple times…
So is it safe to assume that optical connection isn’t the holly grail for noise isolation? Or the measured noise might appear from another source?
I guess the implementation is just badly done… I never measured issues with other streamers.
In this instance the noise is probably injected in the signal before conversion to light. On the other end, the signal is simply converted back, including the noise.
No, optical isn’t a/the holy grail. If it would be, we all would use Toslink.
In case of this iFi streamer: just hook it up with a regular Ethernet cable, put a switch before it (one we tested adequate) and the result is good.
That…
Kind of ironic that an iFi product introduces noise with a supplied power supply, don’t you think?
Question, what are your settings for the dScope III? I typically use 100K input impedance, sample rate 192K, 32K FFT Prism-7 window and average over 8 samples. Your low end measurements look a little off. One thing you also have to remember is the VBUS/GND is tied to the internal supply and therefore you can get ground loops really easy when testing.
Thanks,
Gordon
Interesting test! Thank you. Going above 1000 eur by a coupld hundred the Eversolo Master version should do even better on the jitter front due to the femto clock and op amps. Would be interesting to measuer that one also.
We will test it as well