• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

Review and Measurements of iFi iPurifier S/PDIF Digital Audio Filter

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
13,493
Likes
5,662
Location
Seattle Area
#1
This is a detailed measurement and review of iFi iPurifier. This is a device that is advertised to clean up your coax S/PDIF signal from all ills including the common flu to audio distortions. It is a tiny little box the size of a thumb drive:


Not shown is the fact that it actually needs external power and iFi ships it with their switching wall-wart "iPower" power supply.

This is not a filter device but rather, one that receives digital data over S/PDIF, buffers it (puts it in memory) and then retransmits it. iFi has a nice block diagram of it on their site:

upload_2017-12-27_22-38-33.png

Who wouldn't want their waveform cleaned up in red and turn it into the one in cyan? It is like getting an enema. Not that I have had one but I imagine it being the same.

Because the iPurifier regenerates S/PDIF, it should have good ability to remove incoming jitter. In a way it acts like an Ethernet switch/hub would for that connection. There is a problem here though. S/PDIF signal can drift in speed. It can run slower or faster than the stated sampling rate. For that reason, there is that memory buffer block in yellow. That is not a complete solution though as if the source runs faster than the internal S/PDIF interface in iPurifier, it could eventually fill up. The iPurifier has an undocumented measure for changing its clock frequency. That could cause fidelity issues and is something that is not easy to measure.

I bought my unit in August of 2017 for $149 shipped from Amazon. In other words it costs more than some decent DACs.

Measurements
Since this is a S/PDIF device and J-Test was designed for exactly this application, let's start with that. I programmed my Audio Precision Analyzer to output J-test while jittering the digital output to the tune of 4 Khz sine wave. I pushed that into Topping D30 DAC (review here: https://audiosciencereview.com/forum/index.php?threads/topping-d30-dac-measurement-and-review.2016/) and measured the analog output. I then did the same thing while routing the signal through iFi iPurifier:

AP J-Test induced jitter 44100 Khz Topping D30.png


Focusing on the direct interface in yellow, we see the two classic jitter sidebands. They are at +- 4 Khz relative to our main signal tone at 11.05 Khz. This shows that the Topping has little filtering of incoming jitter.

Running that signal through iFi iPurifier does away with that jitter as if it was not generated at all. So mission accomplished. Or is it?

I had induced quite a bit of jitter in there. Let's go with a real-life scenario of using a Gustard U12 USB to S/PDIF converter driving the Topping D30 DAC and see what inherit jitter we have and how much of it gets cleaned up by iFi iPurifier:

Gustard U12 J-Test 48 Khz Topping D30.png


Not only did we not get a cleaner signal, but we got some added low frequency noise (in red) when we used the iFi iPurifier! Part of that is likely created by the iFi's switchmode power supply which leaks mains current. But there are other spurious signals there that I can't explain. Either way, we went backward here, not forward!

Let's try that with a different DAC, this time the Emotiva DC-1 DAC (full review later). The source is the digital output from my Audio Precision again, but without induced jitter. First let's look at J-test at 96 Khz:

AP J-Test 96 Khz.png


Yeh the DC-1 puts out quite a bit of noise on its own. Alas, the iFi iPurifier did nothing but increase those levels slightly (in red). No good deed is left unpunished here!

Same story at 44.1 Khz:

AP J-Test 44100.png


We see the increased low frequency noise from iFi much better here. Not good.

Let's complete the round by driving the Topping D30 again using the AP:

AP J-Test 44100 Topping D30.png


Same picture again. Increased noise and distortion from iFi and no improvement elsewhere.

Let's go with a much cheaper DAC. Maybe those can benefit from this device. For that, I tested the Prozor 192 Khz DAC which costs $29 shipped!!! Here it is with induced jitter again, with and without iPurifier:

AP J-Test induced jitter 44100 Khz Prozor.png


Yes, there is a whole forest of distortions there. What do you expect for $29 China special???

Addition of the iPurifier did eliminate the induced jitter (in yellow) but has done absolutely nothing for all the rest. Those are jitter/noise generated internally by the DAC and the ifi is ineffective there.

Let's jump to the other extreme and see what the $3,400 Exasound E32 does here:

AP J-Test induced jitter 44100 Khz Exasound E32.png


As expected, a high quality DAC like this one completely ignores the induced jitter on its S/PDIF input. Addition of iPurifier therefore just added noise as show in red.

Finally, let's look at the "eye" pattern of the S/PDIF waveform in time domain. For this, I used the Audio Precision as the reference:

ifi iPurifier Eye Pattern with AP.png


We see that the ifi output is slightly better (rises faster). But that is due to the fact that output of the iFi is directly connected to the DAC/Analyzer where as the output of the source had to go through a cable. In that regard, there is really no improvement here.

Conclusions
The only thing the iFi iPurifier is good at is eliminating high amounts of jitter on S/PDIF signal with lower end DACs. Despite what they show in their marketing material though, typical S/PDIF signal across a short cable is far, far cleaner than what they and their customers assume. The type of distortions and jitter we see in the output of DACs these days are mostly created internally and no external device can fix that.

Typical of these tweak devices shipped with switchmode supplies, combined with small enclosure means that this device can actually harm the output of the DAC! This was clearly visible < 2 KHz with various noise and spurious signals. Mind you, it is not an audible problem but why spend money to make the performance of your DAC worse?

At the higher end of the scale, good DACs have proper PLLs that filter out incoming jitter very effectively and no such device is useful or necessary.

Bottom line, I cannot find a real scenario where ifi iPurifier does any good. Yet it was easy to show that it does some harm. NOT RECOMMENDED.

Save your money for music, or a better DAC driven by isolated asynchronous USB signal.

As always, comments, corrections, feedback, etc. are welcome.

If you find this article useful, please consider making donations through Patreon (https://www.patreon.com/audiosciencereview) or if you are member, by "Upgrading your account" (see: https://www.audiosciencereview.com/...eview-and-measurements.2164/page-3#post-59054). The funding enables me to purchase more products and test.
 
Last edited:

PaperPro

New Member
Joined
Dec 28, 2017
Messages
2
Likes
1
Location
NJ
#2
Interesting results Amir. You mention that the supplied iPower wall wort introduces noise:
"Part of that is likely created by the iFi's switchmode power supply which leaks mains current."

Could your results have been different if a quieter power supply is used?

Disclaimer: I do use an IFi SPDIF iPurifier in my system. I found that the supplied iPower did introduce noise and replaced it with an UpTone Audio Ultracap LPS-1 with much better results. Granted the LPS-1 is $400.00 but it came with a 30 day return policy as did the SPDIF iPurifier. So I thought what the heck I would just be out the nominal shipping costs for the experiment.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
13,493
Likes
5,662
Location
Seattle Area
#3
Interesting results Amir. You mention that the supplied iPower wall wort introduces noise:
"Part of that is likely created by the iFi's switchmode power supply which leaks mains current."

Could your results have been different if a quieter power supply is used?
Some of it yes. I have tested that supply with other products (e.g. ISO Regen) and it definitely contributes mains related noise. I will give it a try with iPurifier and report back.

And oh, welcome aboard! :)
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
13,493
Likes
5,662
Location
Seattle Area
#4
Here we go. This is a test comparing powering iFi iPurifier with my lap supply versus its own iPower power supply that comes with it:

AP J-Test 44100 Topping D30 DAC with iFi iPurifier lab supply test.png


As we see, the low frequency noise goes away completely when I power it with my linear lab supply.

It is just so puzzling that these people sell tweaks the improve the lowest of low levels, then bundle in a switching supply when the power requirements are so low as to easily be met with linear supply. Yes, it requires two versions for 120 vs 240 volts but still.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
13,493
Likes
5,662
Location
Seattle Area
#5
And for completeness, here is the output of the Topping D30 DAC when driven direct vs going through the iFi iPurifier powered by my lab supply:

AP J-Test 44100 Topping D30 DAC with iFi iPurifier and lab supply test.png


There is slightly more noise around 24 to 30 Khz with iPurifier (yellow) but performance is essentially the same with and without it.

In other words, if you want the iPurifier to do no damage, it needs to be driven by a linear power supply.

I should note that these are not audible problems as the levels are quite low. Just lack of good engineering and system design.
 

Cosmik

Major Contributor
Joined
Apr 24, 2016
Messages
2,103
Likes
840
Location
UK
#7
Amir said:
There is a problem here though. S/PDIF signal can drift in speed. It can run slower or faster than the stated sampling rate.
Not "can"; it does. Whether it is by 1 part per billion or 100 parts per million, the S/PDIF clock is faster or slower than the stated sample rate.
For that reason, there is that memory buffer block in yellow. That is not a complete solution though as if the source runs faster than the internal S/PDIF interface in iPurifier, it could eventually fill up. The iPurifier has an undocumented measure for changing its clock frequency. That could cause fidelity issues and is something that is not easy to measure.
It is just another PLL-based buffer similar to the one in the DAC that follows it, and it does adjust its sample rate from time to time - a form of jitter, so it is not true to say that it eliminates all jitter.
 

mindbomb

Active Member
Joined
Nov 11, 2017
Messages
240
Likes
111
#8
I think one potential limitation of this review is that you were testing the coaxial output of the AP and gustard u12, which might have already been very good. Can you check ostensibly worse sources, like the fiio e10k, a computer, a television, and see if they also show no improvement?
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
13,493
Likes
5,662
Location
Seattle Area
#9
I think one potential limitation of this review is that you were testing the coaxial output of the AP and gustard u12, which might have already been very good. Can you check ostensibly worse sources, like the fiio e10k, a computer, a television, and see if they also show no improvement?
I am working on getting my Chromecast audio working for that reason. I will report back later. And yes, I can dig up other worse performing S/PDIF output but is that really the prime scenario for spending $150 on this device?
 

Jinjuku

Addicted to Fun and Learning
Patreon Donor
Joined
Feb 28, 2016
Messages
618
Likes
221
#10
I am working on getting my Chromecast audio working for that reason. I will report back later. And yes, I can dig up other worse performing S/PDIF output but is that really the prime scenario for spending $150 on this device?
This all proves: Go USB.
 

Wayne

Active Member
Patreon Donor
Joined
May 26, 2017
Messages
170
Likes
34
Location
Los Angeles, CA
#11
From iFi webpage: ( https://ifi-audio.com/portfolio-view/accessory-spdif-ipurifier/ ) From Apple TV to Google Chromecast to PS4/Xbox to 4K Smart TVs, SPDIF is the main digital audio output. As with all mainstream products, the critical SPDIF signal transmitted is of fairly low quality. The SPDIF iPurifier runs the SPDIF signal through 4 comprehensive stages, and the result is a fully-restored, perfect SPDIF with jitter eliminated. The result is sound quality that has greater warmth, resolution and dynamic contrast.

Checking to be sure I understand the product evaluation.

1. re: "perfect SPDIF " @ Amir As I understand your testing did not actually check the SPDIF to see if the signal shape was improved... or can it be induced from the tests? So we (a general term) can not address that claim.

2. re: "jitter eliminated"
Not "can"; it does. Whether it is by 1 part per billion or 100 parts per million, the S/PDIF clock
is faster or slower than the stated sample rate.
Per Cosmik the jitter can not be eliminated and @Amir if I understand your testing, the jittter was not improved?
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
13,493
Likes
5,662
Location
Seattle Area
#12
1. re: "perfect SPDIF " @ Amir As I understand your testing did not actually check the SPDIF to see if the signal shape was improved... or can it be induced from the tests? So we (a general term) can not address that claim.
I showed the eye pattern that demonstrates this. I also did a bunch of tests with my scope and the iPurifier without a cable on its output produces very nice waveform. Then again same tests showed that my Audiophilleo did the same. I will do some captures and post.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
13,493
Likes
5,662
Location
Seattle Area
#13
Per Cosmik the jitter can not be eliminated and @Amir if I understand your testing, the jittter was not improved?
My read of their architecture:

upload_2017-12-28_13-57-34.png


is that there is an independent receiver which captures the data and puts it in memory. It then uses a fixed clock (i.e. NOT tracking the input) to transmit those samples out. In that regard, all incoming jitter is left behind (the memory buffer).

That is of course an idealized explanation. In practice there can be some leakage from input and output. The point is that this is not just a filter. It captures and then retransmits. If done well, as some DACs do, it can eliminate all input vagaries. The price as I explained though is the clock adjustment mechanism to track the input.
 

tomelex

Addicted to Fun and Learning
Patreon Donor
Joined
Feb 29, 2016
Messages
694
Likes
192
Location
So called Midwest, USA
#14
Amir, thank you for your work on this, very revealing and informative.
 

Cosmik

Major Contributor
Joined
Apr 24, 2016
Messages
2,103
Likes
840
Location
UK
#15
My read of their architecture:
"capable of producing millions of frequencies". The central issue is that the sender of the data is not using the same clock as the device, so the device has to adapt itself to match (=a form of jitter), or hope that the latency it introduces is (a) permissible and (b) sufficient not to fill or empty the buffer memory over a listening session.

Isochronous USB introduces exactly the same issues.

A description of the problem and why asynchronous USB is the sensible solution:
https://www.edn.com/design/audio-de...l-memoir-of-engineering-heartache-and-triumph
https://www.edn.com/design/audio-de...rge-should-the-FIFO-be-The-D-A-Diaries-Part-2
https://www.edn.com/design/audio-de...-Distortion-Goals-The-D-A-Diaries-Part-3-of-3
 

RayDunzl

Major Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
5,274
Likes
1,279
Location
Riverview, Fl
#17
Hmmm...

How can I install Asynchronous USB ports to replace the S/PDIF ports in my TV, HDRadio, CDPlayer, CableBox, S/PDIF switch, DEQ2496, and miniDSP that feed my DAC, which does have a USB input that I never use?

 

Wombat

Major Contributor
Joined
Nov 5, 2017
Messages
2,280
Likes
795
Location
Australia
#18
How 'bad' does jitter have to be for humans to hear it?
How does it translate to distortion in the analogue domain?
 

Cosmik

Major Contributor
Joined
Apr 24, 2016
Messages
2,103
Likes
840
Location
UK
#19
Hmmm...

How can I install Asynchronous USB ports to replace the S/PDIF ports in my TV, HDRadio, CDPlayer, CableBox, S/PDIF switch, DEQ2496, and miniDSP that feed my DAC, which does have a USB input that I never use?
I am just taking the implication of the necessity for this device to its logical conclusion. I, personally, would be perfectly happy to use S/PDIF in all the applications where S/PDIF is the best or most convenient option - without duplicating the DAC's input buffer as an extra $149 dongle with a wall-wart. If, however, I can use asynchronous USB without penalty, I will do so: my DAC with the aforementioned 'state of the art' interface cost about £90.
 
Top Bottom