DJM ACTIV Audio Ethernet EMI Filter Review (Bonus!)

[pseudoid]

@amirm, as always, thank you for the your time and the review.

DJM ACTIV Audio Ethernet EMI filter.

I would accept this device as being just a functional EMI filter. Unfortunately, I cannot figure out how it knows that there is 'audio' content (within in the I/O stream of data/packets) that is going through it!
Unless DJM removes the word "Audio" in the name of this device, I will consider this thing to be another audiophool device.

 

[LoFiAudiophile]

It us a tough call in that there totally nonsense tweaks out there which do nothing. This box does what it says.

Amir, your objective test shows that it doesn’t. If you purchase a BER debug tool, it too will show it doesn’t do a thing. This device has zero impact on TCP performance for what it claims on the box - some of us are in the industry, doing this for decades and literally some of the top engineers in this field - saying it won’t do a thing.

There are specialized scenarios for EMI filtering but none that apply to an environment where critical audio listening is important, and even there fiber outweighs the complexity of any device. Even physics says an in-line device on differential pair is pointless for filtering noise. Plus an inline passive device like this one is not useful for Ethernet when NIC PHY already incorporates magnetics for galvanic isolation.

The 10G compatibility is a big sign - it’s a passive device. It probably just has discharge tubes for surge suppression like you find in $10 surge suppressors. It can probably protect an equipment against a lightening strike but is no more effective at anything else than what your Belkin surge suppressor can already do.

This device is $2000 paper weight, all the cool spec sheet notwithstanding. Snake oil with a spec sheet is still snake oil!

 

[Bob101]

Note that their company's core business is EMI filtering and is not focused on audio per se. On that front, this seems like a performant device. I think it would have good application in industrial domain. I have for example heard of CNC machines causing interference over Ethernet.

Finally, I can listen to hi-res Bach sonatas over ethernet while using that industrial CNC machine in my living room.

 

[KSTR]

Amir, your objective test shows that it doesn’t. If you purchase a BER debug tool, it too will show it doesn’t do a thing. This device has zero impact on TCP performance for what it claims on the box - some of us are in the industry, doing this for decades and literally some of the top engineers in this field - saying it won’t do a thing.

You seem to have not yet understood what this filter does. It obviously does not filter the signal, the signal is completely untouched, on a logical level. Physically, it runs through a sort of one-port hub.

It filters/eliminates common-mode noise. Repeat: common-mode noise, common mode noise, common-mode noise.

Some of us are as well in the industry (though maybe a different one than yours) for decades and use these kind of filters all the time and experience the effectiveness of them every day.

And Amir's test was neither intended nor capable to show the full filter action, as explained earlier.

I would accept this device as being just a functional EMI filter. Unfortunately, I cannot figure out how it knows that there is 'audio' content (within in the I/O stream of data/packets) that is going through it!

It does not know that there is audio content and as mentioned numerous times by now it doesn't touch the signal, by definition.
The only difference between their standard model and this one is the housing that is simplified and more appropriate for "desktop use" with regard to the location of the RJ-45 jacks.

This "audio" filter is the only one they quote a price for on their website. I believe they are testing new markets for their technology.

Exactly. Normally, you cannot buy this specialized stuff on the end user market as a private person, for good reasons. So, the audio community (or a subset of it, actually), is a new market they try to exploit and that's just fine.

 

[KSTR]

The 10G compatibility is a big sign - it’s a passive device. It probably just has discharge tubes for surge suppression like you find in $10 surge suppressors. I

It's an active device, did you note the power supply input?
You can see about half of the PCB here: https://6moons.com/audioreview_articles/djm-filter/
There is also some quoted background info from the manufacturer.

 

[LoFiAudiophile]

It's an active device, did you note the power supply input?
You can see about half of the PCB here: https://6moons.com/audioreview_articles/djm-filter/
There is also some quoted background info from the manufacturer.

Half the PCB notwithstanding, there isn’t a microcontroller you can buy for $2000 that can demodulate, modulate and noise-shape 10000BaseT. A switch, sure, but then there are plenty 5 port 10G BaseT switches available on Amazon for ~$200.

It would be settled if we could do a tear-down but I’m not about to blow $2000 on a product whose target customer base is unlikely to pay attention anyway.

 

[Cbdb2]

It filters/eliminates common-mode noise. Repeat: common-mode noise, common mode noise, common-mode noise.

So does the input transformer on all Ethernet switches. So this is probably just a switch.

 

[Cbdb2]

It's an active device, did you note the power supply input?
You can see about half of the PCB here: https://6moons.com/audioreview_articles/djm-filter/
There is also some quoted background info from the manufacturer.

"The filter rejects unwanted signals from 10kHz to
40GHz at better than 100dB. Military grade PASSIVE filtering
technology ensures..."

How does passive filtering filter 10khz to 40Ghz but not filter out the signal (a square wave with many harmonics) unless it only filters the common mode interference like a transformer?

So $2000 for a good transformer and a switch, and all you probably need is the transformer.

 

[Descartes]

This is a review and detailed measurements of the DJM ACTIV Audio Ethernet EMI filter. It was sent to me by the company and costs US $1,950.
View attachment 380788
If there was an attempt to justify the price of a small component like this, DJM has done it. The unit feels super dense and solid and a nice shiny finish and writing which doesn't come across in the picture well. There is an ethernet port and external power supply feed on the left of the indicator LEDs. The power supply is a proper, Meanwell supply. On the right is the outbound Ethernet port. Operation was automatic and without any impact despite me connecting and disconnecting cables a number of times.

This is a proper company building this device, unlike many high-end audio tweaks seeing specifications and certification not only in the company website, but also back of the device (abbreviated here):
View attachment 380789

Here is the filter specs:

View attachment 380790

I wish a transfer function graph was provided as I can't quite tell where the filtering starts from, other than full 100 dB at 10 kHz. Alas, I don't have equipment to measure noise and interference on Ethernet filter either so can't provide plot it that. But I did find an indirect way to do so at the end of this review.

DJM ACTIV Audio Ethernet Filter Measurements
As frequent readers of ASR know, my main focus is to see if the output of an audio device changes as opposed to what goes in it. That is what we hear and what ultimately matters. To that end, I pulled out the yet to be reviewed Eversolo DMP-A8 streamer for this testing (full review to come shortly) for the measurements here. Naturally Ethernet connection was used to stream test signal to the A8 and balanced analog outputs measured. Here is the dashboard of DMP-A8 with volume reduced one notch and Ethernet connection without the EMI filter:
View attachment 380791
Superb performance as we expect from Eversolo. However, there are some noise and distortion spikes (well under audible threshold < -130 dB). Can some of this be from Ethernet link? Let's insert the AVTIV Audio filter in the path:
View attachment 380792
There is no difference other than run to run variations. Note that this doesn't mean noise was not filtered. But that the A8 is implemented well as to be isolated from Ethernet. What is there as far as spurious noise components is inherent in the DMP-A8, not from incoming noise on Ethernet.

To see if any ultrasonic noise is filtered, I ran 1 MHz FFT with and without EMI filter:

View attachment 380794
Again, we see noise components but they must be internal as there is no difference with and without filter. I also tested with J-test signal with similar outcome:
View attachment 380795

I then decided to just look at the spectrum of noise without streaming anything to DMP-A8:
View attachment 380796

Again, no difference.

For the final test, I decided to use my AC transformer that powers my headphone measurement gear. This has proven effective in other interference tests and did the job here:
View attachment 380797

We see that there is no difference at mains 60 Hz interference but as we go higher in frequency, we progressively see more attenuation. Keep in mind though that the the graph is massively magnified with the top of the graph at whopping -130 dB or 15 dB below threshold of hearing. So this noise would not be audible in either case. But we do see that filtering does exist.

Conclusions
I commend DJM Electronics for not only offering to send this filter, but insisting that I test it despite me cautioning them that I have not found any of these devices to make a difference that would be audible. Note that their company's core business is EMI filtering and is not focused on audio per se. On that front, this seems like a performant device. I think it would have good application in industrial domain. I have for example heard of CNC machines causing interference over Ethernet.

For audio, my tests on high performance streamers shows that the product is not needed. However if you are experience data errors on Ethernet due to special situation you might have, the DJM Electronic's ACTIV Audio Ethernet EMI filter may help deal with that.

------------
As always, questions, comments, recommendations, etc. are welcome.

Any donations are much appreciated using: https://www.audiosciencereview.com/forum/index.php?threads/how-to-support-audio-science-review.8150/

Another ripoff!!!

 

[im_gumby]

I've been in enterprise/commercial networking for decades, this device is utter snake oil. If there's errors, it's almost always an equipment issue, faulty cable, faulty SFP, etc (or a config issue, although less likely these days). As far as "industrial" equipment goes, most companies with a need would be using ruggedized switches and firewalls/routers, but for EMI resilient gear, you're getting into the niche. I worked at a company [that shall remain unnamed] that may or may not have made specialized networking gear for military applications made to withstand EMI pulses during a nuclear war. That said, we're talking about an entirely different application.

Yeah.
Today... your high speed networks are going to use Fiber so no need for any RFI filtering when its optic. Your transceivers will be at the network switch converting back so you're already in shielded environment so no need for any RFI shielding.

So yes, this is a bit of snake oil even if it does what it claims to do. Its unnecessary.
Note: Within the rack you can see short runs using copper but most everything else is fiber.
Not to mention if you're running ethernet cables... they are shielded. I've never been in any hi RF environment... so I can't say that the cables aren't shielded enough.
(e.g. CAT 7 or CAT 8 as the latest spec.)

 

[Cbdb2]

Yeah.
Today... your high speed networks are going to use Fiber so no need for any RFI filtering when its optic. Your transceivers will be at the network switch converting back so you're already in shielded environment so no need for any RFI shielding.

So yes, this is a bit of snake oil even if it does what it claims to do. Its unnecessary.
Note: Within the rack you can see short runs using copper but most everything else is fiber.
Not to mention if you're running ethernet cables... they are shielded. I've never been in any hi RF environment... so I can't say that the cables aren't shielded enough.
(e.g. CAT 7 or CAT 8 as the latest spec.)

I was involved in installing Ethernet cable (mostly cat 6) for over 5 years. In the hundreds of installations we only used shielded cable once. 300' runs in industrial places and never had a RFI interference problem.

 

[dlaloum]

I was involved in installing Ethernet cable (mostly cat 6) for over 5 years. In the hundreds of installations we only used shielded cable once. 300' runs in industrial places and never had a RFI interference problem.

Depends on the type of industrial gear.... and sometimes it depends on its age (!)...

I've been involved in quite a number of industrial installations over the years using Cat5 up to Cat6a, none of it shielded - without any issues.

Then you get to somewhere where they have a 20year+ old arc welder..... and anything within 5 to10m of the welder needs to be shielded.