The DXE-ISO-PLUS filter thread for various network, cable experiments

[One and a half]

Following on from the EtherREGEN thread for various network, cable, power experiences and experiments, the topic of the DX Engineering filters got a bit crowded and veered off the main topic, so for @roman410 and @Superdad, here is the thread for dedicated application, experiments and findings.

 

Yes, the the DX-ISO-PLUS is a pair of filters using a back to back RJ45 Halo HFJ11-1G46E  with filters and caps, that were mainly intended for Ham radio operations to keep noise from Ethernet lines out of equipment. Each pair lists for USD49.99 and depending where they are in the chain to the audio network end point can make a difference to the noise reduction in that line. The DX-ISO-PLUS works very well with the EtherRegen, and is a complementary device.

 

So far known is that filters installed in the line direct from the router -> ER can gain the highest benefit right away, with filters added in between the router, a music server and an end point will also work very well.

 

The short cables supplied with the DX can be used as is, other users report that replacing these with standard plastic UTP cables is a subtle change for the better, it's a matter of experiment which one works well. Using STP anywhere in the chain was not good from my experience with the Etheregen, and the same is similar with the DX.

 

In any case, the DX-ISO-Plus is an inexpensive method, yet yields high value of return. 

 

As far as SQ is concerned:

 

Noise floor is lower, accenting smaller details

Soundstage: depth especially improves

Positioning of instruments and vocals is easily distinguished

All without tonality differences

 

[One and a half]

4 hours ago, sgb said:

Allthough DX Engeneering says it gives effective filtering for 160 through 10 meters.

160m = 1.873 MHz, 10m = 29.975 MHz, for the radio amateur important to know, for us in audio, frequency is the usual norm.

I would really like to know below 1.8MHz, what it does, in particular what attenuation one Halo has compared to well we know what four looks like. SMPS has is around 50-100kHz upwards of noise, must be something else we wouldn't hear it.

[One and a half]

10 hours ago, Egill23 said:

I have two packages on the way from an Austrian shop (about the same price but less hassle for me in the EU).  Since I run optical fiber the longest distince which is between the Sonore OM and the ER I suspect I might have less effect with them. I am thinking about putting them between the router and wall connector, between NAS (that has the music) and router or between router and OM. All those are just short distances around 0.5-1 m and I have cables with floating shielding there now, so we'll see if they're are of any use.

Once you have the filters delivered, installed and listened to, please send a link 🙂

[One and a half]

When the DX filters are fitted inline with the source, whether it from a switch or router to the audio end (start) point, is a good place to start.

With a few spare DX 10 pack left over, I added these to the router, so all three Ethernet cables in and out of the router have DX filters.

 

Cable Modem ----DX-------> Router in

                                                Router out 1 ----DX -----> JCAT NET FEMTO 1

                                                                                            JCAT NET FEMTO 2---DX-- 25m CAT5e ---> EtherRegen A | B---> Lumin U1

                                                 Router out 2 ---DX ------> Switch 24 Port

 

SQ wise I did not expect a big change, and there's a very slight change in tiny details, but will leave them installed for future streaming. service. In any case no backward change. All IT gear in this area is powered by a standby UPS it's output is connected to a 230V/120V Isolation transformer, in TN rules, since I cannot guarantee there is a two pole switch in the front end of the loads.

 

Notes:

The standby UPS is a Cyberpower 1500VA with their passive filtering in non UPS mode. The power from the street is not reliable, not that the voltage goes all over the place, it drops to zero, reestablishes, drops out again and messes up the server PCs. 

IT gear includes, the cable modem, switch, router, PC which houses the JCAT NET Femto, powered by a linear 5V 2A supply.

TN is where secondary neutral of the transformer (1000VA standard type toroid) is jumper to earth (ground). 

Router, Cable modem power from the same 12V open style SMPS. 

The theory is the noise from the IT gear stays in its loop back to the ISO transformer. The division of IT and audio gear is the EtherRegen. If the DX are only used, enough of them should filter the signal to the audio end point. Using the EtherRegen isolates the world of IT and audio which is important.

 

I would recommend that any IT infrastructure equipment is on a different supply than the audio system. Using a linear supply on IT gear, prevents noise placed back on the AC supply, inherently there's no isolation other than the small transformers in the RJ45 connectors. Using a DX confirms that the noise reduction of the small transformers in IT gear RJ45 is not that effective and they pass noise very well to the detriment of the signal which is audible.

 

In case the formatting messes up the spaces.

 

[One and a half]

6 hours ago, Mahler and Bach on Computer said:

Out of curiosity, I bought two pairs and the positive improvements are there, better base, high frequency lack of glare, all the important perspectives you want to improve.  However, I got pops noise here and there, like I am listening to LPs.  Given the positive side, I feel I can even live with those pops.  My digital chain is like following:

NAS-ASUS 5G Router-Netgear GS305 switch - Apple Mac Mini M1 - HQPlayer-SOtM SMS s200 - Mutec USB 3+ - Metrnum Pavan DAC.  Apple Mac Mini and SOtM s200 are linked through Bridge.  The two pair of DX Plus are used between M1 and SMs100, and between GS305 to M1.  Any ideas what can cause those pops.  

I hope the pops weren't there always.... 

Suggest to move one pair between the NAS and the router away from the output of the Mac mini and sms 100. Reason being there could be adverse reaction there.

[One and a half]

6 hours ago, Mahler and Bach on Computer said:

Good suggestions!  That is going to be my next step.  Unfortunately the pops were there always, in the frequency of once every 10 to 15 seconds.  The intensity is trivial but noticeable.  I checked the traffic of information, there was no sign of information spikes at all.  

Ticks are usually due to starving buffers in the music player. I haven't used HQPlayer for a while, but there are settings that can increase the buffer and should remove the ticks.

Can also be due to aggressive USB audiophile type cables, which have too much capacitance, that can also affect playback more severely, like dropout, or white noise.

[One and a half]

On 6/11/2021 at 1:44 AM, Mahler and Bach on Computer said:

Yes, the consideration of starving buffer certainly makes sense. Thank you.  I actually increased the buffer time from 20ms to 50ms in HQP settings.  The pops were completely eliminated.  I should say before I changed the buffer time, the pops situation had already been greatly mitigated over the time, could be the burn-in process as I noticed that,  every time I change the network configuration, even with minor variations, it takes time to stabilize everything.  
 

However, when I was adding the second pair of DX between the switch and the Mac Mini M1, the pops came back.  The pops did not go away even when I changed the buffer time from 50ms to 100ms.  Could be the burning-in again or some other unknown incompatibility.  Well, to me the most important part is between Mac Mini and SoTM 200, given that I transfer the current album to Mac Mini, and temporarily stored there before being processed by the HQPlayer.  I think the entire timing of digital playback should be largely dictated by the Mutec USB 3+.
 

The SQ improvement is quite obvious in my system.  

Wow, ok, the DX has affected the integrity on the input to the M1. I thought that possibly the buffer settings on the network adapter in the M1 could be adjusted, but the settings in my M1 nic also are only basic , the advanced settings are laughable. 

Surely should be possible to find the nic buffer settings, maybe Macos aficionados can help here.

For my  music server settings, the Nic buffers are wound high to 512kb for an Intel on a JCAT NET FEMTO,  but that server is under Windows LTSC, where settings are to hand. 

 

Good to see that the buffer settings were cleared by adjusting HQPlayer,  good work!

[One and a half]

5 hours ago, PYP said:

An experiment with the copper ethernet cable after the eR (and directly to the DAC) suggested that in my system this last leg is critical (significantly upgrading the ethernet cable made a large positive difference in SQ).    If that can be generalized (perhaps not), then one needs to be careful about what follows the eR (meaning, don't muck it up).  Just one system and one data point.  

I found that any STP Ethernet cables didn't work *anywhere* in the chain. The sound was very rough. What does work is to wire power supplies so that at the load and source connector's shells, a measurement of 0.000V between the + of the PSU is important. IF an STP is introduced, that near perfect zero starts to climb 😕

[One and a half]

On 10/17/2021 at 3:08 AM, PYP said:

   Also, the included pigtail DXE connectors didn't sound right to me and I don't use them.  I use short, inexpensive mono price slim line RJ45 and that works for me.  

Finally replaced the CAT7 pigtails supplied with the DXE with bog standard CAT5e cables a little shorter about 5cm (2") less. The original pigtails are a shielded design. Still getting used to the sound, but certainly is not worse.

 

With yesterday's network cable interruption, the EtherRegen is now playing up a day later, with the Lumin U1 displaying a network error. Just needs a power cycle to the ER and the Lumin and DAC off, wait a while, then power up the Lumin and the DAC in that order.

[One and a half]

2 hours ago, PYP said:

Don't know why, but I find that it takes a new CAT cable time to settle in, even if it has been used but out of the system for a while.  Am curious about how your experiment turns out.  

I really would like to invest in boutique cables, but the Wireworld, Nordost and some Audioquest are all shielded designs. Tried shielded even with a parallel ground wire and results are not good :(. Tried a few more sessions earlier on, promising!

[One and a half]

4 hours ago, PYP said:

doesn't Ghent audio utilize the design John suggested for copper ethernet?  My not be boutique, but always well built and uses quality materials.

Hmm, the Ethernet cables are also shielded from Ghent,  even the plugs are metal. 

[One and a half]

2 hours ago, Superdad said:

What matters is whether or not the shield is tied to the metal end shell at both ends.  One can not know just by looking at the cable. For example, the Audioquest cables, even though they have metal end shells and a shield, they only tie the shield at one end—and that’s good.

Taking a continuity test meter to the cable is the only way to know for sure. 

That's all and good, but am not prepared to spend USD450 on a cable that's shielded at both ends the hard way. I guess I have to nag manufacturers direct as none specify how the shields are connected. Audioquest has the direction arrow, only convenient when connecting to the EtherRegen,  so you know which cable is A or B.

[One and a half]

14 minutes ago, TwinPeak said:

I started experiment with the DXE-filters in July, and it was a pleasant surprise at first. Later, after the DXE-filters had settled inn, the synergetic effects levelled out. I bought two sets - one set was placed between internet modem and an Ubiquiti X-router. From the X-router I have a 7 meter stretch optical fibre to the audio network, starting with a Cisco 2960 (8-port). The other set of DXE-filters I tried at different places before EtherRegen - with almost identical result.

When I got my #2 EtherRegen and an OpticalModule (Deluxe), I removed all the DXE-filters in order to be able to hear the performance of two EtherRegens, uninterrupted by other "alien" components. And since then, the DXE-filters have been in the big drawer with audio spare parts - until last week.

 

The #2 EtherRegen used two month or so on fully "brake-inn", and the result was an unbelievable lift in sound quality, sound stage and musical enjoyment. Just for fun I found the DXE-filters from the drawer, and connected a set between Mac Mini and OpticalModule (see my signature). I have to mention that I had already removed the metal shield from both RJ45 plugs, as suggested by a forum member. It was easy to do. All LAN cables between the filters are high quality Ghent JSSG360.

 

 

I played some locally stored HD music files (24/192 and DoP) with and without the DXE-filters, and I noticed an increased richness in the upper frequencies without cluttering the instruments. A female voice, cymbals, a trompet and guitar - all played at high notes was easy to distinguish from each other, just as if there had been added some more space between instruments.

Well, the reintroduced DXE-filters will stay where they are, for now 😊

Tom

Cool, the metal is just clipped on, rather than crimped on. IIRC, the shield cable is tucked under that metal shell. Easy done 😃. The DXE work very well with the EtherRegen A side, but not so good on the B side.

[One and a half]

15 minutes ago, TwinPeak said:

I havn't heard the differences between shielded and unshielded, I just followed an advice. And yes, I have tried the DXE-filters directly onto ER A-side. Now I have an opticalModule in front of #1 ER 😇

Tom

For the major run of cable, say over 1m, shielded cable imparts a very rough sound, the distinction of mids and treble become blurred. Tried a few times, even running a cable in parallel to the CAT but no good. Might see about the JSSG from Ghent.

 

[One and a half]

@Michael Christ, if you buy the 4 pack, then it's possible to add the filters in the chain from the router onwards. Like @PYP mentioned, there is a difference between UTP and the STP cables supplied with the DX-Plus, with STP smearing the image in comparison. It is not necessary to use uber-priced cables as most of these are shielded, cause reverse improvement and are too long. 

Each system is different, depending on where the noise flows, but effective close to the music server and the endpoint.

 

You always need to use a pair of filters, in the same run, this is how the noise is reduced. Router —> dx —> cat6 —> dx —> server, is correct. One is to kill garbage from the source, eg router, the other one to filter noise due to induction in the UTP cable from other AC sources. I use CAT5e, and still obtain significant improvement, if you have CAT6 UTP, then please use it, no problem,

[One and a half]

3 hours ago, Michael Christ said:

Please let me ask a question to @One and a half: have you already made some experiences with DX in a hybride network combining optic fibers and usuall copper connections (Cat 5 or 6). If yes, where have you located the dx filters? Thx again!

Ok, the detailed config is in the AS profile, the link is in my signature, and Happy New Years!

[One and a half]

8 hours ago, Superdad said:

No, those “sockets” each contain multiple tiny transformer cores. Someone upthread the datasheet was posted for those “mag-jack” modules.

Yes, that's right each ethernet pair has a differential T network of inductance and capacitance.

[One and a half]

23 hours ago, WuBai said:

Just ordered two of these units and had a chance to open them up.  There was still a lot of uncleaned flux on my unit.  Also, I was expecting to see some sort of filter but instead what you see is what you get.  Strangely enough, it does work to bring forth a lot of missing detail and resolution, but for me I lost some musicality.  Meaning, the bass was tighter with the DXE but the notes being so tightly defined made them feel like distant and distinct little instances with zero total interaction.  All music felt extremely precise and each instrument and vocalist was its its own little bubble, like each person was recorded independently and then put together in post.  It didn't feel like multiple people in a room playing in harmony.  Took these units out, which again shouldn't make any difference because, well, just look at them, and now all of a sudden everything got softer and rounder and slightly blurrier.  This sounds like a bad thing, but actually the slight bleeding of the bass and the instruments gave everything a sense of togetherness and presence again.  I guess another way of putting it, since I'm already knee deep is that listening with the DXE is like being out at night with zero percent humidity in the air, clear as hell but also dry as hell.  I gave them a shot, but not for me.  I'll still keep them for further experimentation down the line, just thought I'd share my findings.  Thanks to the OP for starting this discussion, wouldn't have know about them otherwise and it was worth the purchase just to learn and experiment with.

 

Thanks for the impressions, can I ask where the DX were fitted in the chain? 

[One and a half]

1 hour ago, Rsbrsvp said:

Has anyone compared the DX to the Network Accoustics "muon" filter?

I compared the price, and that's about the end of the issue right there. 

 

[One and a half]

A couple of months ago, was working on a remote mining site, where the grounding techniques were not that great for the installed CAT6 cables relaying on communications. Our gear complained that the buffers weren't writing correctly, and only played up when a 2200kW VSD was running.

A week ago a colleague went to the same site and packed a pair of DX-ISO filters to see if they cleaned up the signal, sure enough, our equipment didn't complain one bit, as if there was no trouble to begin with. Cable length would have been around 40m in parallel with 400V power lines for quite a few metres, causing much of the drama. So the choice was to re-route the cables, with no guarantee that could fix, the DX filters just worked, simple and effective solution!

[One and a half]

On 11/30/2022 at 10:56 AM, Dev said:

I am interested in trying some but I am wondering if its necessary any benefit to use in pairs instead of just one for shorter runs, like 3ft ?

The DX-Iso are only sold in pairs, as the intended use is for 'longer' cables, to stop induction of noise along the cable and clamp source and destination noise.

 

You can experiment with just the one leave and listen for a week,  then add the other to compare. I use the pair for all cables even 3' since I can't guarantee that induction can be totally eliminated from the cable run.

[One and a half]

3 hours ago, Dev said:

 

Thanks. I actually ordered two pairs last week and will experiment with their placement. I run a 30ft utp Cat6a from the router to my audiograde switch and that segment will get a pair. The two other segments that probably needs attention is the connection from the cable modem to the router and the other being wifi bridge to the router but both of them are shorter runs. Will see how that goes.

I don't use wifi for audio purposes, other than the the iPad to control what to play, and the DX filters are not used on the feed to the Orbi router (used as an access point). I loaned my work the two pairs of DX filters, one of them was the cable modem to the main router, which is now a straight UTP. I wonder if that connection is messing up the Qobuz evaluation, cause the soundstage from Qobuz is badly hobbled.

The long length UP benefit from using the pairs very well.