Popular Post hols Posted November 17, 2019 Popular Post Share Posted November 17, 2019 My struggle with EtherRegen first 72 hours Some background on my system. I have been a 'loyal' Uptone supporter all along and I still am. I have in my possession one JS2, two LPS1.2, LPS1.0 and isoRegen and now two ER. Before using the ER in my system I have been using a SOtM switch (highest spec) and fed by Cybershaft OP20 as 10M reference and using Habst cryo'ed silver clock cable. I have been using the SOtM switch between my Doppio server (designed by my buddy) HQplayer embedded and the NAA and connected via USB to Denafrips Terminator. There is no problem with connection all along. We have tried to stack 2 SOtM switches connected by single mode optical and sounded excellent compared with only one. And we have been looking forward to the EtherRegen and see whether it can achieve that kind of improvement. I first used the Uptone SMPS and internal clock to replace my SOtM and I would say there is a very significant drop in wattage of the resulting SQ. There is shrinkage in soundstage, loss of detail and sound is dull. This is not unexpected because this is the first 15 minutes of connection and one cannot make any conclusion from that. I then powered off the ER and connected the 10M clock input pushed the button to external clock and then powered up again. And I immediately encountered stuttering which lasted for 10 minutes without improvement. I have to power off and then power up again and this time after stuttering for a few minutes it runs normally without dropout. The sound improves a lot compared to without 10M clock but to my ears the sound is not on par with the SOtM. My estimate is only about 50% to 70% of that of the SOtM. This is again not unexpected with a not yet boiled gear. I then changed the power supply to LPS1.2 and I would say there is not much of improvement with using the LPS1.2 and there was again some dropouts in the beginning. I then changed to JS2 giving it a 12V power supply. The sound appears a bit more solid but there is a slightly heavy mid range with JS2. That is about 1.5 hours after using the ER. My friend came over to join me at that juncture and I was playing a DGG Fournier & Gulda Beethoven Cello sonata. One thing that caught us both is that the piano comes off with excellent shape as if the Grand piano just stands up and you can hear all the delicate touch by Gulda. But at the same time the cello becomes shrunken. I guess this is due to a more quiet background and better delineation of the highs and SuperHighs compared with lows that contributes to this spatial delineation of the grand piano. This is a very different presentation compared with SOtM which gives a much richer timbre especially in the range of Cello. We then decided to try to stack the ER behind the SOtM and see whether we can get the best of both worlds(I have studied the manual a few times and know that the B side of the ER should always be closest to the DAC to prevent leakage currents from interfering). I connected through single mode fibre SFPs and when powered up we face stuttering again and this time it persisted for 10 minutes or so so I have to change it to RJ45 ports with ethernet cable. There is some scattered dropouts and it goes on normally afterwards. As for the sound there is a combination of the 2 characters, the balance between Cello and piano is more natural now. The piano is still excellent but there is a less holographic presentation compared with before. But I would say that the SQ still lacks behind that of 2 stacked SOtM switches. The ER is left powered on till next day When I played music through the next day there is not much of improvement compared with before which is again not unexpected because the boiling process need to continue a bit longer. But I encountered some stuttering again when I changed to other files. And I found out from this forum that I am not alone and John is trying his best to deal with it. At that time one of my buddies just communicated that he has no problem on the first day but on second day there is stuttering and then cannot even connect. Hope that John can find out solutions to that. After stabilizing a bit the music seems to be stable and I tried to experiment with my second ER. This time I tried to put it between my Asus 88 router and my Synology NAS using RJ45 ports and ethernet cable. There is stuttering again and this time the stuttering continues. But amidst the dropouts I can hear that there is an improvement in SQ. But I have to give up because after a while there is interrupted connection. Onto the third day today and I've got the biggest surprise which I didn't expect. The system is still on the SOtM +ER stack (without NAS ER)there is some improvement in the sound the orchestra sound more coherent with good layout and yet hearing more details. No stuttering. And I continue to explore by adding a ER between my NAS and router and voila suddenly the orchestra opens up and the widest and deepest soundstage I have ever achieved. And more important is that I can hear the details and ambiance and the air around the instruments are the best ever. The double bass are just rock solid in their place way back in the soundstage and yet giving out a detailed and musical sound echoing the other instruments of the orchestra. The layout is very natural and not the artificial pinpoint focus. One can just feel the rapport amongst the various instruments of the orchestra. As my friend commented it's like being the conductor of the orchestra. And fortunately there is no more dropouts today. And it seems that adding this second ER gives the best reward and is well worth its price. Now the SQ has surpassed the 2 stacked SOtM switches. Bravo Alex. So to conclude my first 72 hours impression with ER (1) Direct comparison with SOtM the ER may not have advantage and will need further boiling before any definitive conclusion (2) The sound character of ER is different from SOtM (3) The stacked SOtM with ER can give additive effect (4) Second ER between NAS and router can give big boost in SQ (5) The power supply to the ER needs further testing And in my system using two ER becomes a must. Having said that I am still confused about the cause of the dropouts and I shall raise the questions over in the installation thread. Thanks again for your great product Alex and John. Superdad, Maceear, rickca and 1 other 3 1 Link to comment
Popular Post hols Posted November 17, 2019 Popular Post Share Posted November 17, 2019 9 hours ago, Superdad said: Hi Hols: Thanks for your report. Just to be clear for others, Hols must have gotten his two EtherREGENs from one of our dealers as I promise we did not ship two units to any single direct buyer in the first batch. Certainly any stuttering is not related to use of external reference clock as we have others using this feature with no issue (and of course we tested ourselves). I agree with that and what I have been quoting is what happened when I tried the ER. And mind you that mys system has been running without any problem with the SOtM switch in place and it is quite logical to ask what went wrong. When using an external clock it is important to consider the power supply for the clock and where its 0-volt/-VE ("ground") is going and what it may be common to. That is because the BNC "ground" of the EtherREGEN's clock input is on the 'B' side of the moat. So depending on what else the clock power supply may be connected to--as well as what upstream gear other 10MHz outputs of that clock may be connected to--one can end up defeating the EtherREGEN's active differential isolation moat and losing some of its benefits. The point is after using the external clock the performance of the ER actually improves so I propose not to add the power supply of the Cybershaft clock into the picture. And as a matter of fact the power supply of the Cybershaft is not floating. Really any dropouts are not related to connectivity of the EtherREGEN. Such is generally a software and throughput issue. I myself experienced such with HQ Player Embedded feeding Roon when files were elsewhere on network, but that was only until I set my server to Gigabit. You mention using HQP Embedded as well, but I am not familiar with Doppio that you mention. Is that some type of player s/w? You may be able to experiment with some buffer settings. You can take it that my Doppio server is a Linux server with the OS and kernel compiled by my buddy and with HQPe installed and functioning normally as any other OS like Audiolinux and there is no stuttering before using the ER. Like most others using the HQplayer it is briged with the endpoint NAA. And it seems that bridging is one of the issues that John is asking. No the issue we are working on has only to do with connectivity and dropping of IP leases. John may have found the root cause, but it will take a couple of days until he can fully investigate. Based on what he told me tonight about what he found, I suppose there is a minor chance that the stuttering you got could be a side effect of the actual problem. I guess if IP connection was bouncing on and off or if there was some crazy excess negotiation traffic going on--such could cause some data delay issues. I am just speculating. I'd like to see a diagram of your network and what with the second EtherREGEN you were feeding from its 100Mbps 'B' side. Attached a photo of my network diagram. Well that's great! That is quite the turnaround from where you started. Wonderful! beautiful music and Superdad 2 Link to comment
hols Posted November 17, 2019 Share Posted November 17, 2019 5 hours ago, hieukm said: So @hols you are daisy chain ER and SOTM switch both clocked to Cybershaft OP20 and used between Router to server/render while another ER was in between the Router and NAS?? You are right that I daisy chained the SOtM switch and the ER both clocked to Cybershaft OP20 but it is used between the server HQPe and the endpoint NAA. Another ER was placed between Router and NAS without clocking because physical distance too far from Cybershaft. Link to comment
hols Posted November 20, 2019 Share Posted November 20, 2019 On 11/17/2019 at 9:31 AM, Superdad said: Hi Hols: Thank you for the diagram! A few questions and comments if I may.... Questions: 1) What is this connection with dashed line you mark as “Bridged”? 2) Why is your Linux server being used—with two Ethernet ports—to make the connection from your router to the SOtM and UpTone switch? 2) What type of port do you use for the connection between the SOtM switch and the ‘A’ side port of the EtherREGEN? 3) What power supplies are you using for the SOtM switch, the Cybershaft clock, and the EtherREGEN? If you are using a JS-2 for one of those, what if anything is connected to the other output of the JS-2? Comments: a) It seems very unwise to have your server with two Ethernet ports and having it act as the connection (for all downstream equipment) to your router and the NAS. b) With that second EtherREGEN between your NAS and your router—and with use of HQ Player Embedded (knowing what I do about how it draws data) you are creating a data bottleneck, as the 100Mbps ‘B’-side is not meant for that purpose. Plus, what really is the point to cleaning up from your NAS just to feed to your router? This application is really not what most people think of when they think of using more than one specialty switch for audio. c) You are most likely defeating the EtherREGEN’s active differential isolation moat by using the same clock for both it and your SOtM switch—as the BNC shell of the EtherREGEN is common to its ‘B’-side power domain. There is also a good chance you are defeating the isolation via the power supplies as well. Given all of the above I am not surprised that you sometimes get stuttering. And I also doubt that you are realizing the isolation moat benefits of the EtherREGEN. I suggest you make the following changes to your network configuration—all at the same time: a) Plug your SOtM switch directly into your router; b) Remove the Ethernet connection between your Linux server and your router, leaving it connected only to the SOtM switch; c) Plug your NAS directly into the SOtM switch, and set aside the second EtherREGEN for now; d) For the moment, to assure effect of the EtherREGEN’s ADIM, connect the Cybershaft clock to only either the SOtM switch or the EtherREGEN—but not both at the same time. e) Remove whatever that cable is—marked “bridge”—shown between the server and endpoint. You may have some reasons for your diagramed configuration, but I think it is far from ideal, not how we intend the EtherREGEN be used, and that you will acheive much better results using my suggestions above. We look forward to reading your answers to my first questions and also about what you hear when you make the recommended changes. Best, —Alex C. Hi Alex, It was a pity that you only announced a while ago that the connection difficulty that we faced is due to some firmware issue. I spent the past 2 days trying various methods but still cannot get the ER to sound properly. The stuttering just comes and goes but never goes away. Unlucky me -- futile 48 hours of work. Anyway I shall try to answer your questions and comments below. Thank you for all the comments and suggestions. First I must say that the network diagram I posted is not exactly right. My fault. I found it out just now. (I was in a hurry at that time when I made it ) Please see the amended network diagram attached which should help to answer some of questions. And my answer to your questions are listed below: 1) What is this connection with dashed line you mark as “Bridged”? It does not exist. My poor drawing made it wrong. 2) Why is your Linux server being used—with two Ethernet ports—to make the connection from your router to the SOtM and UpTone switch? By bridging the 2 ethernet ports it does not matter because the 2 Ethernet ports are regraded as one. And for SQ wise I have compared this configuration with your suggested configuration( Remove the Ethernet connection between your Linux server and your router, leaving it connected only to the SOtM switch) many times before and I find my present configuration gives a better sound. 2) What type of port do you use for the connection between the SOtM switch and the ‘A’ side port of the EtherREGEN? RJ 45 copper. I have tried to use single mode fibre SFP and it stutters then so I changed to use copper RJ45. As I have described before it also stutters for a while and then improves so RJ 45 is used. 3) What power supplies are you using for the SOtM switch, the Cybershaft clock, and the EtherREGEN? If you are using a JS-2 for one of those, what if anything is connected to the other output of the JS-2? The SOtM switch is supplied by Uptone LPS1.2 energized by Paul Pang LPS. Cybershaft comes with its own Linear power supply not shared with any others. ER is supplied by JS2. The other output of JS 2 is vacant. a) It seems very unwise to have your server with two Ethernet ports and having it act as the connection (for all downstream equipment) to your router and the NAS. already explained above because the 2 ports are bridged and SQ comparison is the best with this config b) With that second EtherREGEN between your NAS and your router—and with use of HQ Player Embedded (knowing what I do about how it draws data) you are creating a data bottleneck, as the 100Mbps ‘B’-side is not meant for that purpose. Plus, what really is the point to cleaning up from your NAS just to feed to your router? This application is really not what most people think of when they think of using more than one specialty switch for audio. I fully agree with you there is no theoretical basis for putting it there especially with the B side to the NAS. The reason I did this initially is because my NAS is located over another corner of my listening room with the modem & router. And my previous impression of using NAS is always not as good as local files in hard disc located in my server. And I have made many attempts to improve the sound from NAS and that is why I would experiment with the ER positioned there. Actually I have also tried connecting it to the A side but the SQ is not as good. And when it is connected to B side the sound becomes excellent. And the surprising question is why suddenly the sound is so good that night. Actually I have tried to do it again it stutters like mad. My best guess is that (1) I used a $1500 Ethernet cable(sorry I don't mean to introduce expensive ethernet cables into the picture, but it was what happened that night) (2) Suddenly there was no stuttering that night and everything runs smooth. I just paused the HQplayer and added the ER and music just goes on and me and my friend can both got the great improvement in SQ. I failed the next day probably because I need to start the connections again and the stuttering prevented me from doing anything more. (3) A lot of AS members has expressed that 100Mbps gives better sound than 1000. Could it be that's the reason? Or now that we understand it is the EEE problem then maybe the reversed connection does not trigger the EEE. Purely guessing. c) You are most likely defeating the EtherREGEN’s active differential isolation moat by using the same clock for both it and your SOtM switch—as the BNC shell of the EtherREGEN is common to its ‘B’-side power domain. There is also a good chance you are defeating the isolation via the power supplies as well. I understand that this is your main concern of not getting the benefit of the ER ADIM and also afraid that the clock connection would pollute your B side. I have checked my Cybershaft clock output receptacles with my multimeter. The 3 outputs are all independent of each other and also not connected to the chassis of the Cybershaft . There is no connection between the shield/ground of the BNC cable of one output to the other outputs. So this makes your worry not likely. As for the power pollution we have explained earlier there is no other gears connected to the JS2. I suggest you make the following changes to your network configuration—all at the same time: a) Plug your SOtM switch directly into your router; b) Remove the Ethernet connection between your Linux server and your router, leaving it connected only to the SOtM switch; c) Plug your NAS directly into the SOtM switch, and set aside the second EtherREGEN for now; d) For the moment, to assure effect of the EtherREGEN’s ADIM, connect the Cybershaft clock to only either the SOtM switch or the EtherREGEN—but not both at the same time. e) Remove whatever that cable is—marked “bridge”—shown between the server and endpoint. I have moved my NAS to near my server and tried the above suggestion( which is basically one of the configurations I have tried months before). I have taken off the Cybershaft clock to both switches to make things more simple And there is no physical bridge between server and endpoint. And the result is all stuttering once the ER is added. There is no way one could hear anything meaningful. I have also tried to use the ER alone using all your standard recommendations including Uptone switching supply, no clock, but there is still stuttering. I have also tried to reduce the network speed to 100Mbps and the result is the same. And then this morning when I woke up I found your announcement that we need to do a firmware update. I guess the number is more than the 6-8 you mentioned. there were already 3 ER amongst me and my friends that has stuttering going on. So anticipating for the firmware update this weekend so we can have another go for better SQ. Cheers. Link to comment
Recommended Posts
Create an account or sign in to comment
You need to be a member in order to leave a comment
Create an account
Sign up for a new account in our community. It's easy!
Register a new accountSign in
Already have an account? Sign in here.
Sign In Now