|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
104.238.169.94
In Reply to: RE: From another thread... posted by Mercman on January 26, 2017 at 03:23:25
No. Just think. What you 'optimise' is not optimisation if small changes in your system brings either better or worse SQ.
Follow Ups:
Just think.Good idea.
What you 'optimise' is not optimisation if small changes in your system brings either better or worse SQ.
False argument - see above. "Optimising" in the PC audio context means configuring an OS to favour audio reproduction, typically by disabling non-audio-related functions. It may, as you argue, be a misuse of the term but the rest of us know what it means even if I prefer to say "slimming". Whatever, a carefully configured OS is more likely to be responsive to minor changes in the overall setup, not less.
In passing and at risk of saying the obvious, the microRendu has not been "optimised", just not bloated in the first place.
Soundchekk is right to say: "Keep mains separated from DC and signal cables; keep HF stuff . . . separated from the rest; review your grounding scheme". And he was right many years ago when he recomended a decentish PSU for the switch that connects to the PC though it was only a start.
Soooo, today's tip?
If you drive an audio PC via ethernet, try running it at 100-BaseT or (better) 10-BaseT. That's an old tip now but the new one (at least I've not seen it anywhere else) is to use LAN cables that connect the green and orange pairs only (pins 1&2, 3&6). Omit or remove 4&5 and 7&8 as 10/100-BaseT doesn't need them. Best use Telegartner plugs and a decent stock LAN cable such as Excel (see link) or, probably, BJ. No need IMHO for boutique products, not even the well-documented Meicord cable. I bought two of those but they're no longer in use. Ditto ViaBlue CAT7, which are as good as Excel CAT6 but much pricier.
I don't know why but the difference in my system made by simply removing the unused pairs was very marked even though I'm already using a fibre-optic link to the server with the local media converter and other gubbins powered by an Uptone "supercapacitor" PSU and separated from the PC by LAN isolators. Viz, the biz. The sound is way better than that from ten-times-the-price Meicord cables and in a different league to Amazonic stock crap. Yup, I've tried them all. I'd wager that it compares well with Audioquest's mega-buck audio bling but that's beyond my reach.
And, yes, my audio PC was "optimised" / "slimmed" till its eyes watered.
HTH
D
Edits: 01/26/17
Hi there.
To me at least it's not new that cutting certain pairs made a difference.
I think we discussed it around 5 years back.
I recently cut a broken rather long Meicord into pieces and terminated these also with Telegaernter MP8s. The MP8s are quite nice.
I didn't run any ethernet cable comparisons lately.
***
And nowadays I run GBIT LAN on my PI3. Actually I achieve 300MBit/s with a 10Euro USB3 ethernet dongle which comes with potentially better ethernet chip. The handling became much snappier and my files are loaded much faster into RAM.
I also applied an idle sitting AQ Jitterbug to that ethernet dongle for having some filtering in place.
And. Since I had some of Sbooster filters around, I hooked these up
to my Cisco ethernet-hub which sits right next to my Pi and to my FritzBox power supplies on the other end.
All that sounds extremely clean to me.
I'm wondering if LAN isolators are worth it -- if you go the UTP route.
I heard stories that people didn't notice any differences. That's why I left it alone.
I didn't jump on the fiber train, since the known devices have no hubs at the end.
All that - A never ending story.
Enjoy.
To me at least it's not new that cutting certain pairs made a difference.
Interesting - thanks for the feedback and for the RJ45> USB idea.
I didn't do any structured LAN cable comparisons, just played about with what I had to hand and pinched other folks' ideas. The Telegartner tip came from the CA forum; I genuniely thought the snippie bit was my idea but maybe I saw it somewhere. Who cares? It seems to work.
I'm wondering if LAN isolators are worth it if you go the UTP route. I heard stories that people didn't notice any differences.
I certainly hear the difference even with vasectomied LAN cables and an LPS-1. I use two, one either side of a pair of hubs sited after the media converter. I'd love to use them elsewhere but the loss of SQ is immediately apparent. H/W Quickie Piccie of the setup (moved to a desk, not on its rack).
I didn't jump on the fiber train, since the known devices have no hubs . . .
For me, the sound is definitely better with the fibre link, Zytek hubs and EMO isolators though the latter probably make less difference than the hubs. The LPS-1 was a big step up from linear PSUs.
If you were to say "you're mad", I'd probably agree. If you were then to say "there must be an easier way to do it", I'd definitely agree but, so far, I've not found it.
In any case, there comes a point where you write off the cost and effort and just play music while reminding oneself that the setup cost less than many fancy LAN cables and would, I suspect, do well in comparison.
Just wait till I've optimised it . . .
D
From the inmate who 'doesn't' talk optimisation!
Your cable looks like mine. I did put a nice black jacket on to make it look serious, audiophile and - most important - pricey. ;)I do find your DC wires quite a bit "exposed".
And you just use the 100MBIT fiber. OhhKaay.I do agree. The HUB itself makes quite a difference.
I mean a HUB "reclocks, regenerates, rebalances, refilters, rebuffers, resignals, re...whatsoever" the stream.
(Thorsten are u listening??? The world needs an iNurifier (N for network) ;) )****
These EMU filters just cut the ground/shield. That you'll have with a UTP cable. Or u just don't connect the shield to the MP8.
And the EMU will cope with surges/spikes. What more at 129$?As explained on the Meicord page. The jacks/plugs do cause trouble on their own. You introduce two more of them with the filter in place. Hmmh.
Nope. I won't even try one of these devices.
****
I'm not a fan of the LPS-1 for several reasons.
1. It's much too weak.
2. It doesn't really act like a battery with the LDO at its output.
3. I do also consider that LDO overkill inside a pre-regulation stage
4. People don't actually know what's causing the experienced improvements!
Sometimes just reviewing and rearranging your grounding scheme does more then a DAC upgrade or a new PS.
5. I really like your heatsink on top of it.
6. Let's not talk about the price.
Enjoy.
Edits: 01/31/17
I did put a nice black jacket on to make it look serious, audiophile and - most important - pricey. ;).
Ah. I removed a nice lilac-coloured jacket from the smaller leads so I could make shorter connections with tighter bends. The audio-macho look came with. NB that the pairs are slightly separated and cross at right angles. You could be right about the DC wiring but then you didn't see the first hookup . . . what would you suggest?
The 10/100MBit media converters are half the price of faster ones. The Zytek switches (as recommended on a CA thread) are also 10/100MBit only and do actually sound slightly better than the little gigabit Netgear boxes I used while experimenting. They cost slightly more so they must be better.
I agree that the LPS-1 is expensive given its low output ( < 1.2 amps) but supercapacitors are an emerging technology at least in audio. They are IMHO a significant breakthrough whatever their weaknesses.
Sometimes just reviewing and rearranging your grounding scheme does more than a DAC upgrade or a new PS.
The setup shown in the pic (and, of course, the other end) took me several months to get 'right' but was such an improvement sonically over the Netgear/linear PS as suggested on your blog a few years back that I still can't quite believe it.
Yes, easily as good as a new DAC. Moving data from a remote server over optic fibre can work very well but what I learned was that you do have to do it right. Skip the regen razzamataz and you are likely to be disappointed.
++++
These EMU filters just cut the ground/shield. That you'll have with a UTP cable. Or u just don't connect the shield to the MP8. And the EMU will cope with surges/spikes.
Not so - see datasheet. They decouple data pairs with transformers as well as cut any shield. They were designed for the medical sector where lawyers seem to like protecting patients from the mains. The Baaske may also use transformers - I don't know - but the EMOs definitely do. Whether they're worth the money is a fair question (I bought mine well before I got into the fibre thing) but I know that the difference they make is readily audible even in UTP links.
As explained on the Meicord page. The jacks/plugs do cause trouble on their own. You introduce two more of them with the filter in place
I'm not sure that that's what the Meicord papers say. See also the BJ link. Both suggest that few LAN cables come even close to spec and that the biggest problem is typically the plugs. They don't seem to claim that connectors inevitably degrade the signal to below spec, only that designing quality plugs is not trivial. Obviously, any plug degrades to a degree but the suggestion that they do so to an extent that inevitably makes in-line filters useless is not supported by data or, in my case, experience.
I tried a couple of the Telegartner connectors with Excel cable and found they sounded as good as the Meicord products or, post-vasectomy, better. Both are way better than Amazonic stock but a one-metre Telegartner/Excel cable still only costs about $10, the Meicord about ten times that.
BTW, AQ LAN cables use Telegartner plugs with, I suspect, a small markup. IME, no need for screened Telegartner plugs. Both types work fine on UTP but the screened, though slightly more robust, are harder to rework.
We compared supercaps with other quality caps.
I can't confirm that these are superior in general.
It depends on the application.
They IMO work better without that LDO behind it.
Then you also won't suffer power limitations.
Each ethernet jack has a decoupling transformer inside!!
They are all isolated.
We compared supercaps with other quality caps.
"We" being? Any details, links?
Each ethernet jack has a decoupling transformer inside!!
Is that right? And there was me thinking that the regs for medical equipment might just be a bit more stringent than those for consumer kit. Silly me.
So the differences I hear are all in my imagination? Funny old stuff, psychology.
D
We did some tests at a friends place.
"So the differences I hear are all in my imagination"
Not at all. You change the physics. You change the sound.
'"Optimising" in the PC audio context means configuring an OS to favour audio reproduction, typically by disabling non-audio-related functions'
I don't know how this type of statement can be made with any degree of authority, when nobody can determine exactly what to disable to achieve what you said. If you go back to CMP (your favourite it seems), disabling certain processes or altering them can make a system sound worse. The whole thing depends on what system is used and what hardware and software it is loaded with.
There are competing software suites that are claimed to optimise audio replay. The fact that they all sound different and the fact that the OS used has an effect mean that there is no optimisation of audio playback. The fact that you can make changes to a USB or LAN cable to affect SQ materially is another obvious argument against so called 'optimised' systems that some claim or imply that are superior to those used by others.
I am not going to argue with you about Dictionary definitions. I am simply saying that it makes no sense for anyone to claim that he/she has an optimised system that is immune to even small hardware, software and setup changes.
"I am simply saying that it makes no sense for anyone to claim that he/she has an optimised system that is immune to even small hardware, software and setup changes."
I agree with this. Computer optimization or slimming of the OS does not make the system immune to other changes. I honestly don't remember ever implying this for myself.
I don't think of you as a claimant, but you have used the term loosely and so can imply it.
However, others have made claims that point to the 'superiority' of their 'optimised' systems which contain obvious non-optimised features.
I have raised these points repeatedly because the claims do mislead inmates and more recent computer audio adaopters.
nt
Mac (with Audirvana + ARC2 Acoustic Room Correction plugin) -> Cat 6 cable -> Merging Hapi -> AES/EBU XLR Vovox SD -> PMC TwoTwo.8
Indeed: very marked SQ improvement!Thanks for the feedback. Are you using decent cable/plugs? Have you done every possible cable? If not, you may well have further improvements in store. Meanwhile, as promised, comments on your earlier post.
It highlights a 'hidden' problem that I also had in my system
WRT the Juergen Reis interview, I'm not sure you're right to suggest that "these 'back door leakage currents' are also better known as: ground loops". My take was that by "back door noise" he meant how different PSUs in the same system tend to interact with each other in ways that the spec sheets don't (can't) describe even when properly configured that nevertheless degrade the sound. Ground loop issues are well understood even if tackling them can - as you've found - be time consuming.
John Swenson makes a similar point on the Uptone Audio web site - see link. He suggests that:
There are two methods by which a power supply interacts with the AC mains and the difference is not well understood by most people (even many engineers). The solutions to deal with them are very different but get all jumbled up in most peoples minds [snip]:1) noise between the hot and neutral (and maybe ground wire)
2) leakage current between the AC line and the DC output of a PS#1 is pretty straight forward and what most people think of when they talk about "noise injected into the mains". There are many types of filters, power conditioners etc that deal with this.
#2 seems to be much harder to understand. It is created by the power supply, some of the AC voltage on the mains sneaks through parasitic capacitances in the PS and winds up on the DC output. The strange thing here is that it is between the the AC line as a whole (hot, neutral, GND) and the DC output as a whole (+, -). It does NOT exist between hot and neutral or between the DC - and +. It is between the two groups.
Because of this any type of normal AC line filter or DC filter or regulator is completely useless at combating this. All these devices work by "filtering" the noise between hot and neutral or between + and -. But this type of noise does not exist that way.
Later, he advises folk to take any filters for "type 1" noise out of their systems and instead connect everything via a multiplug to a single mains socket. I can't comment on how well that works but I can say that the Swenson-designed LPS-1 "Supercap" PSU which I'm currently using to power the DAC end of the fibre optic (and gubbins) link between my system's audio PC and its server transforms (sorry) the SQ of my system.
BTW, every other box in my system except the amp connects to the mains via a Jon Risch-style DIY mains isolator (i.e. the audio PC, an Intona, the USB> I2S board, the DAC and the DAC's I/V section, five devices (ten transformers and 15 coils . . .) all told. A transformer good enough to isolate a power amp is too expensive. I don't know if the isolators are dealing with "back door noise", "Type 2 noise" or just noisy neighbours but I do know that if I take them out there is a dramatic drop in sound quality. I'm hoping soon to get a second LPS-1 for the PC.
I feel pretty sure it's not ground loops I've cured as discussed in the Jim Brown or Bill Whitlock papers as I have no connections between system ground and safety ground. (There are safer ways of keeping them separate BTW so don't take this as a tip even if I got it from a Charles Hansen post a while back.)
Finally, though removing redundant pairs from my LAN cables also makes - as you've confirmed - for a jump in SQ, it is definitely not because it fixes a ground loop issue. LANs use differential pairs that have no ground; LAN sockets always come with an isolating transformer and I have in any case two in-line medical-grade isolators not to mention that pesky ground-free LPS-1 PSU. My guess would be that removing the wires (and running at 10-BaseT) does wonders for Near-End Crosstalk on the spare pairs. But it's a guess. What's not a guess is that it was an effective tweak.
HTH
Dave
Edits: 01/27/17
Hi Ryelands,
Friday evening I hurried of too a friend too show him your LAN tweak. And also in his system it gave a very marked SQ improvement. Even bigger as in my system.
I use these brandless OEM connectors
EAN productcode 8716065132205
See:
https://intronics.nl/en/product/keystone-and-modulaire-connectoren/modular-connectors/intronics-oem-rj45-8p-8c-modulaire-connector-for-round-cable-with-solid-conductors-td108m-
I buy these and the UTP-cable at https://www.netwerkwinkel.com
I have not done every possible cat 5e cable. For the experiment I only modified a 1,5 meter UTP cat 5e cable that sits between the Baaske LAN cable isolator and the MAC mini build-inn LAN socket. I only modified the RJ45 connecter that connects too the MAC mini LAN-socket.
Between the Merging Hapi LAN socket and the Thunderbolt LAN Adapter a LAN cat 6 cable is used that came with the Merging Hapi. I did not modify connectors on this cable.
From the Baaske LAN isolator there is about 20 meter of UTP cat 5e cable traveling through floors, walls and ceilings too the breaker panel closet. In the breaker panel closet I keep the rest of my internet equipment (cable modem/router/switch box, managed Netgear Prosafe switch, Synology DS216 NAS, D-link DIR 868-L WiFi-access point, Philips Hue bridge).
At what more locations in your wired LAN network did you also changed the connections in the RJ45 plugs?
My be I put aside Juergen Reis too quickly as just another manufacturer of highly overpriced 'High Futility Audiophool' gear. He after all also said (16:11 min) "when every thing is connected balanced than the influence of the cable is really shrinking", "too a minimum". Most HiFi manufacturers would rather die first, than admitting the need of such basic, best practice.
I looked again at the video and Juergen Reis points at 'capacitive coupling' as being part of the loop. Capacitive coupling bridges the gap and creates/closes a loop. But what exactly is the noise source (driving force), stills stays unclear too me. Same for John Swenson's text. I do believe him on his word that "some of the AC voltage on the mains sneaks through parasitic capacitances in the PS and winds up on the DC output". I wished John had taken more time not only too point at that there is a the leakage pathway from AC-side too DC-side in the PSU, but that John also would have explained how the source (driving force) is created and how the noise loop is closed (in most systems).
The above mechanism may be the same as on Page 16 in the PDF with Ugly showed too us. Just speculating.
http://www.hottconsultants.com/pdf_files/APEC-2002.pdf
"and instead connect everything via a multiplug to a single mains socket"
This is the oldest remedy I know. This was how I became aware of having a huge ground loop in my system. I temporary connected my system via a multiplug in a single mains socket because my living room needed fresh painting. This unmistakably sounded better. Which made me realize, I had a ground loop in my original setup for years.
Would you consider posting this effective tweak as a separate post?
Your effective tweak is now hidden as an off-topic subject in Mercman's post.
Not many inmates will find this magnificent tweak here in this post !
Mark
Mac (with Audirvana + ARC2 Acoustic Room Correction plugin) -> Cat 6 cable -> Merging Hapi -> AES/EBU XLR Vovox SD -> PMC TwoTwo.8
Dave has done something that deserves its own post.
that one cannot 'optimise' a system through conjecture. When optimising in the technical sense, the cause and effect of each variable is generally understood and optimisation involves the give-and-take in terms of achieving a final result.
When you talk about OS optimisation, the precise effects of many elements of the OS on audio replay quality are not known and so called 'optimisation' is based on conjecture, often speculative ones at that.
So if you say that your system is 'optimised', then you would have had to understand and install power supplies that are 'optimal'. Otherwise, if you make a small change to these and get a material change in SQ, what does 'optimisation' mean? Similarly you LAN cable and other peripherals.
Power supply interactions are complex and using two transformers back to back to provide optimal isolation' is conjecture. You are altering the source impedance significantly and it can be that you are hearing an improvement because something else in your system is 'not right'. As long as you accept that this outcome is fine (optimal for you) and do not pontificate on pseudo technical reasoning for others to follow, then this is fine by me.
As an aside, I know for a fact that, apart from filtering and current leakages, the effect of mains frequency can cause strong changes in SQ (my preferred regenerator frequency into my dac being 73 Hz ???????). As I said above, I simply have to accept that this is so for my system without dwelling into the technology and pretending that I know better than others.
As for isolation transformers, there is no substitute for spending the necessary money to buy units deigned as such and if possible have technical data to back them up.
There are transformer manufacturers who know what they are doing and can supply units to requirement at reasonable cost.
that one cannot 'optimise' a system through conjecture.
Some copy missing?
When optimising in the technical sense, the cause and effect of each variable is generally understood and optimisation involves the give-and-take in terms of achieving a final result.
Sounds good but not sure it's relevant here.
When you talk about OS optimisation,
I don't talk about OS optimisation - see above.
the precise effects of many elements of the OS on audio replay quality are not known and so called 'optimisation' is based on conjecture, often speculative ones at that.
Nonsense. Many OS functions have no role in the audio replay chain. Though it can be tedious, disabling or deleting them is generally trivial and has been done to varying degrees across the audio sector for over a decade. Something tells me you've not tried it even if you do experiment with this and that.
So if you say that your system is 'optimised'
Sigh. I don't use the word 'optimise' but neither do I feel the need repeatedly to belittle those who do with gobbledegook about power supplies.
then you would have had to understand and install power supplies that are 'optimal'. Otherwise, if you make a small change to these and get a material change in SQ, what does 'optimisation' mean? Similarly you LAN cable and other peripherals.
Again, nonsense. If, say, I disable enumeration in the USB stack, I definitely lose flexibility and may or may not get better SQ but PS quality has nothing to do with it. If you want to call snipping a couple of pairs from a LAN cable 'optimising' it, who am I to stop you?
Power supply interactions are complex and using two transformers back to back to provide optimal isolation' is conjecture.
You've changed the subject and accused me of saying "optimal'. Again.
You are altering the source impedance significantly and it can be that you are hearing an improvement because something else in your system is 'not right'. As long as you accept that this outcome is fine (optimal for you) and do not pontificate on pseudo technical reasoning for others to follow, then this is fine by me.
Now you're the one who's guessing. Of course the setup raises source impedance which is why one and all stress that it isn't suitable for dynamic loads. We're discussing powering devices that present a low and more-or-less constant load well within the transformers' power rating. You seem also to miss the point that we're talking about isolation and filtration with an eye on leakage currents.
As I said above, I simply have to accept that this is so for my system without dwelling into the technology and pretending that I know better than others.
But almost every post you make has you pretending you know better than others though not always convincingly.
As for isolation transformers, there is no substitute for spending the necessary money to buy units deigned as such and if possible have technical data to back them up. There are transformer manufacturers who know what they are doing and can supply units to requirement at reasonable cost.
I know of a couple of UK manufacrurers who sell reportedly excellent audio-grade isolation transformers but they are not cheap. They're too expensive for me and, in any case, overkill for circuits drawing ten watts or less. Among the "stuff" I tried were ordinary isolation transformers from reputable suppliers but they didn't impress.
''I don't talk about OS optimisation'' Quote 1 from you
''"Optimising" in the PC audio context means configuring an OS to favour audio reproduction, typically by disabling non-audio-related function.''
Quote 2 - from you http://www.audioasylum.com/forums/pcaudio/messages/16/162783.html
So, what are you on about?
It is no good quoting bits of what I said and responding in bits and pieces. You need to make cogent arguments if you wish to discuss. After all, you invited criticisms of your post and now it seems that you cannot take comments seriously.
So, what are you on about?
You have a bee in your bonnet about and sounded off on the word 'optimise' so I explained the sense in which other people use it in our context even though I don't use it myself. I don't see what's difficult about that.
The link you provide with that conjuror's flourish confirms that. I wrote: "It may, as you argue, be a misuse of the term but the rest of us know what it means even if I prefer to say "slimming".
You need to make cogent arguments if you wish to discuss.
You miss the point - I really, really don't wish to discuss with you but, as you were talking nonsense about 'optimisation' and more nonsense about Jon's little circuit, I felt obliged to correct the record.
I should've let it go . . .
D
Ho! You mean you with YOUR definition.
...but you appear to have a comprehension issue, again.
issue you have invented again.
"I feel pretty sure it's not ground loops I've cured as discussed in the Jim Brown or Bill Whitlock papers as I have no connections between system ground and safety ground."
If I remember correctly the Risch "isolators" are RC filters, right? RC's are for "type 1", aka differential mode, problems.
"Type 2" noise is a subset of the more broad set of problems known as common mode noise.
It's pretty safe to say that in general solutions for differential problems are almost completely useless for common mode issues and vise versa. All systems will have a bit of both differential and common mode noise present pretty much throughout the system. It's only a matter of how much and is it a problem? Many times the most effective solution is a bit of both differential and common mode filtering.
There is only so much that even the best power supply designer can do in design and layout to prevent emi. Mother nature disallows perfection from an emi standpoint. Unfortunately, in some cases, filters are the best option. When all that can be done in these areas has been, you are left with filters to get you the rest of the way. Filters are a necessary evil.
Henry Ott is considered one of the old masters when it comes to understanding EMC issues. Slide 16, in the link about common mode problems from power supplies, gives a visual for the power supply leakage current "ground loop" path from energy transferred via transformer inter winding capacitance and via parasitic capacitance to nearby conductive objects.
As Ott points out, in the slides but also in his several excellent books on the subject, leakage current generally responsible for causing conducted problems. Conducted problems can be handled with filters.
Though an important point to note, by my estimation, is that conducted can easily convert to radiated and vise versa under the proper conditions so getting a filter in just the right location is critical to prevent high frequency conducted problems from converting to radiated energy.
Diff mode problems are easy, you can easily see them with any oscilloscope. With common mode, not so much.
The hard part is figuring out if you have common mode problems big enough to worry about and all it's mechanisms and nature of being. With that info dealing with it is a snap. However, to set yourself up with the necessary gear and knowledge to do well at it is beyond practical for most.
Avoiding problem children at the time of new power supply purchase seems logical to me. But it's not as if the power supply companies I'm aware of are making it easy by supplying emi scan captures or anything useful like that. It's annoying.
After that, most of us are pretty much left trying stuff.
Getting into designing and having success with getting the audio system to sound better via common mode filter application is gonna be tricky for a noob. Getting the current handling capacity, as in power amp filter for example, is easy as getting enough wire gauge for the current. It's getting the filter tuned to and getting enough attenuation at the frequency of noise problem you have, putting the filters in the right spots in the proper orientation is where all the action is.
Since it's tough for us mortals to easily take a look at the problem it's about like shooting in the dark when trying to get the right filter put into the right place.
Some educated guesses are no brainers though in my opinion. For example it's almost guaranteed a PC power supply, even if covered in emi certs, will be injecting some amount of stuff back into the power line as well as onto its other cables. Playing with blocking noise from conducting out of the various cables emanating froma typical PC isn't too risky in my opinion. In fact it seems very popular on this forum now days. Your mileage may vary.
wow guess mine went a bit long too.
Many times the most effective solution is a bit of both differential and common mode filtering
Thanks for an interesting post and links. I admit that I don't get how the Type 1 / Type 2 distinction or the notion of "back-door noise" adds to what is already well known. It's been several years since I looked at Henry Ott's paper. Who knows? I might get more from it this time . . .
Whatever, JR used isolating transformers because they allow high capacitance across live and neutral without excessive leakage currents (see link). He specified split-bobbin transformers because they were easy to get for a "cheap and cheerful" project and for their low inter-winding capacitance. In practice, they get pretty hot and tend to buzz. Because I had suitable toroids to hand, I used them instead, fitting those with inter-winding screens on the input side. (Don't know why but grounding the screens on both transformers regularly trips the ELCB.)
As you suggest, designing and measuring decent filters seems beyond many audio designers, let alone us hapless DIYers. Once I grasped the importance of clean power for audio, I did much as you said and "tried stuff". After lots of "stuff", much of it discarded, my rules of thumb became:
1. A dedicated power line for digital kit and another for analogue. With care and a bit of luck, both will have lower impedances than the typical home supply (esp with UK-style "ring" circuits) and be relatively free of crud.
2. No cheapo SMPS PSUs on the above. If you need one for e.g. a data server, run it from a separate line and, if possible, connect via the latest fad, optic fibre. (It is a fad but a very effective one if done right.)
3. Instead of JR's high values of C across L&N, fit C-L-C filters between the transformers. I copied a commercial mains-rated design several cuts above those IEC socket things. By using one isolator per line-level device (~15 watts each), no real need for large coils.
Hopefully, unlike JR's circuit, I'm addressing both noise modes and, by using separate isolators & filters for each device, helping to prevent noise from one device affecting another. I don't see how a simple connecting strip or system-wide filter can do that. At least in my system, the difference that several discrete isolators makes is very apparent even though I'm using decent PSUs.
Feel free to criticise.
D
It sounds like you are already using a tuned 2 stage line filter with common mode filter first stage and differential mode filter second stage. That sounds like a really good starting place for trying out in various locations.Not that you didn't already think of it or maybe already even doing this but since you didn't mention it a couple potential tweaks for tuning the common mode performance of the first stage come to mind.
Maybe running the CLC in a balanced differential CLC configuration with respect to live and neutral will help enhance common mode rejection performance of the first stage without sacrificing the second stages differential mode performance. Ie you'd still have the same X caps spanning line and neutral but the balanced configuration would have an a separate L (but half the L as compared to an equivalent single ended version of the filter) between the caps on both the line and neutral.
You could also try adding, in a balanced manner, 2 identical and appropriately rated Y caps. One from from live to earth and one from neutral to earth. I'd try variations on where they belong, ie on the transformer primaries, transformer secondaries, after the CLC??? Some where around around 1nF ought not put you up against code violation most locales and is probably a fair place to start experimenting. Adding the Y's like this is to try and enhance common mode noise rejection of the first stage.
The only other thing I might add is that remembering to swap the order of the stages with respect to source and load to see which way works better during experimentation is another thing that may very well be worthwhile.
Edit: maybe there is a use for a damping resistor from line to neutral after the CLC to enhance it's performance.
Edits: 01/28/17 01/28/17
You could also try adding, in a balanced manner, 2 identical and appropriately rated Y caps
Thanks again for your reply. I did ponder doing as you suggest but decided I was perhaps making things a bit over-complex what with five (hopefully) separate devices. I've also got the units buried out of sight and difficult to get to. The temptation to fiddle is reduced but I might just experiment with a spare.
maybe running the CLC in a balanced differential CLC configuration with respect to live and neutral will help enhance common mode rejection performance
H/W the circuit though I didn't fit the 0v link or the 1M resistors.
Because there's a 240V > 24V transformer on the input, the filters are working at low voltage. I didn't see rating as a worry and used a pack of cheap Y-rated 220nF caps. (I'm aware that some places would need X-rated caps were they at mains potential.)
OTOH, currents are higher at the lower voltages even if no isolator is in practice passing more than about 3/4 of an amp. I guessed that 3-amp-rated Murata coils with DC resistance 0.03R would be fine. Well, they're not getting hot . . .
maybe there is a use for a damping resistor from line to neutral after the CLC to enhance it's performance
Agreed but I didn't feel confident in calculating a value so I left them out. Again, I could experiment.
D
I enjoy chatting about this stuff. No need for thanks here. I appreciate your sharing your secrets so I can see how much I'd like to try myself.
Imagine your transformer in place of the common mode choke...This is what I was envisioning Note the Y caps orientation with respect to earth. It might be worth playing with to try and increase impedance to common mode noise.
Though, regarding your filter schematic, I haven't seen anybody using an inductor on earth that way leading me to wonder what it's function in the circuit is. Presumably it can't do much until there is appreciable earth current which in my locale is strictly forbidden above a tiny level.
With double insulated, tinyish transformers, even the authorities presumably wouldn't have any problem with you not using special safety rated parts.
...including the two Henry Ott books in my collection. "Noise Reduction Techniques in Electronic Systems" has been a go-to reference since the 1980's.
These are words invented by some PC users without any justification or basis. Optimisation in the engineering sense means something totally different. The case in point is simply the advocation of using 2 or 3 'optimising' software suites to get some sort of 'best' sound out of an OS. In Mercman's case, he didn't even use the vendor favoured OS in his review of 'AO'.
Optimisation in the engineering sense means something totally different.
So what DOES it mean "in the engineering sense"? You've been making similar points at intervals for at least eight years - since (IIRC) cics first used the word on this forum - but have yet to put us right. You also, as here, seem to confuse what you feel is mis-use of the word with, again in your opinion, a process that is in and of itself flawed.
To help clarify the first point, I looked up "optimise" in the Collins Dictionary. The definition includes "to find the best compromise among several often conflicting requirements, as in engineering design".
Is that wrong? If so, why? If not, can the word not reasonably include configuring an OS, whether performed entry by entry (as with that team-effort, post-cics cMP2 slimming marathon), by Delphic pronouncement (as with your recent Reg-edit tips) or (for those with better things to do in life) buying a little program that more or less does it automatically? Seems everyone else feels it can.
Dunno why but I also looked up "snipe". Its definition includes "to criticize adversely a person or persons from a position of security".
WRT the second point, I repeat that the microRendu's OS was not "optimised" as in slimmed down from a commerical release but configured from the ground up as and only as an audio player. Mercman's finding that his mega-buck Sonore PSU may perform less than perfectly has nothing to do with whether or not he understands what "optimisation" is. That said, I agree with you, soundchekk and hfavandepas that it's likely a system-level issue that may be eased by BandAid filtering but not necessarily "fixed" by it.
I'll respond separately to hfavandepas' points.
D
like a friend of mine, the fmakster "doesn't hold the {fill in the blank} dictionary as any authority" when using language. :)
" In Mercman's case, he didn't even use the vendor favoured OS in his review of 'AO'.
The use of Windows 10 Pro was intentional since many more people have Win 10 than Server 2012 or Server 2016. But I've told you this before.
My review clearly discussed that Server 2012, and better yet, Server 2016 could achieve superior results.
you insist on wireless,networks,and anti virus, your PC system cannot be said to be optimised for best audio replay quality.
I use a fancy Ethernet cable to connect my computer to my LAN and no antivirus software. Fred, you win. My computer isn't optimized, consider it "slimmed".
Edits: 01/27/17
Your remote? How does it work?
Any computer on the LAN can be a remote. Even the computer that I am tying this on that is connected with an Ethernet cable to the router.
Since you don't have any respect for me, given your comments at Critic, why bother even discussing things with me. You have made your feelings on this issue as clear as you can.And in the end, everything matters.
Enough already!
Edits: 01/26/17
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: