Ethernet switch

Looks like you can go a bit too far with some of this Paul Pang stuff - externally clocked switch 

Could you explain in exactly what way this is a 'huge upgrade'? A 'huge upgrade' for £100 sounds distinctly unlikely to me. A tiny improvement maybe, but again - in exactly what way? 

I hope you do realise that Naim streamers actually re-clock incomimg data flows to remove any jitter? So using another re-clocking device in a switch seems a bit superfluous.

The external clock just replaces the basic oscillator module that would normally be in the switch Adam - it's not a re-clocker. The picture i posted is a higher spec version of the Paul Pang switch. I was just a bit surprised as to how far some of these companies go...

Fair enough-  I though it was re-clocking the data flow.

Whoa there audiophilia peeps,  forget the clock stuff,  lets stick with the OP's basic switch (please) & HH's question about huge upgrade,  define "huge" please.  I might be one of the few on the forum who have listened to a PP switch & yes the PP was pretty good, I found just found the difference to not be so much better. ......  lets say I still have my GS105

Hi. Sorry my enthusiasm got the better of me. In the context of my system it made a significant difference. More texture to the sound and more musical, a click into focus without being clinical.

When searching for the ength degree you sometimes pay thousands for a small difference, well this is just £100 and it now means my streaming matches/betters my cd replay for the first time. Something I have been trying to achieve for 3 years.

Please specify the PP switch model and the model of linear PS you are using to allow us to check them out (on line anyway).

The pp switch is the basic one without the external clock. The ps is an mundorf caps high end bespoke made type.

Adam Zielinski posted:

I hope you do realise that Naim streamers actually re-clock incomimg data flows to remove any jitter? So using another re-clocking device in a switch seems a bit superfluous.

Adam, this makes no sense what so ever... you are talking about two completely different functions that do very different things.. so can't be superfluous. It would be like saying my car has shock absorbers and steering, but because it has shock absorbers the steering is superfluous ! But clearly in this case if the performance is poor in either the overeall handling will be impacted.....

But essentially clock jitter produces a wider range of RFI.

RICHYH posted:

On Monday I fitted the Paul Pang ethernet switch into my high end streaming system, with linear power supply, and have to totally recommend this as a huge upgrade over the netgear 105. You won't get a better £100 upgrade.

Interesting, they seem to be addressing what I have seeing for some time now... I will see if they post details of the jitter in the Ethernet BaseT clock... to me I would put this well ahead of boutique Ethernet cables (for engineering reasons)... although the two together will make a difference..

Now I also predict someone will start providing Ethernet equipment for the audio consumer with BaseX SFPs ... perhaps need some of the high end audio manufacturers such as Naim to use BaseX interfaces... of course you can use BaseT SFPs as an interim and swap out later.

james n posted:

Looks like you can go a bit too far with some of this Paul Pang stuff - externally clocked switch 

It's good to see someone finding a use for old steam iron 'mains flex' at last.

if every upgrade has produced improvements when is it enough.  All changes may change but do they improve.  One thing is the cost of real good units cost less ,that is the biggest improvement

I also note from the photo that Paul Pang device seems to offer three fixed flavours of 802.1p frame priority ports; High, Medium and Standard. I haven't seen this before on consumer stuff.. that is also interesting and will potentially allow more  consistent Ethernet frame timing performance irrespective of switch load... 

This is one example of a  SQ improvement technique you can do with a commercial managed switch such as a Catalyst 2960.

audio1946 posted:

if every upgrade has produced improvements when is it enough.  All changes may change but do they improve.  One thing is the cost of real good units cost less ,that is the biggest improvement

Good question... and I am not sure all so called upgrades are neccessarily performance improvements although may sound different.

However one thing I find hard to understand is that peeps spend a sizeable fortune on audio equipment, speakers, interconnects, fancy Ethernet cables, dedicated mains radials, quality supports etc.. but when it comes to networks and media server platforms any old thing will do...as long as it functions...

I guess to answer your question, you stop upgrading when all your components are of equal quality and in balance....

Simon-in-Suffolk posted:

Adam Zielinski posted:

I hope you do realise that Naim streamers actually re-clock incomimg data flows to remove any jitter? So using another re-clocking device in a switch seems a bit superfluous.

Adam, this makes no sense what so ever... you are talking about two completely different functions that do very different things.. so can't be superfluous. It would be like saying my car has shock absorbers and steering, but because it has shock absorbers the steering is superfluous ! But clearly in this case if the performance is poor in either the overeall handling will be impacted.....

But essentially clock jitter produces a wider range of RFI.

 

Simon - possibly mixed two re-clocking processes up.

I've read that Paul Pang fitted a re-clocking mechanism. Presumably to stabilise the incomimg stream. Somethink akin to what you reffered to it in your other post as a server jitter, but this time caused by a network/

Naim streamers re-clock incomimg digital streams to remove jitter from digital sources.

I presumed the functions of those would double up. By the looks of it I was wrong.

Adam

Simon-in-Suffolk posted:

...

However one thing I find hard to understand is that peeps spend a sizeable fortune on audio equipment, speakers, interconnects, fancy Ethernet cables, dedicated mains radials, quality supports etc.. but when it comes to networks and media server platforms any old thing will do...as long as it functions...

...

Simple, they don't understand the effects on the audio replay, and in many cases people don't really understand the network components themselves.

And companies like Melco aren't helping with their barely believable stories about audiophile quality SSDs, without any sensible indication as to what changes they have made to a industry standard device (indeed one that's a closed system component) and how they have managed to make them so much better.

Adam, FWIW, the reasoning behind switch clock accuracy is not about frame timeing or anything to with the data per se ... it's the transport clock.... it's the stability  of the clock that all sent communication on the ethernet physical  port is effectively synchronised to. Actually there will be two clocks.. one on each end. The streamer will have a send clock, and the switch will have a send clock. 

Now if the switch clock has jitter, then a set of intermodulated frequencies (sums and differences) will most likely appear in the receiver's electronics, such as coupled powerlines, ground plane or electromagnetic field. In a data world this is inconsequential ... but when the receiver is an audio streamer this may couple into ithe internal components that directly impact the audio such as digital audio clocks or analogue circuitry.

The effect is not totally dissimilar to modern SPDIF transport jitter, albeit the here there is only one clock from send to receive. Again here the transport jitter has no direct impact (unless a very old design) on the sample clock, but does cause coupled effects.

Simon

Huge, I don't disagree. However having been a product manager in my career, I do get that in the marketing and sales collateral you often have to use the language of the majority of your customers. Albeit it can not be an untruth, it is often very simplified, and also the language that addresses  your customer's aspirations, fears, uncertainties and doubts is often used to market your product or service.. We used to use the concept of  'white paper' when we provided a more technically accurate and technically specific sales and marketing collateral. I am not sure I have yet seen a Melco white paper.

Thanks Simon. I feel better about using the word huge now. It was a big statement but correct for £100. As you say the timing is excellent now, good before also.

Huge posted:

And companies like Melco aren't helping with their barely believable stories about audiophile quality SSDs, without any sensible indication as to what changes they have made to a industry standard device (indeed one that's a closed system component) and how they have managed to make them so much better.

Hi Huge,

Earlier this year I had an email exchange with Alan Ainslie at Melco and thought you might be interested in his reply to my question about 'audio grade SSD.' This is what he said:

Audio Grade SSD was put in place as a result of poor performance of several IT SSD when evaluated in complete machines vs HDD. There are clear advantages to SSD of course, but two important downsides – one being the use of wear levelling SW running on the derive.  Second is large power spikes due to synchronous addressing of data. So high grade Flash is used with no wear levelling (or very minimal and disabled in normal operation) and sequential addressing to minimise the PSU spikes.

ATB, David

I am not sure where I could or should include an "improved" switch like this in a network such as mine.

I have a Virgin modem (their standard router operating in modem-only mode); a better-quality Netgear router (R7000); two HP 36 port switches; and a gateway / distributor thing that feeds ethernet cable throughout the house.

Where would an "improved" switch fit?  In place of one or both of my HP switches (but obviously I would need the same number of ports)?  In a daisy chain between the Netgear router and the HP switches?  Or between one of my wall sockets and my streamer?

Wherever the new switch sat, wouldn't the effect just be dragged down to the level of the lowest-quality piece of equipment in the network?  

I would put the better quality clocked switch connecting directly to your audio enabled network components such as a streamer. 

The clocking I am referring to only is meaningful across a network segment, i.e. The physical connection between a switch and host such as a streamer... so no you won't be dragged down to the lowest dominator from elsewhere in this regard.

Yes in the limit there could well be couplings from other network segments, but at each coupling the effect will be dimished if detectable at all.

Simon-in-Suffolk posted:

Adam, FWIW, the reasoning behind switch clock accuracy is not about frame timeing or anything to with the data per se ... it's the transport clock.... it's the stability  of the clock that all sent communication on the ethernet physical  port is effectively synchronised to. Actually there will be two clocks.. one on each end. The streamer will have a send clock, and the switch will have a send clock. 

Now if the switch clock has jitter, then a set of intermodulated frequencies (sums and differences) will most likely appear in the receiver's electronics, such as coupled powerlines, ground plane or electromagnetic field. In a data world this is inconsequential ... but when the receiver is an audio streamer this may couple into ithe internal components that directly impact the audio such as digital audio clocks or analogue circuitry.

The effect is not totally dissimilar to modern SPDIF transport jitter, albeit the here there is only one clock from send to receive. Again here the transport jitter has no direct impact (unless a very old design) on the sample clock, but does cause coupled effects.

Simon

Thanks Simon - learned something new today. Much appreciated.

dsc posted:

Huge posted:

And companies like Melco aren't helping with their barely believable stories about audiophile quality SSDs, without any sensible indication as to what changes they have made to a industry standard device (indeed one that's a closed system component) and how they have managed to make them so much better.

Hi Huge,

Earlier this year I had an email exchange with Alan Ainslie at Melco and thought you might be interested in his reply to my question about 'audio grade SSD.' This is what he said:

Audio Grade SSD was put in place as a result of poor performance of several IT SSD when evaluated in complete machines vs HDD. There are clear advantages to SSD of course, but two important downsides – one being the use of wear levelling SW running on the derive.  Second is large power spikes due to synchronous addressing of data. So high grade Flash is used with no wear levelling (or very minimal and disabled in normal operation) and sequential addressing to minimise the PSU spikes.

ATB, David

That's interesting:  For most SSDs those are simply configuration parameters - i.e. the operating mode in which the device software uses the SSD, rather than differences in the SSD hardware itself (although in some SSDs, these features can't be turned off)

So audiophile system software not audiophile SSDs!