Fibre Isolated Network

Good stuff Dave - interested to hear how you get on with this.

James

Well on that basis I assume your router is not supplied by BT and your switch is half way decent.

I have read and tried to understand Simon's reasoning for why it *could* make a difference, but frankly I don't buy it. But hey its only money right?

Fill your boots.

Draytek router and Cisco SG 300 switch. Can't be any worse than spending £1600 on a 1m Ethernet patch lead? It all adds to the rich debate 

Dave

Dave - I was looking at something similar last week and concluded the fibre to wired ethernet converter would be the weak link so didn't pursue further. I would be interested t hear how you get on - but I suspect the major improvement will be when Naim support fibre ethernet interfaces directly.

Gary - if you don't follow my rambling musings - I don't blame you - you are not alone   but try reading the Texas Instruments white paper on wired Ethernet noise and  design mitigations I posted on the Audioquest Cable thread. It really is all quite straightforward - as long as you remember to think real engineering world rather than simplified marketing-sales collateral  conceptual 'white papers'.

Simon

The mini GBIC fibre adaptor for the switch arrived on Friday and had been plugged in. The log file shows it as being recognised so all is well at the switch end. The media converter and fibre patch lead should arrive on Monday along with the components to build the linear PS. I've made a start at designing the PCB for the rectifier and regulator so progress is being made.  

The fibre patch and tranceiver arrived today and have been connected up to conduct some initial tests. At present evertthing is still in my comms cupboard in the garage loft and running off the standard wall wart power suppliers.

 After a bit of fiddling with port settings the connectivity to the NDS has been established so I'm off to pour a glass of wine and do a bit of listening. The linear PS will be constructed over the weekend.

I will of course continue to report on progress for anyone who is interested.

Dave

Dave, I looking forward to your comments... Where are you putting the fibre to UTP converter?

Simon

Ultimately it will be on a separate radial feeding a linear PS, right next to my NDS. This won't be possible until I change my listening room, so for now I'm just testing with known limitations. My current room is about 12m away from my switch and the longest single-mode patch cable I can find is 10m. I could make up a longer cable, but don't fancy shelling out a grand for a termination toolkit!

The new room will not be ready for a few months, so in the meantime I may pull the fibre as far as possible into the main house loft and only run the last couple of metres over copper.

Thanks for sharing this Dave!

Will be interesting to read about whatever differences you find compared to a direct copper connection from your main switch to the NDS.

Good stuff - keep the updates coming Dave.

James

By all means keep us informed but be brave enough to admit it sounds no different if that is the case.

The trouble with spending money and making effort is that you are fairly pre disposed to hear a difference

I will certainly do my best to remain objective... Much easier when the total outlay is less than £400 compared to 4 times this for a Chord Sarum TA...

Gary, or it could make the Naim sound less good, as I discovered with the Revive LAN isolator in my particular setup.

Simon

Just finished a test assembly of the Linear PS to make sure everything fits before I solder it all together tomorrow. It's a very simple circuit based on a 7809 and 7812 three terminal regulator with oversize 10000uF smoothing caps. 2 separate regulators are fed from separate secondaries on the toroidal transformer. Decent case as it's going to be on display. The 12v will feed the switch and the 9v the media converter. The switch will allow me to either leave the outputs floating or tie them to ground to see if either option sounds better.

With regard to sound quality definately no negative impact noticed, and perhaps a slight improvement in detail. Too early to comment really.

Dave

For anyone who's interested, I have now completed the power supply for the media converter and switch. No serious listening so far as I just want to check everything's stable for 24-36 hours. The eagle-eyed amongst you may notice that the PCB layout has changed. In the testing of the earlier version I had not properly calculated the heat dissipation from the regulators and had undersized the heat-sinks. In this version I have included some power resisters before the reg to drop a few volts under load to minimise the loss in the regs. I've also had to mount the 9v reg on the chassis as this is the one with the highest current draw from the media converter.

Next step is some serious listening. For optimal use I have to finish my listening room as I can't drag the fibre to the current position easily.

Dave interesting. Out of interest why didn't you choose a toroidal with a lower secondary voltage?

S

Dave, I'm watching this place closely.  

I'm very excited  -- much the same as I was when watching your exercises with Slim Devices Transporter SMPS  

But as soon as PS was not a subject of this exercise why just not to employ one of the Naim's power supplies, like NAPSC or HC2?

Unless, of course, assembling PS is an easier way for you rather than get one from a local dealer on home demo.

Simon, I needed a 9v and 12v supply, but wanted a single transformer with 2 secondaries to keep the size of the case small. The transformer I chose was the best compromise that gave me enough headroom for the 12v reg. Unfortunately the 9v circuit draws more current, so with the increased input voltage to cater for the 12v side was dissipating far too much. If I could have found a transformer with 2 different secondary voltages it would have avoided the load resistors.. The one in the 12v circuit is not really needed, but gave me a convenient way of calculating the current drawn from the voltage drop so I left it in place.

AMA, you are right that the PS is a secondary concern in this case, but it was easier for me to build a simple custom supply than try to use anything off the shelf. Naim's PS units are designed for a particular job and in any case cost far more than the components I needed for this. I also enjoy getting a soldering iron out from time to time 

Dave

Nice work 

Thanks James. The next step is to work out whether it's all worth it. My initial thoughts are positive, but I still have about 15m of cat6 between my new isolated switch and the NDS so it's not a true test. The good news is that with the fibre transceivers and single mode fibre I've got a maximum distance of 15km to play with now; any placement worries are gone! 

Dave, I used to believe that ethernet input in NDS has transformer decoupling and hence galvanically isolated from router lines. Is that the case?

Ama, I'd be surprised if it didn't, most Ethernet NIC cards have galvanic isolation to avoid ground loops.

Simon

Yes, the NICs line driver is via a transformer, so electrically isolated although noise could still be coupled. The fibre approach should still offer higher degrees of isolation so I'm hoping to get some improvement, but as I said in the other post it's difficult to know since I've added a 552 to the equation.

The first point of galvanic isolation that the NDS white paper mentions, is after the incoming TCP stream has been unpacked, buffered and clocked for the DSP. So just before the DSP.

Dave, are you planning to do a comparison of your solution VS a more conventional copper connection? Thanks anyway for the story, it's been interesting!

Felix

Felix, I believe AMA was referring to NIC galvanic isolation, which is part of standard Ethernet NIC design for safety and DC ground isolation across the Ethernet connection.

The NDS like the NDAC and other premium Naim components use opto isolation to seperate  key modules of the design from each other to help reduce noise in the ground planes of the modules affecting each other.

Strictly speaking for these to be electrically galvanically isolated, as indeed mentioned in the Naim white papers, each module needs a seperate ground all the way back to the mains transformer separate    windings or  actual separate transformers and the grounds are completely seperate from each other with no DC path between them with each module having its own seperate  0 volts.

Simon