ROON ROCK with a linear power supply?

Using a HDPlex 200W linear PSU which has several output rails at varying voltages: 15v into NUC running ROCK, 5v into fibre media converter (feeding a microRendu), 9v into network switch, another 5v into external USB SSD on which resides my Roon library.

I tried out a Benchtop LPSU (with amp meter and the NUC only draws 0.7A at 15v under SRC to DSD), a Teddy Pardo 12v supply (very nice) before settling in the HDPlex because of its multiple outputs. Sound quality difference? Yes and not unsubtle - higher frequency stuff had a harsh, brittle sound with the SMPS that came with the NUC. Couldn’t really tell a difference between the LPSUs.


What resistance?

If the NUC has a power draw of x Watts then it’s going to pull x Watts - which means the only way it can pull x Watts if the input voltage falls is to increase the current draw.

The NUC is not a light bulb or heater (not a resistive load) - it is a multitude of switching DC-DC convertors - which will keep increasing their current draw as the voltage falls until the input voltage gets too low and they stop working.

Your NUC is not directly to DAC? And you still benefited from using a linear power supply to it? And to your router? Wow! This is the opposite of what everyone is saying here!

Lowered line voltage can place stress on power supply capacitors:

Reduced input voltage can cause excessive power supply heat dissipation, resulting in short equipment life. What’s behind this overheating? While trying to maintain constant DC output as the line voltage declines, the DC-to-DC converter circuit has to draw from the reservoir capacitor. With line voltage reduced, this capacitor experiences deep discharges between the twice-per-cycle charging periods.

But if you’ve got a DC supply that puts out a lower voltage to a NUC, then I would expect the NUC to draw less current unless there are further power supply stages inside the NUC itself. That may be the case. Apologies for the distraction.

No. A NUC will always need to draw a certain amount of power. Indeed if it needs to draw 15 Watts to operate, if you lower the input voltage the current has to rise to compensate in order to keep delivering 15 Watts. Intel specifies that a NUC will work at any voltage between 12 and 19 volts which means there are further voltage regulation stages. So a NUC is not a simple resistive load. It is a complex active component and the resistance it presents as a load can and will change depending on supply voltage. And the “stressing capacitors” comment is a real red herring in this context. Intel tell you what voltage and power is needed, if you remain within their recommendations then assume any capacitors are designed to work correctly.


Yes that’s right: NUC->ethernet->switch->ethernet->fibre media converter->fibre patch->fibre media converter->ethernet->microRendu->usb->DAC

Which fiber converters and patch do you use? Are you in US?

I use a pair of TP-Link MC100CM with a 1m Lindy patch. Cost about £60 all in. By the way, I went with the 10/100 rated FMC’s rather than the full-fat gigabit due to lower noise circuits.

I’m based in England.

It is more than that. I used to design switching buck regulators used in DC-DC converters. Almost all, NUC uses its own switching buck regulators for the CPU, memory, northbridge chipset, USB and other stuffs. This type of buck regulator (not to be confused with linear regulator) when input voltage is high, the current drawn will go low. Power efficiency will go up when input voltage is high.

Therefore it is more advisable to operate it at higher voltage, say if the NUC is rated at 19V though it still can run at 12V but at a lower power efficiency. Good power efficiency ensures maximum power transfer to the device.


Hi Rik,

comparing the MC100CM to TL-SM311LS, do you think there could by any differences or pro/cons regarding SQ?


I’ve not heard of the TL-SM3 before. If I understand this correctly, it plugs directly into an ethernet switch (assuming it has an SFP/SFP+ port) which is good as it eliminates two sets of RJ45 patch cables. There are no disadvantages to this as any electrical noise arising from close proximity to the switch is vanquished as part of the conversion to optical. And, then, presumably there’s another TL-SM3 at the receiving end also plugged into a switch. If I’ve understood this correctly then I’d say this is possibly neater than the MC100CM option.

I don’t think this will lead to any change in SQ. Neat device.

I agree and there is no need for additional PSs to run the converters.

Actually I run audio by direct eth connection (Nuc 7i5BNH > eth > Squeezebox Touch > Toslink > DAC), so I’m wondering why not to plug both TL-SM3 directly into both devices LAN ports?

(Regarding to SQ I had to realize, in my system Toslink (SB Touch > DAC) for the audio signal works better compared to USB. Also the response time UI controls are very fast).

1 Like

Hı all.

Just a quick note on experience.

Running NUC with a linear PS makes a good difference on my part, although the NUC is not connected directly to the DAC.

Here is my digital chain

nuc7i5 (WD MyBook Duo on USB for data) > PaulPang Switch > MC100CM > 10m 50/125 turquoise optical cable > MC100CM > 75cm Audioquest Diamond Ethernet > Acoustic Revive RL-1 > dCS Network Bridge

Powering the switch and the media converters with battery is a big plus. Also the optical cable type makes a huge difference. 62.5 Orange ones degragades the sound.

The WD is now powered with its own supply with an iFi DC iPurifier. May move that to Linear as well.

The PS I am using is a Paul Hynes SR-5 15v which I had lying around from an earlier setup.

Next is moving the NUC to a fanless case which I will receive in a couple of days. Getting rid of the fan dirtying up the power should also provide beneficial, as well as silencing the already not very loud spinning noise.


1 Like

OK. For the heck of it, I gave it a try: I bought a SOtM SPS-1000 and used it power the NUC w/ ROCK, as well as my Ethernet Switch.

I am thrilled to report that I am happy with the results! Noticeable improvement in SQ. No question !

Using a Lumin T1 as a Roon endpoint.

Not sure which did the trick, powering the switch or the NUC or both, but I don’t care!

Give it a shot guys. Even if NUC is not the player. You may be pleasantly surprised

What switch do you have?

John Swenson ‘s approved Netgear N105

1 Like

Thanks. Was that meant to say GS105?

1 Like

Sorry. FN105. The original finding from JS. DC output from sps-1000 grounded

Thanks again.

Did you try with using the sPS-100 with only the switch and only the NUC?

Nope. I put them both at the same time the SPS-1000 arrived