NAD CS1 Network Streamer Review

[eliash]

we are far from the era of quality of squeezbox...
end for me
;-)

...obviously not

 

[Kagemusha]

Good guess but since they don't publish measurements like I do, no one would know the DAC is lousy as to then be ready for upsell.

I think the Argon Solo Streamer seems to be cheaper. It supports Airplay2 as well. I dont use Roon by the way.

 

[eliash]

I think the Argon Solo Streamer seems to be cheaper. It supports Airplay2 as well. I dont use Roon by the way.

In the german test, they mention DLNA as a hidden additional feature, probably its not certified but as long as works...too bad it has no Eth...

 

[Bernard23]

I bought one back in March just when they were released because it was the cheapest Roon end point. It was the cheapest solution to effecting room correction at the time, in that role it works fine, I don't use the internal DAC (the one in the Wiim is allegedly junk too). However, I've just read today that the Wiim Pro is also now certified Roon ready so no reason to buy the NAD. FWIW you can stream from the qobuz app using cast. The biggest gripe I have is the unreliable connectivity to WiFi, I'm having to reboot it occasionally which isn't great. I also have a sonos gen 1 also connected to my external dac. The only advantage of the NAD is it's compatibility with Roon, and it's PEQ; and the occasional 24 bit remastered tracks. I'd bet that having room EQ has a far bigger impact on SQ than the differences in device performance between the NAD, the Sonos and the Wiim.

 

[eliash]

Got a NAD CS1 set up here.
Works nicely via Eth or Wifi with Android/BubbleUPnP or Windows/foobar which play my (44.1, 48, 88.2, 96, 192 16/24) FLAC content flawlessly (no drop outs, gapless, 24bit deep with swift & quick operation, just like an USB-DAC). Bubble and foobar play through WAV conversion (FFmpeg/AudioCast:0ff and foo_out_upnp).

2 things to moan about:

There is a silly unsecured WiFi node built in, blasting the NAD´s presence to anybody around. Worst is, if one knows the set mgmt IP address (it has a built-in web server) a foreign person may screw up the whole thing, including false or destructive SW update w/o any password protection.

- If anybody knows how to silence that, I would be grateful to learn how (I am attempted to completely turn off that WiFi module manually, since I solely use Eth anyway)!

Electrically, the analog and the SPDIF outputs share the same ground, which is probably also a reason for the poor built-in DAC measurements, Amir has documented.
They omitted the (mandatory) SPDIF output transformer isolation. So, not relying solely on my RME ADI-2 fs to isolate any EMI via SPDIF, I decided to use Toslink...


Late rem: This is what everybody in range may find by logging into the NAD CS1 built-in web server via its open WiFi node: Full control available, always, also in stand-by!

 

[Bernard23]

Got a NAD CS1 set up here.
Works nicely via Eth or Wifi with Android/BubbleUPnP or Windows/foobar which play my (44.1, 48, 88.2, 96, 192 16/24) FLAC content flawlessly (no drop outs, gapless, 24bit deep with swift & quick operation, just like an USB-DAC). Bubble and foobar play through WAV conversion (FFmpeg/AudioCast off and foo_out_upnp).

2 things to moan about:

There is a silly unsecured WiFi node built in, blasting the NAD´s presence to anybody around. Worst is, if one knows the set mgmt IP address (it has a built-in web server) a foreign person may screw up the whole thing, including false or destructive SW update w/o any password protection.

- If anybody knows how to silence that, I would be grateful to learn how (I am attempted to turn off that WiFi manually, since I solely use Eth anyway)!

Electrically, the analog and the SPDIF outputs share the same ground, which is probably also a reason for the poor built-in DAC measurements, Amir has documented.
They ommitted the (mandatory) SPDIF output transformer isolation. So, not relying solely on my RME ADI-2 fs to isolate any EMI via SPDIF, I decided to use Toslink...

I'm not using the onboard DAC, but is there any technical benefit of optical v coaxial? I had presumed latter, but only because optical is limited to 96k and some of the qobuz files are (pointlessly) 192k and I'm not entirely sure where the downsampling would take place. If I'm using Roon then that can do it, but if I used bubble or heaven forbid one day qobuz gets it app act together and stream direct then I'm uncertain what happening where.

 

[eliash]

In former times DACs with SPDIF or Toslink input didn´t have a perfect jitter suppression. Therefore SPDIF had the advantage of a higher S/N, thus introducing less random jitter than Toslink. Additionally earlier Toslink implementions where not really 192Kb/s capable. Both problems are not really relevant anymore, according to my perception. RME claims to have a more or less perfect SPDIF/Toslink jitter suppression implemented.

Did you recognise the presence of that open NAD CS1 WiFi node as well ?

Late rem: As you can above, the SW version stars with "eng", which could stand for an engineering version - which version do you see?

 

[eliash]

When I had a look at the inside of the NAD CS1 I noticed that there is a SPDIF transformer for isolation, as required and properly layouted. Inserting a foil between chassis and the combined Toslink/SPDIF connector plus opening the connector´s shield screw bore yielded the required SPDIF isolation (right connector in the pic). Connected my RME ADI-2 fs via SPDIF again.



Next step is terminating Wifi and BT into 50 Ohms plus having already disconnected the NFC coil to optimize any possible EMI/RFI pickup in my analog (esp. turntable) setup (and avoid being hacked by wireless).

(The a. m. open Wifi node is used only for configuration purposes to set up the CS1 with Google Home or some Apple app. Having seen it disappear, it reappears when not using the Home app anymore, as I have understood my test results so far. Besides not having been able to get sound from Bubble, running Home in parallel, I gave up further investigation.)

Using Ethernet, similar as Amir used in his test, is anyway best suited for my personal hi-end aspiration.

 

[morillon]

the galvanic isolation via transformer can and is often rather installed on the dac side....

 

[eliash]

...both sides is better, especially when RF (BT/Wifi/NFC) is the game (transformer coupling capacitance!)...besides, one of my DACs has no SPDIF input isolation...
Using the NAD CS1 without its analog outputs, it´s even better to ground the chassis in addition (via PE), since the Eth. port has no shield.

 

[morillon]

...both sides is better, especially when RF (BT/Wifi/NFC) is the game (transformer coupling capacitance!)...besides, one of my DACs has no SPDIF input isolation...
Using the NAD CS1 without its analog outputs, it´s even better to ground the chassis in addition (via PE), since the Eth. port has no shield.

for a long time the pulse transformers type murata pulse have made enormous progress.. but despite everything there is a very slight impact on the rising front.. so the doubled if not necessary is a little clumsy and especially not so useful.. .seems to me...

 

[eliash]

...maybe from a signal perspective 2 soften the edges a bit more, but RF-isolation is even more important to me.
Connecting the CS1 to an ADI-2 via SPDIF always involves 2 tranformers, no matter if the CS1 is modified in the above way or not.
The difference is the SPDIF cable´s ground connection:
>unmodified - CS1 chassis connected with all the impact from its power supply and RF coupling,
>modified - CS1 chassis decoupled by the CS1´s transformer coupling capacitance

 

[morillon]

...maybe from a signal perspective 2 soften the edges a bit more, but RF-isolation is even more important to me.
Connecting the CS1 to an ADI-2 via SPDIF always involves 2 tranformers, no matter if the CS1 is modified in the above way or not.
The difference is the SPDIF cable´s ground connection:
>unmodified - CS1 chassis connected with all the impact from its power supply and RF coupling,
>modified - CS1 chassis decoupled

this fight against rf seems to be your hobby ;-)
but do you have his measurements etc concerning the impact of one or two transformers?
(from the front edge it is easily observed at the typical scope, less than 0.5ns)..and has a slight impact on the jitter..the hobby of many others...seems to me that the compromise 1 transformer is a choice of reason seen in this way .. ,-)

 

[eliash]

...in a way yes, since I learned the impact of electrical and magnetic fields on circuits, designed to maintain a S/N of >110dB the hard way, doing some hi-end audio design.

Coming from an GHz-speed mixed signal design background a long time ago, it is fascinating and nerve wrecking at once, how such RF experience translates into maintaining a (unwanted noise) coupling ratio of less than 1/300,000 in the audio domain (even <1/1000,000 in case of some chinese DAC&AMP circuit designers - hats off!).

Once again, regarding the signal quality, data and clock recovery require noise shaped signals anyway for min. bit-error rate.
More important are reflections on the cable (from poor impedance matching - other than 75Ohm in/out) which will superpose the original signal edges and therefore create unnecessary jitter.
Therefore going through coaxial cable (or POF), the length has to be long enough to move reflections out of the rising and falling signal edges, which translates into a minimum length of some 2.5m...as I re-calculated once...assuming typical SPDIF signal rise/fall times...

 

[morillon]

I know the tdr... manipulating the concept is one thing... measuring it in situ is another... observing the impact in situ at our scale our needs our relevance in terms of objective or subjective results is another... so factually in the subject


mesureaments?
ps we work not in giga but in few mhz....

but the serious murata pulse d impulsion transformers cost next to nothing.
;-)

 

[eliash]

...oh yes GHz, Wifi @ 5GHz...the issue is also about non-linear components and baseband demod. Fortunately not less than 2.5GHz, but already 100mW of power.
Have detected audio opamp input stages (OPA 211) oscillating at 100's of MHz, according to the associated unwanted stub length path at the input. Such are certainly also capable of demodulating Wifi pick-up and one starts to wonder why suddenly something doesn't sound right.

...SPDIF measurements have to wait until IPEX terminators arrive, then I will pull out my Rigol for you, when opening the CS1 again...

 

[Bernard23]

In former times DACs with SPDIF or Toslink input didn´t have a perfect jitter suppression. Therefore SPDIF had the advantage of a higher S/N, thus introducing less random jitter than Toslink. Additionally earlier Toslink implementions where not really 192Kb/s capable. Both problems are not really relevant anymore, according to my perception. RME claims to have a more or less perfect SPDIF/Toslink jitter suppression implemented.

Did you recognise the presence of that open NAD CS1 WiFi node as well ?

Late rem: As you can above, the SW version stars with "eng", which could stand for an engineering version - which version do you see?

If you mean it appears on my home network, yes; that's how (the only way from what I gather) to install it using Google Home. Latter can reset volume, so I remove it from the Home network. Mine is using wifi, my system isn't good enough to warrant installing a satellite router just for the NAD.

 

[eliash]

Finally, after some experimenting to attenuate the unwanted open WiFi nodes generated by the CS1, this was my most efficient and also cheapest HW approach. Two 51 Ohm resistors (1206), instead of the IPEX connectors (one broke during clipping off an IPEX/SMA cable, so I removed both, since I will not use WiFi for this anyway but only Eth networking) yield about 90% WiFi attenuation, good enough for not broadcasting outside of our home (any add´l shielding acted like a WiFi antenna!).

...Admitting, not really an optimum mod, but it serves the purpose...best SW config is the NAD default, letting my FritzBox router do a permanent IP-adress assignment. Somehow an (above shown) IP address, set in the NAD, is not effective.
Fortunately sound performance and stability are flawless. Another nice observation is much less DPC latency in the server-PC going via Eth, compared to the previously used USB driver...

 

[eliash]

Another late remark, when using the CS-1 to a larger extent, I found a bad bug:

Playing UPnP through Bubble (using the FFmpeg transcoder from Flac to wav), extensive metadata let´s the CS-1 crash, better say "go through reset" when starting a respective title.
- It looks to me like any Interpreter, Album or Title field with 32 characters or more causes this crash.

Whether this is caused by Bubble or the CS-1 ist not really clear to me, but the foobar UPnP renderer does not show this fault. The Bubble guys won´t follow up this one (as they informed me, very quickly - Thanks anyway!), since they think it is an NAD problem...

Meanwhile I got ChromeCast mode to work in the CS-1 as well, without crashing it with large metadata as described above, this by configuring Bubble AudioCast to wav transcoding, even runs 192K/24b files, but not gapless...

This UPnP area looks like a minefield to me meanwhile, interoperability seems to be an ongoing challenge...the foobar renderer being the most reliable in this context...

 

[eliash]

Fortunately the UAPP (USB Audio Player Pro for Android by eXtream Software Development) has UPnP (in addition to USB) integrated now as well.
Meanwhile nicely working beta SW with the NAD CS-1, thanks to "intensive care" by their developer - maintaining their high end replay capabilities through Android!