WolfX700 Measurements of Holo Audio Spring 2 Lv2

[YSC]

One more thing, did you measure the NOS frequency response? I really enjoyed my unit in NOS, and wonders is the warm and smooth sound resulted from the HF roll off in NOS

 

[Veri]

I really enjoyed my unit in NOS, and wonders is the warm and smooth sound resulted from the HF roll off in NOS

Simple answer, yes warm/smooth means the high frequencies are rolled off due to NOS.

 

[YSC]

@WolfX-700 did you keep the power always on and only mute it with front panel button or switch off at the back each time?

 

[WolfX-700]

@WolfX-700 did you keep the power always on and only mute it with front panel button or switch off at the back each time?

I just shut down my Poweramp and let Spring2 run all the time (my Roon library is big enough) I will measure it after a few days (should be this weekend)

So you have to wait for the frequency response.

 

[YSC]

Sure for the waiting, really curious will the channel becomes more balanced in nos pcm mode (I bet not).
Just wonders do you objective audio guys believe in the manufacture claims that this discrete R2R thing need the power always on to keep the internal thermally stable for optimum performance (sounds kinda makes sense to me as discrete resistors arnt thermally trimmed or compensated)
Or is it just some marketing snake oil.

also wonders what tolerance resistors are they using here, be it 0.01% or worse ones is interesting as they managed to get into the SINAD of some of the higher tier DS dacs

 

[YSC]

actually anyone have idea whether the spring 2 have full R2R ladder of 24 bits or some kind of software side implementation like the denafrips Ares? I remember it accepts 24 bits pcm but only have 20 bit ladder

 

[WolfX-700]

update:

 

[bennetng]

Could you try 997Hz? In both NOS and the highest possible OS mode.

 

[YSC]

So this is update on NOS pcm mode? Seems like the designer theory of the running in actually worked! And call me surprised that the SINAD actually is in the second quarter of the best dacs! Amazing for NOS R2R

 

[WolfX-700]

So this is update on NOS pcm mode? Seems like the designer theory of the running in actually worked! And call me surprised that the SINAD actually is in the second quarter of the best dacs! Amazing for NOS R2R

Yes NOS

 

[YSC]

Yes NOS

Thanks WolfX! Really surprised me, I was expecting the measurements won’t change before your update (as a lot of running in myth), but glad to know your random purchase unit measures this well and it ended up channel balance being almost perfect! I can say happy goodbye to my purchase money now

 

[YSC]

Any idea about the filter and roll off of various modes?

 

[WolfX-700]

Any idea about the filter and roll off of various modes?

Same as Spring:

https://www.audiosciencereview.com/...-measurements-of-holo-audio-spring-dac.10047/

 

[Veri]

Yes NOS

Thanks WolfX. That's interesting indeed

 

[YSC]

Any idea after running in did linearity or IMD etc. Would improve, Aside of the physical properties of the resistor ladders being ran in I believe more on the FPGA self learning made some improvements? If not I can’t believe Jeff actually can use 0.01% tolerance (I bet he did?) resistors to create a randomly purchased unit (I think @wolfx just buy online and not let Jeff hand pick the golden sample? As proven by first trial measurement channel difference) to perform such matching and low SINAD on a discrete R2R product! Interested in how a MSB which cost an arm and leg will fare

 

[WolfX-700]

Any idea after running in did linearity or IMD etc. Would improve, Aside of the physical properties of the resistor ladders being ran in I believe more on the FPGA self learning made some improvements? If not I can’t believe Jeff actually can use 0.01% tolerance (I bet he did?) resistors to create a randomly purchased unit (I think @wolfx just buy online and not let Jeff hand pick the golden sample? As proven by first trial measurement channel difference) to perform such matching and low SINAD on a discrete R2R product! Interested in how a MSB which cost an arm and leg will fare

I'm surprised you are surprised by this result ... (Although I was also surprised when I measured Spring before) because I put the measurement results of Holo Audio May in the forum ...

https://www.audiosciencereview.com/...may-probably-the-best-discrete-r2r-dac.10161/

 

[YSC]

I'm surprised you are surprised by this result ... (Although I was also surprised when I measured Spring before) because I put the measurement results of Holo Audio May in the forum ...

https://www.audiosciencereview.com/...may-probably-the-best-discrete-r2r-dac.10161/

Indeed the May is a lot better and ranked inside the absolute top rank, what I am surprised is in normal circumstances when a new top end model is announced and the more budget option will got cheaped out on selecting best matching components and leaving those for the flagship, and seems holo didn’t do that, kudos for them and your effort. Also it’s the random purchased unit having self correct the channel balance on a good way (increasing SINAD for poorer channel) but not a logic to reduce the better performing channel is pretty amazing

 

[YSC]

Dear dr WolfX, as in China we have nothing to do and bored to death, did you enjoy much listening the Spring 2 in NOS? or you just use other dacs/OS mode for enjoyment

 

[Miska]

Not better than 96/24 (even considering 4db)

View attachment 44543

For best performance at PCM you need to set "DAC Bits" to 20, so that output is dithered to 20-bit level. With TPDF dither, but you could also use something like NS9 or LNS15 noise shaper at output rates of >=352.8k.

Note that when I measure I take into account whole output spectrum up to 5 MHz or whatever needed to reach point where analog filter's response falls below constant noise floor of the DAC. If you run the DAC at 96k (NOS) there's massive amount of correlated images (distortion) in the output spectrum, so badly incomplete reconstruction. At 1.5 MHz the analog filter manages to remove practically all images and thus you are getting pretty decent reconstruction of the analog signal.

Here's -120 dBFS signal with 24-bit data:

And here's the same signal with 20-bit NS5 noise shaped dither, at 1.4112 MHz rate:

 

[Veri]

Here's -120 dBFS signal with 24-bit data:
View attachment 56306

And here's the same signal with 20-bit NS5 noise shaped dither, at 1.4112 MHz rate:
View attachment 56307

Okay so #2 is cleaner, hands-down agreed. But we are talking... -140dB down ...?

What's the point.