Review and Measurements of Holo Audio May --- Probably the best discrete R2R DAC

[amirm]

So what's the measurement device that came up with the above measurements then ?

Picoscope is a high-speed, PC based oscilloscope. Since it runs at such high speed, its ADC has poor dynamic range for audio applications. Depending on model, picoscope ADC is rated from 8 to 16 bits. I have yet to see whether they actually achieve 16 bits but even if they did, that is 96 dB SINAD and -96 dB noise floor which is totally inadequate for performance testing of audio gear.

Picoscope gets used because it is so much cheaper than Audio Precision. It is the wrong tool for audio measurements beyond rudimentary levels. If a company can't afford an Audio Precision or at least Prism Sound dScope, then little can be said about their attention to best performance.

 

[maxxevv]

Picoscope is a high-speed, PC based oscilloscope. Since it runs at such high speed, its ADC has poor dynamic range for audio applications. Depending on model, picoscope ADC is rated from 8 to 16 bits. I have yet to see whether they actually achieve 16 bits but even if they did, that is 96 dB SINAD and -96 dB noise floor which is totally inadequate for performance testing of audio gear.

Picoscope gets used because it is so much cheaper than Audio Precision. It is the wrong tool for audio measurements beyond rudimentary levels. If a company can't afford an Audio Precision or at least Prism Sound dScope, then little can be said about their attention to best performance.

Exactly what I was alluding to in my questions.

If the device itself is unable to achieve the precision and/or resolution of measurement displayed, how can it be claimed that the performance of the device actually achieves the claimed levels ?

Its like using a hand stopwatch to time a cycle of the LHC particle loop at CERN !!

 

[Miska]

Picoscope is a high-speed, PC based oscilloscope. Since it runs at such high speed, its ADC has poor dynamic range for audio applications. Depending on model, picoscope ADC is rated from 8 to 16 bits. I have yet to see whether they actually achieve 16 bits but even if they did, that is 96 dB SINAD and -96 dB noise floor which is totally inadequate for performance testing of audio gear.

Picoscope gets used because it is so much cheaper than Audio Precision. It is the wrong tool for audio measurements beyond rudimentary levels. If a company can't afford an Audio Precision or at least Prism Sound dScope, then little can be said about their attention to best performance.

I use both, plus 200 MHz bandwidth 8-bit device (1 GS/s)...

So far it is best resolution I've found for 5 MHz or more bandwidth. I would actually prefer 10 MHz bandwidth with 16-bit or more.

Good for testing out of band signals.

To stay on topic, here's Holo Spring 2, with OS PCM enabled, 0 - 22.05 kHz sweep at 44.1 kHz input rate, peak hold mode:

Since the on-board digital filter outputs at 8x rate, you can see images around multiples of 352.8 kHz sample rate.

Here's the same test signal, but upsampled to 1.4112 MHz in HQPlayer and then sent to the DAC now running in NOS mode:

For some comparisons, Focusrite Forte ADC/DAC, playing same test signal:

It also works nice for studying modulator noise floors, like ESS Sabre:

 

[pkane]

@amirm

Tim from Kitsune audio offered this:

 

[Killingbeans]

https://www.audiosciencereview.com/forum/index.php?threads/msb-dacs-measurements.9112/

That thread is one of the reasons why I wrote this earlier:

I'd really like to see an MSB DAC get tested. I'm far from an expert, but from what I've gathered so far they seem be a little "creative" when getting numbers from FFT measurements.

I'd really like to see some more independent measurements of the MSB DACs before I make any personal conclusions about them.

 

[Esotechnik]

If the device itself is unable to achieve the precision and/or resolution of measurement displayed

For measurements of high-quality systems, before the appearance of measuring complexes, band-pass filters and a voltmeters were usually used.
Audio Precision x555 has Residual THD + N (22 kHz BW):
-117 dB +1.0 μV
In "high bandwidth mode" (SAR) - specified only 24 bits and 2.5Ms/s, without noise figure and THD.

 

[Dominique-T]

I am also considering whether to buy this unit. There is a good reason for me to consider paying: it looks so pretty inside.

It's a look that will fascinate people like me.

View attachment 40891

Very nice looking...
Would you have the same kind of photos showing the power supply?
I agree with you, expensive, but tempting, looking different with amazing perfomances !
Dominique

 

[Veri]

Very nice looking...
Would you have the same kind of photos showing the power supply?
I agree with you, expensive, but tempting, looking different with amazing perfomances !
Dominique

I saw more pics here:
https://audiophilestyle.com/forums/...ac-r2r-dsd512/?do=findComment&comment=1011425

Power supply:

if overengineering was an art these guys got it down to a T

 

[Dominique-T]

I saw more pics here:
https://audiophilestyle.com/forums/...ac-r2r-dsd512/?do=findComment&comment=1011425

Power supply:

if overengineering was an art these guys got it down to a T

Very impressive... 4 transformers!

 

[NTomokawa]

This has probably been asked before:

Scientifically, what is the advantage of a resistor ladder DAC vs. an IC-based DAC?

 

[Veri]

This has probably been asked before:

Scientifically, what is the advantage of a resistor ladder DAC vs. an IC-based DAC?

Advantage? None. Both delta sigma modulation and R2R can result in excellent performance. Worth browsing through this thread:
https://www.audiosciencereview.com/forum/index.php?threads/delta-sigma-vs-“multibit”-what’s-the-big-deal.5359/

The only difference I could personally hear was with NOS ladder DACs rolling off the treble quite early. But that has nothing to do with the ladder itself. Scientifically, the AK4499 in a Topping D90 will sound the same as this Holo May DAC. There's no reason it would not.

 

[pma]

Picoscope gets used because it is so much cheaper than Audio Precision. It is the wrong tool for audio measurements beyond rudimentary levels. If a company can't afford an Audio Precision or at least Prism Sound dScope, then little can be said about their attention to best performance.

This is misleading and not true. Picoscope is used for the reason that it has much higher bandwidth than AP and is used for different kind of measurements. Picoscope can measure what AP cannot and vice versa. Picoscope sampling rate is >=100MHz and analog BW >=10MHz, depending on type. It is good for measurement of HF noise spectrum and of time domain responses. AP is excellent for audio band analysis and usable to some 1MHz. Not to 10MHz or more, not usable for HF measurements and not usable for fast rise time measurements. There is no reason to state any competition between these instruments, each is determined for very different measurements.

 

[Esotechnik]

resistor ladder DAC vs. an IC-based DAC

IC-based resistor network (AD1862, PCM1704, AD5791) is more precision and stable because all resistors (laser trimmed) are located on a common heat-conducting base. Less thermo-emf effects than FR-4 PCB (therefore MSB use 4-8 parallel discrete boards). R2R DACs have guaranteed voltage accuracy (up to 1 ppm) and fixed conversion time.

1-bit deltasigma DACs can be detected without any instruments. It is easy to hear by applying an inaudible 0.1 Hz pure sine.

 

[amirm]

This is misleading and not true. Picoscope is used for the reason that it has much higher bandwidth than AP and is used for different kind of measurements. Picoscope can measure what AP cannot and vice versa. Picoscope sampling rate is >=100MHz and analog BW >=10MHz, depending on type. It is good for measurement of HF noise spectrum and of time domain responses. AP is excellent for audio band analysis and usable to some 1MHz. Not to 10MHz or more, not usable for HF measurements and not usable for fast rise time measurements. There is no reason to state any competition between these instruments, each is determined for very different measurements.

You are repeating what I said without regard to what was asked. Question was what device was used by MSB to characterize their DAC:

I explained it was Picoscope and it was poor man's AP as used by some audio designers to save money. The above graph stops at 1 Mhz so the point you are making regarding bandwidth of PIcoscope is irrelevant (and what I explained anyway). Above they are talking about noise yet are using a measurement device with far worse noise performance than their own DAC. Wide FFT is helping them mitigate this but it still isn't the right tool for the job.

Seeing how we can only hear up to 20 kHz, the 1 Mhz bandwidth of APx555 is ample for any measurement whatsoever related to what we hear. There is no excuse for using Picoscope other than save money. Given the fact that MSB sells DACs that costs way above an APx555 costs, they really have no business not using it for such measurements.

 

[Music1969]

View attachment 40938

View attachment 40939

@JohnYang1997 can you comment on these results, since you asked for this test?

Anything good/bad about these results? I don't fully understand what this was testing but am interested in understanding the results.

 

[Music1969]

The developer told me very lightly: Spring2's measurement is indeed better, but you should measure my new product: May

Will you also measure Spring2 Level 2?

 

[Veri]

@JohnYang1997 can you comment on these results, since you asked for this test?

Anything good/bad about these results? I don't fully understand what this was testing but am interested in understanding the results.

If that was in NOS mode the 'grass' you see is because of the 44kHz sampling without any oversampling, which is why 192kHz is so much cleaner there. Non oversampling DACs are generally a mess measurement-wise. They do not adhere to sampling theorem.

 

[WolfX-700]

Will you also measure Spring2 Level 2?

I will, i bought one

 

[Music1969]

If that was in NOS mode the 'grass' you see is because of the 44kHz sampling without any oversampling, which is why 192kHz is so much cleaner there.

@WolfX-700 can you confirm if this was with NOS mode?

 

[YSC]

I found some Denafrips Terminator's not very authoritative measurement. I don't think it is comparable to May.

Of course I can measure it if I have a chance. I will make a more careful comparison.

View attachment 40859

and：

AP test chart from this website：https://6moons.com/audioreviews2/denafrips/1.html

I did a similar test and adjusted the scale to be basically the same：

May is the red line

View attachment 40862

Sidetrack a bit, do you have a measurement of the Spring or spring 2 vs. the may and terminator to show?