NOS DAC in-band distortion

[RayDunzl]

Here's the output of my DAC running in 24/44.1 NOS mode (1kHz sine wave input):

How many bits were used in the generation of the tone?

 

[mansr]

In a recent BorderPatrol DAC review in Stereophile, JA refers to it as "aliased image energy". This makes no sense to me. My understanding is that "aliasing" occurs during analogue-to-digital conversion, where any signal above the Nyquist frequency is 'folded back' in-band. "Imaging" occurs during digital-to-analogue conversion, where any signal beyond the Nyquist frequency is 'replicated forward' out-of-band.

(I've noticed Amir using "aliasing" when talking about DACs too.)

Is JA correct to use the term "aliased image energy" for NOS DAC in-band distortion? If not, what's causing it?

Your understanding is correct. JA doesn't know what he's talking about.

 

[manisandher]

So, "aliasing" and "imaging" really shouldn't be used interchangeably.

JA and AM ought to know better .

Mani.

 

[manisandher]

How many bits were used in the generation of the tone?

24 bits. Why?

 

[RayDunzl]

24 bits. Why?

16 bit can be ugly down low...

 

[gvl]

I've seen many times images called aliases, people have problem understanding terminology but you usually know what they mean through the context.

 

[mansr]

So, "aliasing" and "imaging" really shouldn't be used interchangeably.

No more than differentiation and integration should.

I've seen many times images called aliases, people have problem understanding terminology but you usually know what they mean through the context.

Sloppy use of terminology when the meaning is clear is one thing. JA's case is different. He seems to have a very distorted understanding of how these things work.

 

[manisandher]

16 bit can be ugly down low...

My 24 bit 1kHz tone is a bit cleaner than that :



Mani.

 

[DonH56]

Ideally a 24-bit noise floor should be around -216 dBFS. In the real world it doesn't come close, and in simulation is often (usually) limited by FFT resolution (FFTs using >2^24 points are rare...) and choice of signal frequency (if not chosen properly spurs will land in the same bins, raising the noise floor).

 

[LTig]

The 22nd harmonic lies at 22000Hz. The first image of this beyond Nyquist is at 22100Hz. The difference is 100Hz.

It's definitely IMD .

Mani.

Use 1001, 1005, 1010 Hz instead of 1000 Hz and see what happens.

 

[Blumlein 88]

Use 1001, 1005, 1010 Hz instead of 1000 Hz and see what happens.

Or 1500 hz. I've taken to doing IMD measurements at 18.5 khz and 19.5 khz. That way the sum and difference from the various even and odd orders don't end up on top of each other.

 

[YSC]

I am more interested to see 20khz measurements, for NOS it should look shit

 

[pkane]

I am more interested to see 20khz measurements, for NOS it should look shit

It does. At -80dB, it looks like noise.

 

[YSC]

It does. At -80dB, it looks like noise.

then if -80db sounds likely still inaudible, the waveform is interesting though

 

[pkane]

It does. At -80dB, it looks like noise.

My scope can't resolve down to that level, it has 8-bits resolution with 12-bits software interpolation. Maybe someone with a higher resolution scope can post this. Here's 20kHz @ -70dBFS, Holo Spring NOS mode:

 

[YSC]

My scope can't resolve down to that level, it has 8-bits resolution with 12-bits software interpolation. Maybe someone with a higher resolution scope can post this. Here's 20kHz @ -70dBFS, Holo Spring NOS mode:

View attachment 48492

ok, that don't look really like a sine wave but it is not as bad as I would expect

 

[pkane]

ok, that don't look really like a sine wave but it is not as bad as I would expect

It probably looks better because of the interpolation done by the scope software

 

[KSTR]

My scope can't resolve down to that level, it has 8-bits resolution with 12-bits software interpolation. Maybe someone with a higher resolution scope can post this. Here's 20kHz @ -70dBFS [...]

20kHz @ 70dBFS (ref +19dBu) NOS waveform ("NOS" of my Adi-2 Pro):

20kHz @ 44.1kHz, 70dBFS (ref +19dBu) "standard" reconstruction waveform ("SD Sharp" of my Adi-2 Pro):

 

[manisandher]

Use 1001, 1005, 1010 Hz instead of 1000 Hz and see what happens.

Or 1500 hz.

OK, so here are the in-band FFTs for a 1.5kHz tone. First the 8x OS filtered output:

The third and fifth harmonics are the largest, both sitting at around -107dB.

And the NOS output:

The 'spikes' sit exactly 300Hz apart. (They were 100Hz apart for the 1kHz tone.)

So is this distortion (the 'spikes') down to IMD?

Mani.

 

[manisandher]

I wanted to see how bad things were at 10kHz.

With filter:

NOS:

Here, the 'spikes' are 100Hz apart. Due to IMD of images? And what's causing the modulation?

Mani.