Lavry Engineering Paper on Hi-Res

[Talk2Me]

"Of course one can enjoy great music on a table radio too"

 

That is not the point, and far from it. I am talking about listening with a quality 2 channel set-up with enough dynamics to simulate a near concert experience.

 

[PeterSt]

I don't think it is negative at all. It is just true. Of course, that is assuming Lavry is right (which he may not be)

 

Well, that was my point of course. It is the whole subject, so to next bring it forward as truth, seems a little strange. And with that assuming the worst, is ... negative ? But never mind, it's just a bit teasing.

 

You tell us whether he is right or wrong.

 

I tried to put it into some context here : Ratio and stuff

 

Does that help ? or isn't that what you meant ?

 

[PeterSt]

It possibly won't say a single thing (at least I myself must think about the merits), but when my 1704's are fed with 16/44.1, 1KHz -138dBFS, the signal nicely sticks is nose above the field (and a normal 8K FFT, nothing averaged).

 

No dither or anything. But, "upgraded" to 705.6KHz sample rate into 24 bits.

 

The signal itself (as a sine wave) looks nicely like a sine as always, but of course a lot of modulated noise rides on it. But still it works.

I can add that the signal itself is -3dBFS, so actually it should be at -141dBFS (leaving out the fractions per 6dB step).

 

I am not sure I expected this myself. Ok, I did not. :-)

Mind you, the original signal is 16 bits. The signal fed to the DAC should be wobbling around 1 bit only ...

So, no dither, but inherent noise might nicely provide that.

 

Also : at -96dBFS (-99dBFS) there's nothing like further harmonics visible, and the signal looks "perfect" in the FFT (now sticking out some 45dB which makes it well judgeable).

 

Whether this tells that a 24bit R2R like the 1704U-K is accurate to the last bit ... I don't think so. At least I never thought it would be (also looking at the specs).

What my differentially setup throughout here contributes ... not sure. The 2 pairs of 4 1704's all together for sure will do a few things, but them canecelling out eachothers LSBs inaccuracy - I don't think so. Or at least I don't see how.

 

Peter

 

 

 

[Mark Powell]

67 years have made me think the worst of many things. Like our local member of parliament, who has just been caught for telling the police that his ex-wife was driving when he had an accident. Not even a serious accident, but lying to the court is very serious. And he was a government minister. Now he is a government ex-minister

 

And the cheap binoculars, advertised as 'See fifteen miles'. I can see the Andromeda Galaxy, two and a half million light years away, without any binoculars at all. Not during the daytime of course

 

So I'm cynical.

 

[Jud]

In the end, it is simply about statistics, instead of having two samples for 20 kHz tone you have more than eight.

 

This. ^^

 

I've been going on a little about bit depth because it was mentioned so much in the Lavry paper. But bit depth is only part of the story in converting analog to digital and back. I wish I knew more of the mathematics involved, but will take Miska's word in this case, since plainly material recorded at very low bit depths (SACD/DSD) can be used to produce what is accepted as audiophile-quality sound.

 

(I'd be curious, Miska, about what the calculations are that you did to determine 8x sample rates are reasonably sufficient.)

 

 

 

[PeterSt]

Lavry talks of 'near 24-bit at 50-60Hz' So it will be lower at 192K.

 

I have already asked PeterSt, who makes DACs, whether it is possible today or not.

 

Regarding my last post - judge yourself, but maybe draw better conclusions than I do at this time.

But at least that is at 705.6KHz (768KHz really would not be different - better actually).

 

Of course there are some prerequisites to allow judgement in the first place. For example, when the whole lot starts to express noise at e.g. -110dB, a. I could never have seen that signal and/but b. Mr Lavry would immediately say : see ? This happens at these high rates.

But there is the difference; nothing happens at these high rates here, which is exactly what I tried to tell about in that "Ratios" post.

 

Whether it is all that important to perceive "accuracy" of any kind at these lightyears away from audible levels, is something else.

 

Peter

 

 

[PeterSt]

(I'd be curious, Miska, about what the calculations are that you did to determine 8x sample rates are reasonably sufficient.)

 

If you'd ask me, I don't think it is sufficient;

Regarding this, it may be good to know that I hunted for the 768 (705.6) not for the numbers, and not because of any math I could predict it with, but because the step from 176.4 to 352.8 made a siginificant difference sound wise (mind you, this is just about "upsampling" 16/44.1). So I thought : when that makes (still) a difference, then who guarantees me that going from 352.8 to 705.6 will not.

 

Sadly, the step from 352.8 to 705.6 made an even bigger difference (meaning with sadly : I want to go beyond that now), and it is also measureable why. I can't dig it up right away (it wouldn't tell a thing anyway), but I have lists of tables of where harmonics fall, and how the shift from 352.8 to 705.6 solves a couple of things (while nothing regarding this was solved from 176.4 to 352.8).

 

But please see : this is only about my own means of filtering for the 24 bit PCM chips, and this specific subject will be totally unrelated to SDM DACs and their means of filtering.

For the SQ result of either it is of course, but never mind that now.

 

Peter

 

[PeterSt]

I have made up my mind this morning to apply for a certain job in Germany. For 20 months only.

I will make a mess of it for sure, and by the time I must resign (which is after 20 months) I will be receiving 200K euros for pension annually, a secretary and a car, and a place to put the secetary in. Most probably the car can go in there too.

For life.

Only the cigar fails to be there.

 

So yes, I understand. Btw, I am known as the foremost biggest nagger on the planet.

The difference might be that I try to not do it by standard. My father in law does that already.

 

Oh well.

haha

 

[Talk2Me]

Translation from Peter: I just got a great job with benefits, so, I don't care what you think or hear.

 

[Jud]

I will be receiving 200K euros for pension annually, a secretary and a car, and a place to put the secretary in.

 

Put the secretary in the car and sell his/her place. ;-)

 

[Mark Powell]

Where do I apply?

 

You can put the secretary in the car, then the car in the garage. That will not need so much space as keeping them separate

 

[Jud]

And the cheap binoculars, advertised as 'See fifteen miles'. I can see the Andromeda Galaxy, two and a half million light years away, without any binoculars at all.

 

Reminds me very much of my friend who saw an ad for a sort of umbrella to hang above a satellite dish, advertised as being able to combat the effect of "rain fade" on signal reception, who said to me, "I wonder what they do about the other 28,000 miles?"

 

[Mark Powell]

28000 miles

 

Thank you for giving me the biggrst laugh I have had all day!

 

[Miska]

I'd be curious, Miska, about what the calculations are that you did to determine 8x sample rates are reasonably sufficient.

 

I base my calculation on a filter that would have proper 20 kHz passband and enough attenuation. And the impulse response would fit into length of a half-wave 20 kHz sine.

 

 

[PeterSt]

I tried to dig up something in English : http://news.xinhuanet.com/english/world/2012-02/17/c_131416841.htm

 

A successor to Wulff has to be elected by parliament within 30 days.

 

So be fast ! one week left ...

 

My story about this is not in this text, but true anyway.

 

[Jud]

Only the cigar fails to be there.

 

Was it Freud or Marx (Groucho) who said "Sometimes a cigar is just a cigar"?

 

[Talk2Me]

No, I think it was Linda Lovelace.

 

[PeterSt]

... only know her first name.

I was one of them.

 

[1audio]

ADC accuracy means different things depending on application. Its not so simple. TI's best ADC chip for accuracy http://www.ti.com/product/ads1281 is listed as having the following specs:

 

High Resolution:

130dB SNR (250SPS)

127dB SNR (500SPS)

High Accuracy:

THD: –122dB (typ), –115dB (max)

INL: 0.6ppm

 

Its also listed as a 24 bit converter for industrial applications. Notice the 250 and 500 Hz sample rates.

 

Accuracy can mean several things. Rather than try to repeat them here is a good into link:

http://en.wikipedia.org/wiki/Digital-to-analog_converter#DAC_figures_of_merit

 

Many of us in the trade have referred to real bits and marketing bits. Real bits in practice rarely exceed 20. The rest are marketing bits and have been for years. The 32 bit converters may have a 32 bit internal processing path but cannot get anywhere near that performance (or better than 24 bits) due to physical laws. Either the voltage out needs to increase (a lot) or the resistances inside need to decrease (a lot). Neither are particularly possible. Otherwise you are limited by thermal noise.

 

Sampling at higher resolution at higher sample rates is possible. At the other extreme you can get 12 bit at 1 GHz sample rates http://www.ti.com/product/ads5400 If you can make use of them. The first generation digital audio products were 12 bit and they actually worked.

 

There are fewer uses for DAC than ADC's industrially so the DAC chips are not getting as much attention but this is a 2.4 GHz DAC with 14 bit precision: http://www.ti.com/product/dac5670-sp. Miska should be able to calculate what resolution can be created using it at closer to 200 KHz sample rates. At $7K ea I'm not rushing to experiment.

 

Lavry's point is that there is a speed/resolution trade-off. I think that has been blurred in current products since the new chips all run at the same internal rates.

 

I think PeterST is trying to say that Delta Sigma "cannot deliver the goods" but has not been clear enough on how to show it in technical measurements, and claims to hear it.

 

Miska's position is almost diametrically opposite, in using Delta Sigma technology to capture and reproduce the stream. And suggests keeping it in native Delta Sigma (DSD) the whole way if possible.

 

I don't see an easy resolution to this.

 

I know Keith Johnson is not a fan of DSD but like any skilled artist knows how to get the most from the instrument at hand, whichever it may be, and that is all any of us can ask for. Using the larger dynamic range available on a 24 bit recording can make it better, but can also make it unplayable in a domestic situation. Turn it up for the quiet parts and the loud parts are are too loud, bordering on hearing damage or system damage. The engineer's challenge is to fit that into something that you can play at home and enjoy.

 

 

 

[Talk2Me]

Love your Constellation gear, Demian.

 

[PeterSt]

I think PeterST is trying to say that Delta Sigma "cannot deliver the goods" but has not been clear enough on how to show it in technical measurements, and claims to hear it.

 

Hi Demian,

 

I don't think I said that anywhere, and I hope not to have implied it either.

It is only that I measured my DAC the other day throughout for low level information, and found (a bit to my surprise) what I reported a few posts back in the thread.

 

Thermal noise will be the biggest problem for PCM/R2R chips. So, they (the resistors) may be accurate alright, but for THD YMMV per bit combination active, plus the switching noise adds up.

 

What I further said (and always talk about) is that for these PCM chips real THD figures can be measured with test signals, that being a real representative for music once the filtering is a real interpolating means. Not so with SDMs; they can only show test signal THD, while music will turn out to be nothing as accurate the THD figures show (the music influencing itself by means of the filtering).

 

*Because* I can do all ahead in PC software and the DAC does nothing, I can easily compare any filtering means (useful for PCM). So, THD figures can easily be better for the higher frequencies by means of different filtering, behavior will be similar to any DAC then, but the accuracy is lost and the sound (relatively) awful.

This is how I said "who can test this ?" (similar), implying "me only".

 

I am not saying SDM can't be accurate. I think it even can be measured but it will require some changes in hardware and things have to move to the software side of matters (for both being accurate and being able to measure the real merits). I am not up to that though.

Miska is. Or otherwise he will be way more close than I am.

 

Regards,

Peter

 

[Miska]

Currently one of the best audio ADC is ESS9102 (http://www.esstech.com/index.php?p=products_ADC). It outputs 32-bit PCM up to 384 kHz sampling rates, but the actual converter has much less bits and much higher sampling rate. Probably similar to their DACs, meaning 6-bit at 40 MHz. But with noise shaping it achieves good audio resolution, meaning that the noise floor is heavily tilted towards high frequencies.

 

As you can see from the specs, it's SNR is worth 21 bits and linearity is worth 20 bits.

 

 

[Miska]

*Because* I can do all ahead in PC software and the DAC does nothing, I can easily compare any filtering means (useful for PCM).

 

I'm just doing the same for both SDM and PCM. I don't actually like to position myself for SDM any more than for PCM. But I do consider cost and reduced complexity of SDM to be beneficial.

 

Even though I don't mandate it, there are ways to do "NOS" DAC for either PCM or SDM. The biggest difference to Peter's approach is that I don't like those being filterless, I consider analog reconstruction filter being crucial part of the design. Discrete component SDM is also much more practically feasible.

 

 

[jacobacci]

Thank you all for the great debate over the last two days. I have learned a lot about the different approaches (Peter's NOS DAC for example). From what I gather, the state of the art has moved on a bit since 2004 (60khz/24bit) but not a quantum leap. True 192khz/24bit seems to be more like 20bit. But then, who needs >120db of dynamic range at 20khz, when most of us seem to be older than twenty and our hearing is going downhill.

Thanks also to the mastering artists, who produce wonderful music with this technology. After all to me it is all about technology becoming inaudible (Barry's mic feed) and the soul of the music coming through.

 

 

 

[Jud]

I agree, with one addition that seems key to me.

 

My limited understanding of what I've read here is that ~20 bits will be the practical limit due to thermal and various other forms of noise at 192kHz, *but also at the 2x rates*, so this would not be a reason to prefer lower sample rates.