English Amiga Board Amiga Lore


Go Back   English Amiga Board > Main > Amiga scene

 
 
Thread Tools
Old 12 October 2016, 19:14   #41
meynaf
68k wisdom
meynaf's Avatar
 
Join Date: Nov 2007
Location: Lyon (France)
Age: 44
Posts: 2,362
Quote:
Originally Posted by pandy71 View Post
Well it looks like you not confused but OK let assume that 8 bit DAC is 8 1-bit DAC stacked together (it is not like this from circuit perspective and they are various circuit topologies) - what is source for multibit DAC nonlinearities accrodingly to you ? - as a coder you see only numbers and registers assuming that DAC is plain and linear extension where in real life analogy between arithmetic in ALU and real electric current is not direct.
I know how SID and AY tricks works but how this is linked with Paula?
The AY trick is same as 14bit on Paula - combine channels to get more output levels.
I do not know what is the source of nonlinearities from the electric perspective, but you can probably explain this, huh ?


Quote:
Originally Posted by pandy71 View Post
Problem is that you didn't explain anything - side to this you are ignoring fact that calibration assume less bits than in theory available (did you perform re-quantization on 8/6 sample part? - asked you about algorithm used in calibration)
Of course calibration assumes less bits than available in theory. But in theory we have 16 bits available !


Quote:
Originally Posted by pandy71 View Post
You didn't explained anything and this is problem from my perspective as you expect me to believe when i'm trying to understand.
How output is different on final output when compared to input signal (assumption is 16 bit samples at input) - please briefly describe steps from input to output.
Steps from input (2x 16bit) to output (4x 8bit) are :
Code:
 lea d14b+$10000,a3
.loop
 move.w (a0)+,d0        ; left
 move.w (a0)+,d1        ; right
 move.w (a3,d0.w*2),d0
 move.w (a3,d1.w*2),d1
 move.b d0,(a2)+        ; low left
 lsr.w #8,d0
 move.b d0,(a1)+        ; high left
 move.b d1,(a6)+        ; low right
 lsr.w #8,d1
 move.b d1,(a4)+        ; high right
 subq.l #1,d2
 bgt.s .loop
 rts
The steps to build the table are quite complicated and i can't tell how exactly it's done.


Quote:
Originally Posted by pandy71 View Post
Understand how to achieve real 14 bit DAC accuracy on Paula.
This has been explained many times, sorry if you don't get it.


Quote:
Originally Posted by pandy71 View Post
How there can be more than 14 usable in Paula with calibration - please explain as you didn't explain anything.
My theoretical level is that on Paula with calibration you may have between 10 and 12 bit DAC accuracy - you can fit even 64 bit data but at the end it will not more than 12 bits.
There are 65536 possible output levels, of which many are not useful. At the end we get something like 16384, maybe even a little more.
What is so hard to understand here ???


Quote:
Originally Posted by pandy71 View Post
I can tell you when you connecting Amiga to amplifier input without any signal as power plane is so noisy in Amiga... And there is substantial difference between real audio system capable to deliver real 16 bit performance and Amiga.
Oddly, i've found the exact opposite.
I connected an Amiga to a very powerful amp where the smallest fly's fart would have blown the loudspeakers away. I just got beautiful silence (try this with an Atari ST and you'll understand what noise really means !).
If your Amiga has noise then it's because you removed the shielding and/or added noisy components. Else an Amiga has much less electronic noise than the average PC.
So yes, that "real audio system"'s D/A are perhaps 16bit, but it doesn't mean it provides better audio experience.


Quote:
Originally Posted by robinsonb5 View Post
I think the point Meynaf's trying to make is that when two channels are combined to do 14-bit, the channel that's at 1/64 volume is still an eight-bit DAC, and some of its range overlaps the range of the full-volume channel.

This means we can input 65536 unique codes into the combined DAC, rather than the 16384 that a true 14-bit DAC could accept.
Yes, this is the point.


Quote:
Originally Posted by robinsonb5 View Post
It's highly likely that non-linearities in the DACs mean that the vast majority of those 65536 codes give a unique output level, and with careful calibration (that the current calibration software may well not be doing sufficiently well) it may be possible to achieve higher than 14-bit resolution, but only in very specific (and not very useful!) parts of the range.
Exactly how many useful values we get out of 65536 is something i don't know. I doubt it's just 4096 though.
meynaf is offline  
AdSense AdSense  
Old 12 October 2016, 20:15   #42
pandy71
Registered User
 
Join Date: Jun 2010
Location: PL
Posts: 1,526
Quote:
Originally Posted by meynaf View Post
The AY trick is same as 14bit on Paula - combine channels to get more output levels.
Nope it is not the same as Paula using Nyquist-Shanon and PCM and AY is non-PCM so there is tricky apporach for searching particular combination that corespond to linear sample value, SID is even more funny as they use internal triangle waveform to create 12 bit samples (of course they can use attenuator to form crude DAC)

Quote:
Originally Posted by meynaf View Post
I do not know what is the source of nonlinearities from the electric perspective, but you can probably explain this, huh ?
There is plenty of sources, power supply quality is one of factors as it is used indirectly (to power Vref) or directly (as Vref itself).
But if you are really interested there is lot of papers explaining common non-linearity problems in DAC - i will recommend for example http://www.analog.com/library/analog...erters%20F.pdf to see how various non-linearity can be visible in typical DAC .
But generally 8 bit DAC is not simple stacked 8 1 bit DAC's - if this was so simple then decent 16 - 20 bit monolithic DAC will be not so costly - DAC like PCM63 cost around 50 - 80$ depends on grade... (and it consist two 19 bit DAC connected).

Quote:
Originally Posted by meynaf View Post
Of course calibration assumes less bits than available in theory. But in theory we have 16 bits available !
Nope both partially overlapping so there is maybe 65535 but only 16383 are usable and after calibration there is even less.




Quote:
Originally Posted by meynaf View Post
The steps to build the table are quite complicated and i can't tell how exactly it's done.
Thx for code, i was really interested to see what kind of technique is used for calibration - i have source for calibration software but it cover various system aspects and i'm not a developer



Quote:
Originally Posted by meynaf View Post
This has been explained many times, sorry if you don't get it.
I disagree but OK - it is very important for me to tell you that i was curious and it was not personal and i respect you very much and i'm not arguing with your coder skills.


Quote:
Originally Posted by meynaf View Post
There are 65536 possible output levels, of which many are not useful. At the end we get something like 16384, maybe even a little more.
What is so hard to understand here ???
Nothing but more or less we are closer to one of my first statement - IMHO 14 bit Paula will be after calibration close to 10 - 12 bit depends on calibration accuracy, DAC accuracy and power supply quality.

Quote:
Originally Posted by meynaf View Post
Oddly, i've found the exact opposite.
I connected an Amiga to a very powerful amp where the smallest fly's fart would have blown the loudspeakers away. I just got beautiful silence (try this with an Atari ST and you'll understand what noise really means !).
If your Amiga has noise then it's because you removed the shielding and/or added noisy components. Else an Amiga has much less electronic noise than the average PC.
Which Amiga (mostly motherboard revision).
In modern PC analog audio power section design is usually very well designed (separate analog ground plane, linear regulator, way better Vref etc)

Quote:
Originally Posted by meynaf View Post
So yes, that "real audio system"'s D/A are perhaps 16bit, but it doesn't mean it provides better audio experience.
Well trust me - most of them, even cheap players from China can provide 92 - 98dB SNR.


Quote:
Originally Posted by meynaf View Post
Yes, this is the point.
I understand where is difference between us - you see this purelly from arithmetic perspective and for you this is just code that operate in deterministic way and always provide same result for same data - from my perspective electric current is not so simple. Please understand difference between DAC resolution and DAC accuracy - this is completely different - very frequently there is specified something like ENOB https://en.wikipedia.org/wiki/Effective_number_of_bits - it is not uncommon to find for example 10 bit ADC that offer ENOB like 6.4 bit and there is nothing wrong - this is nature of electronics.
http://www.analog.com/media/en/train...als/MT-003.pdf


Quote:
Originally Posted by meynaf View Post
Exactly how many useful values we get out of 65536 is something i don't know. I doubt it's just 4096 though.
It may look very bad but real 12 bit is IMHO very good result.
Attached Files
File Type: c Calibrate.c (25.8 KB, 20 views)
pandy71 is offline  
Old 12 October 2016, 21:08   #43
meynaf
68k wisdom
meynaf's Avatar
 
Join Date: Nov 2007
Location: Lyon (France)
Age: 44
Posts: 2,362
Quote:
Originally Posted by pandy71 View Post
Nope it is not the same as Paula using Nyquist-Shanon and PCM and AY is non-PCM so there is tricky apporach for searching particular combination that corespond to linear sample value, SID is even more funny as they use internal triangle waveform to create 12 bit samples (of course they can use attenuator to form crude DAC)
I don't care about theorems. The Amiga outputs PCM and AY can be turned to output PCM as well, even though it's not designed for that.
Said otherwise, i care more about real life usage than pure theory.


Quote:
Originally Posted by pandy71 View Post
Nope both partially overlapping so there is maybe 65535 but only 16383 are usable and after calibration there is even less.
Here you take as granted that we go to 14bit before calibration.
This is just not the case !


Quote:
Originally Posted by pandy71 View Post
Nothing but more or less we are closer to one of my first statement - IMHO 14 bit Paula will be after calibration close to 10 - 12 bit depends on calibration accuracy, DAC accuracy and power supply quality.
Same problem as above. We would perhaps get 10-12 bits if we took 14 bit and then calibrate on them.
But we take 16 bits as input and calibrate just that, ending with 14 bits.


Quote:
Originally Posted by pandy71 View Post
Which Amiga (mostly motherboard revision).
It was an unexpanded A600 (labeled A300, i don't remember other details). I don't have it anymore.
I've never plugged my A1200 to something really big, but i've never heard any noise coming from it either.


Quote:
Originally Posted by pandy71 View Post
In modern PC analog audio power section design is usually very well designed (separate analog ground plane, linear regulator, way better Vref etc)
But nevertheless the amount of noise it receives from other components is way bigger...


Quote:
Originally Posted by pandy71 View Post
Well trust me - most of them, even cheap players from China can provide 92 - 98dB SNR.
Yeah, on the paper. But the mere fact they are 16 bit doesn't mean they really provide 96dB. They might have very poor signal/noise ratio and end up with 12 bit quality.


Quote:
Originally Posted by pandy71 View Post
I understand where is difference between us - you see this purelly from arithmetic perspective and for you this is just code that operate in deterministic way and always provide same result for same data - from my perspective electric current is not so simple. Please understand difference between DAC resolution and DAC accuracy - this is completely different - very frequently there is specified something like ENOB https://en.wikipedia.org/wiki/Effective_number_of_bits - it is not uncommon to find for example 10 bit ADC that offer ENOB like 6.4 bit and there is nothing wrong - this is nature of electronics.
http://www.analog.com/media/en/train...als/MT-003.pdf
And ? Calibration is there precisely to raise accuracy, not resolution.
In addition, don't forget that the sound doesn't come directly from the DAC, there is filtering hardware behind. And what matters for me is the final output.


Quote:
Originally Posted by pandy71 View Post
It may look very bad but real 12 bit is IMHO very good result.
If so, why the heck would true 16-bit DAC be better ? They're not linear either.
meynaf is offline  
Old 13 October 2016, 01:31   #44
pandy71
Registered User
 
Join Date: Jun 2010
Location: PL
Posts: 1,526
Quote:
Originally Posted by meynaf View Post
I don't care about theorems. The Amiga outputs PCM and AY can be turned to output PCM as well, even though it's not designed for that.
Said otherwise, i care more about real life usage than pure theory.
But they not work the same as such they can't be used as comparable techniques... at least based on description for https://www.msx.org/forum/developmen...mples-poor-psg - there is no PCM code - there are transition in internal state machine so you not converting sample values but instead you folding waveforms...

I will not comment your opinion about theory.

Quote:
Originally Posted by meynaf View Post
Here you take as granted that we go to 14bit before calibration.
This is just not the case !
Even if you have 65535 samples some of them are lost due overlapping between DAC (6 bit attenuator)

Quote:
Originally Posted by meynaf View Post
Same problem as above. We would perhaps get 10-12 bits if we took 14 bit and then calibrate on them.
But we take 16 bits as input and calibrate just that, ending with 14 bits.
Well... not if you have 14 bit DAC then you don't need to calibrate it - if you need calibrate DAC then you no longer have 14 bit...


Quote:
Originally Posted by meynaf View Post
It was an unexpanded A600 (labeled A300, i don't remember other details). I don't have it anymore.
I've never plugged my A1200 to something really big, but i've never heard any noise coming from it either.
Perhaps we have different perception of noise.

Quote:
Originally Posted by meynaf View Post
But nevertheless the amount of noise it receives from other components is way bigger...
Other compnents ?

Quote:
Originally Posted by meynaf View Post
Yeah, on the paper. But the mere fact they are 16 bit doesn't mean they really provide 96dB. They might have very poor signal/noise ratio and end up with 12 bit quality.
Measured not on paper... nowadays this is not a problem.

Quote:
Originally Posted by meynaf View Post
And ? Calibration is there precisely to raise accuracy, not resolution.
In addition, don't forget that the sound doesn't come directly from the DAC, there is filtering hardware behind. And what matters for me is the final output.
At a cost of resolution... other factors are also important as such we going back to what started this discussion - 14 bit resolution maybe real life maximum 12 bit accuracy or less.

Quote:
Originally Posted by meynaf View Post
If so, why the heck would true 16-bit DAC be better ? They're not linear either.
Real 16 bit DAC has 16 bit linearity and this is significant difference between 2x 8 bit DAC and 1 16 bit DAC.
pandy71 is offline  
Old 13 October 2016, 09:16   #45
meynaf
68k wisdom
meynaf's Avatar
 
Join Date: Nov 2007
Location: Lyon (France)
Age: 44
Posts: 2,362
Quote:
Originally Posted by pandy71 View Post
But they not work the same as such they can't be used as comparable techniques... at least based on description for https://www.msx.org/forum/developmen...mples-poor-psg - there is no PCM code - there are transition in internal state machine so you not converting sample values but instead you folding waveforms...
Appears you didn't get it. How it works is simple : it finds three 4-bit values for every wanted output level (there are 256), and attempts to remove the level transition "spikes" (due to the fact the cpu is too slow) by choosing in which order the three levels are changed.


Quote:
Originally Posted by pandy71 View Post
Even if you have 65535 samples some of them are lost due overlapping between DAC (6 bit attenuator)
Of course some of them are lost, but not as many as 60000.


Quote:
Originally Posted by pandy71 View Post
Well... not if you have 14 bit DAC then you don't need to calibrate it - if you need calibrate DAC then you no longer have 14 bit...
We do not start with 14 bits ! We start with 16 !


Quote:
Originally Posted by pandy71 View Post
Perhaps we have different perception of noise.
If you care about noise that can't be heard at all, maybe.


Quote:
Originally Posted by pandy71 View Post
Other compnents ?
Things making lots of electromagnetic garbage, like pc cpu.


Quote:
Originally Posted by pandy71 View Post
Measured not on paper... nowadays this is not a problem.
Isolated from any complete computer then measured, doesn't count.
The D/A may be the best in the world, if its output gets parasite signal it won't help.


Quote:
Originally Posted by pandy71 View Post
At a cost of resolution... other factors are also important as such we going back to what started this discussion - 14 bit resolution maybe real life maximum 12 bit accuracy or less.
How many times will i have to repeat you that the resolution isn't 14 bit but 16 ???


Quote:
Originally Posted by pandy71 View Post
Real 16 bit DAC has 16 bit linearity and this is significant difference between 2x 8 bit DAC and 1 16 bit DAC.
Oh, so you mean that our poor 8 bit DAC aren't linear but nowadays 16 bit ones are, even cheapo chinese ones ?
meynaf is offline  
Old 13 October 2016, 12:13   #46
paul1981
Registered User

paul1981's Avatar
 
Join Date: Oct 2013
Location: England
Posts: 121
I've connected A1200's and A600's to various setups, one a very good surround system and they're clean...this is with bog standard Amiga PSU's too. If someone has noise, hum or distortion my first guess would be leaking caps, particularly audio output caps.

On the other hand I've owned laptops that are literally unusable on the audio outputs, buzzing, clicking, humming various low frequency crap due to self interference!
paul1981 is offline  
Old 13 October 2016, 12:46   #47
vulture
Registered User
 
Join Date: Oct 2007
Location: Athens , Greece
Posts: 661
@paul1981

that's pretty much my experience as well, my miggies are silent, even ones with an accelerator and no rf shield inside. I've to say that whatever audio noise trouble I've had with laptops have been eliminated by using a better PSU though. I've a good quality universal psu in particular that has saved me from audio noise every single time.
vulture is offline  
Old 13 October 2016, 14:01   #48
pandy71
Registered User
 
Join Date: Jun 2010
Location: PL
Posts: 1,526
Quote:
Originally Posted by meynaf View Post
Appears you didn't get it. How it works is simple : it finds three 4-bit values for every wanted output level (there are 256), and attempts to remove the level transition "spikes" (due to the fact the cpu is too slow) by choosing in which order the three levels are changed.
Thank you for explanation - so 3x 4 bit DAC are required to recreate sub 8 bit DAC... it is quite similar to what i pointed - can't stack 2x 8 bit DAC and call them 16 or 14 bit...

Quote:
Originally Posted by meynaf View Post
Of course some of them are lost, but not as many as 60000.
How do you know - did you checking every code word or only some of them - perhaps only 2 LSB from upper part and 2 MSB from lower part?
To linearize DAC you need to check every code, also you need to address dynamic distortion of a DAC - this prevent simplistic approach stacking multiple low accuracy converters to form single high accuracy (but you may improve dynamics by paralleling multiple DAC's - each doubling gives you something like 3dB improved dynamics).

Reducing level by 6.0203dB means you loosing half of level values... so from 65356 there is only 32768 left - it is nothing wrong as 1 bit giving you 6.0203dB SNR but 2 bits giving you 12.0406dB SNR (so from 2 values you have now 4 values). So yes, there can be only 1500 - 5000 useful sample values in 2x 8 bit DAC stacked after calibration and this is enough to drastically improve quality as single DAC may have something like 150 - 200 useful sample values - see this from this point not inherently stable digital arithmetic...

Quote:
Originally Posted by meynaf View Post
We do not start with 14 bits ! We start with 16 !
We start with 16 but this means nothing as they overlap i.e. same bits represent same signal level (2 LSB from upper 8 bit are covered by 2 MSB's from lower 8 bit part - AUDxVOL on lower half is set to fixed -36dB attenuation in analogue domain and thanks to PWM this it is stable and almost perfect as such less problems for calibration).

Quote:
Originally Posted by meynaf View Post
If you care about noise that can't be heard at all, maybe.
Well... this is called psychoacoustic and that's why every human hear slightly different.


Quote:
Originally Posted by meynaf View Post
Things making lots of electromagnetic garbage, like pc cpu.
And this is confirmation for things already pointed by me - with default Amiga M/B design 14 bit quality is unrealistic, i would rather agree for something between 10 and 12 bits maximum.

Quote:
Originally Posted by meynaf View Post
Isolated from any complete computer then measured, doesn't count.
The D/A may be the best in the world, if its output gets parasite signal it won't help.
Why isolated? You always testing whole system as even software quirks may change overall result.
Yes, it is true, DAC can be best but bad application of DAC will ruin everything (and this is another thing pointed by me that Amiga IC Paula design is probably far from best DAC design as designers need to squeeze lot of audio not related logic within same chip).


Quote:
Originally Posted by meynaf View Post
How many times will i have to repeat you that the resolution isn't 14 bit but 16 ???
How many times i need to repeat that with 8 bit on one DAC and 8 bit in another with attentuation -35.9868109891dB there are values that overlapped so finally you will have 14 not 16 bits.
And side to this how many times i need to repeat to not use terms resolution as accuracy - resolution of DAC can be even 32 bits but it may have less than few bits accuracy if you put lot of bad resistors inside.
Imagine that you building own 16 bit DAC (like this one https://hifiduino.wordpress.com/2014...he-rest-of-us/ ) and you using +-5 or +-10% resistors - at the end what quality you expect at the DAC output...?

And what kind of quality (tolerance) those internal resistors have in Paula (10 bit? maybe 12? or perhaps 14?) - this was my question from very beginning - what kind of quality can be expected from for example 8364R4 - i seriously doubt that 8 bit DAC in Paula has better accuracy than +-0.5LSB - it may be +-0.75LSB or +-1 LSB.

Once again - fact that you feeding to DAC register value 200 doesn't mean that at the DAC analog output you will receive voltage equal to Vref*(200/255) as it depends on DAC quality and can be for example equal to code 198 or 203 or for example 201.76. This should be clear for everyone - for typical low quality DAC errors can be between +-4LSB or low as +-0.25LSB for good DAC.

Quote:
Originally Posted by meynaf View Post
Oh, so you mean that our poor 8 bit DAC aren't linear but nowadays 16 bit ones are, even cheapo chinese ones ?
So i mean that 8 bit DAC has usually 8 bit (or less) linearity and 16 bit DAC usually have 16 bit linearity (or less) - even if you link together multiple Opel Corsa you will never create Formula 1 race car (even if from your perspective combined power of all Corsa engines will be same as in F1 engine) and China is not guilty of this - they can produce good and bad quality - all depends from customer (his requirements and his quality control).
pandy71 is offline  
Old 13 October 2016, 15:41   #49
meynaf
68k wisdom
meynaf's Avatar
 
Join Date: Nov 2007
Location: Lyon (France)
Age: 44
Posts: 2,362
Quote:
Originally Posted by pandy71 View Post
Thank you for explanation - so 3x 4 bit DAC are required to recreate sub 8 bit DAC... it is quite similar to what i pointed - can't stack 2x 8 bit DAC and call them 16 or 14 bit...
The situation isn't the same. You forget that these 2x 8 bit DAC have 6-bit volume control.


Quote:
Originally Posted by pandy71 View Post
How do you know - did you checking every code word or only some of them - perhaps only 2 LSB from upper part and 2 MSB from lower part?
To linearize DAC you need to check every code, also you need to address dynamic distortion of a DAC - this prevent simplistic approach stacking multiple low accuracy converters to form single high accuracy (but you may improve dynamics by paralleling multiple DAC's - each doubling gives you something like 3dB improved dynamics).
Yes, checking code words.
What do you think the calibration program is doing ?


Quote:
Originally Posted by pandy71 View Post
Reducing level by 6.0203dB means you loosing half of level values... so from 65356 there is only 32768 left - it is nothing wrong as 1 bit giving you 6.0203dB SNR but 2 bits giving you 12.0406dB SNR (so from 2 values you have now 4 values). So yes, there can be only 1500 - 5000 useful sample values in 2x 8 bit DAC stacked after calibration and this is enough to drastically improve quality as single DAC may have something like 150 - 200 useful sample values - see this from this point not inherently stable digital arithmetic...
Again, you seem to forget that one of the DAC has its output shifted 6 times.
Even if there are only 150 useful sample values - which i seriously doubt - the first sample (8 bit) would have 150, the other (6 bit) 150/4, and that would lead us to 150*(150/4) =5625 values at the end (which is a minimum, for 200 it goes up to 10000), that's many more than your 1500-5000 claim here.

Btw. If we reduce to 14 bits before calibrating, it implies that non-calibrated 14bit has more unique output levels than calibrated...


Quote:
Originally Posted by pandy71 View Post
We start with 16 but this means nothing as they overlap i.e. same bits represent same signal level (2 LSB from upper 8 bit are covered by 2 MSB's from lower 8 bit part - AUDxVOL on lower half is set to fixed -36dB attenuation in analogue domain and thanks to PWM this it is stable and almost perfect as such less problems for calibration).
They do not overlap. Same bits don't represent same signal level. If they did, they would be called linear.


Quote:
Originally Posted by pandy71 View Post
Well... this is called psychoacoustic and that's why every human hear slightly different.
This has nothing to do here. Other ppl confirmed here that the noise level of our amigas is low.


Quote:
Originally Posted by pandy71 View Post
And this is confirmation for things already pointed by me - with default Amiga M/B design 14 bit quality is unrealistic, i would rather agree for something between 10 and 12 bits maximum.
You see a confirmation where there is none. Components on the miggy don't make that much noise.


Quote:
Originally Posted by pandy71 View Post
Why isolated? You always testing whole system as even software quirks may change overall result.
Yes, it is true, DAC can be best but bad application of DAC will ruin everything (and this is another thing pointed by me that Amiga IC Paula design is probably far from best DAC design as designers need to squeeze lot of audio not related logic within same chip).
Yes, isolated. You're speaking about the DAC, the DAC, and the DAC, totally forgetting that it's not the DAC that produce the 14 bit effect.


Quote:
Originally Posted by pandy71 View Post
How many times i need to repeat that with 8 bit on one DAC and 8 bit in another with attentuation -35.9868109891dB there are values that overlapped so finally you will have 14 not 16 bits.
Repeat as many times as you want, it won't change the fact that what you say here is true only if these DAC are linear, which they are not.

But you're not just saying we can't go above 14 bit. You're saying we can't even reach that. Then a 16-bit DAC fed with 14-bit data (2 low bits cleared) would give significantly better results than pseudo 14-bit on the amiga. Experience shows that it does not.

If there were so much value overlapping, the calibrated output routine wouldn't be able to use many possible values in the 65536 possible ones. How many do you think it uses ?


Quote:
Originally Posted by pandy71 View Post
And side to this how many times i need to repeat to not use terms resolution as accuracy - resolution of DAC can be even 32 bits but it may have less than few bits accuracy if you put lot of bad resistors inside.
Then define what you call accuracy. As resolution is easy to check but nothing is perfectly accurate.


Quote:
Originally Posted by pandy71 View Post
Imagine that you building own 16 bit DAC (like this one https://hifiduino.wordpress.com/2014...he-rest-of-us/ ) and you using +-5 or +-10% resistors - at the end what quality you expect at the DAC output...?

And what kind of quality (tolerance) those internal resistors have in Paula (10 bit? maybe 12? or perhaps 14?) - this was my question from very beginning - what kind of quality can be expected from for example 8364R4 - i seriously doubt that 8 bit DAC in Paula has better accuracy than +-0.5LSB - it may be +-0.75LSB or +-1 LSB.
The lowest output levels required to do 14bit aren't generated by Paula.
Internal resistors, as you call them, don't need more than 8 bit quality. Paula outputs 8 bit, and doesn't need to output anything else.


Quote:
Originally Posted by pandy71 View Post
Once again - fact that you feeding to DAC register value 200 doesn't mean that at the DAC analog output you will receive voltage equal to Vref*(200/255) as it depends on DAC quality and can be for example equal to code 198 or 203 or for example 201.76. This should be clear for everyone - for typical low quality DAC errors can be between +-4LSB or low as +-0.25LSB for good DAC.
And this is exactly the reason why more than 16384 output values can be obtained.


Quote:
Originally Posted by pandy71 View Post
So i mean that 8 bit DAC has usually 8 bit (or less) linearity and 16 bit DAC usually have 16 bit linearity (or less) - even if you link together multiple Opel Corsa you will never create Formula 1 race car (even if from your perspective combined power of all Corsa engines will be same as in F1 engine) and China is not guilty of this - they can produce good and bad quality - all depends from customer (his requirements and his quality control).
This implies that a 16-bit DAC would play 8-bit samples (that have been zero-extended to 16 bits) with better quality than a 8-bit DAC.

Accurary is never perfect, it is always within a tolerance level.
Where one expects 1V output, it may be 1.00033V or 0.99995V.
If you have 44100Hz samples it will never be exactly that, for the simple reason no clock is perfect.
So "14 bit accuracy" is meaningless, or you have to define it.

You may pretend accuracy is below 14bit - i can't prove you wrong. Doing so would require precise measurements and i can't do that - i guess you can't either. Perhaps someone else should, settling the matter once and for all ('coz it's starting to be a bit long, in case you haven't noticed).

But if you pretend resolution drops below 14bit - it can be proven right or wrong. And you seem to do so, because you said "only 16383 are usable and after calibration there is even less".
meynaf is offline  
Old 13 October 2016, 16:51   #50
Thorham
Computer Nerd

Thorham's Avatar
 
Join Date: Sep 2007
Location: Rotterdam/Netherlands
Age: 41
Posts: 2,960
All this theory is very interesting, but it has to be put to the test, or it means nothing. In the end it doesn't even matter. What matters is that when I play high quality WAV files on my A1200 they sound pretty damned good.

Quote:
Originally Posted by meynaf View Post
Sorry if i've been rude, my friend.
No worries mate
Thorham is offline  
Old 13 October 2016, 16:58   #51
meynaf
68k wisdom
meynaf's Avatar
 
Join Date: Nov 2007
Location: Lyon (France)
Age: 44
Posts: 2,362
Quote:
Originally Posted by Thorham View Post
All this theory is very interesting, but it has to be put to the test, or it means nothing. In the end it doesn't even matter. What matters is that when I play high quality WAV files on my A1200 they sound pretty damned good.
meynaf is offline  
Old 13 October 2016, 16:59   #52
pandy71
Registered User
 
Join Date: Jun 2010
Location: PL
Posts: 1,526
Quote:
Originally Posted by meynaf View Post
The situation isn't the same. You forget that these 2x 8 bit DAC have 6-bit volume control.
Exactly however you insisted to use AY nonuniform log scale DAC as comparable to Amiga and due internal AY design those DAC can be stacked (added) when Amiga you can introduce only fixed attenuation to lower part.

Quote:
Originally Posted by meynaf View Post
Yes, checking code words.
What do you think the calibration program is doing ?
Definitely not checking all codes as to do this you must provide stimulus signal that use all possible code values... all calibration doing is checking two overlapping bits (IMHO).

Quote:
Originally Posted by meynaf View Post
Again, you seem to forget that one of the DAC has its output shifted 6 times.
Even if there are only 150 useful sample values - which i seriously doubt - the first sample (8 bit) would have 150, the other (6 bit) 150/4, and that would lead us to 150*(150/4) =5625 values at the end (which is a minimum, for 200 it goes up to 10000), that's many more than your 1500-5000 claim here.
It is 150 useful for full scale but nominally rarely sample values can utilize full scale so finally it is always less... and my claim covers all possible problems in analog path.

Quote:
Originally Posted by meynaf View Post
Btw. If we reduce to 14 bits before calibrating, it implies that non-calibrated 14bit has more unique output levels than calibrated...
It may have non-unique output levels but it doesn't be good (as without calibration overall error can be larger) - once again you extending analogies from arithmetic in ALU to DAC which is not justified.

Quote:
Originally Posted by meynaf View Post
They do not overlap. Same bits don't represent same signal level. If they did, they would be called linear.
So there is no lower 8 bit but only 6 - you have 2 overlapping bits - attenuator introduce 6 bit attenuation so in analog voltage scale you shifting 8 bit low by 6 bits.

Quote:
Originally Posted by meynaf View Post
This has nothing to do here. Other ppl confirmed here that the noise level of our amigas is low.
Glad that everyone can express his opinion about perceived Amiga noise level.

Quote:
Originally Posted by meynaf View Post
You see a confirmation where there is none. Components on the miggy don't make that much noise.
Same as above - i respect your opinion.

Quote:
Originally Posted by meynaf View Post
Yes, isolated. You're speaking about the DAC, the DAC, and the DAC, totally forgetting that it's not the DAC that produce the 14 bit effect.
Nope - i'm trying to cover whole analog path but you insist to speak about arithmetic operations on signal.

Quote:
Originally Posted by meynaf View Post
Repeat as many times as you want, it won't change the fact that what you say here is true only if these DAC are linear, which they are not.
Well so if they are nonlinear and you correcting 2 bits how the heck suddenly 14 bit accuracy was born?

Quote:
Originally Posted by meynaf View Post
But you're not just saying we can't go above 14 bit. You're saying we can't even reach that. Then a 16-bit DAC fed with 14-bit data (2 low bits cleared) would give significantly better results than pseudo 14-bit on the amiga. Experience shows that it does not.
What kind of experience - any objective source of this claim?


Quote:
Originally Posted by meynaf View Post
If there were so much value overlapping, the calibrated output routine wouldn't be able to use many possible values in the 65536 possible ones. How many do you think it uses ?
I don't know and that's why i ask and i strongly consider to measure this (but waiting for Vampire 500).

Quote:
Originally Posted by meynaf View Post
Then define what you call accuracy. As resolution is easy to check but nothing is perfectly accurate.
We not talking about perfection only about objectively quantified parameters that precisely describe DAC conversion capabilities.

Quote:
Originally Posted by meynaf View Post
The lowest output levels required to do 14bit aren't generated by Paula.
Internal resistors, as you call them, don't need more than 8 bit quality. Paula outputs 8 bit, and doesn't need to output anything else.
Well, you can't expect 14 bit accuracy from a component that provide 8 bit accuracy (but due volume regulation design in Amiga with perfect 8 bit DAC it may work as Amiga for volume regulation use PWM and as such it is free from requantization issues - you can have 8 bit with AUDxVOL=1).

Quote:
Originally Posted by meynaf View Post
And this is exactly the reason why more than 16384 output values can be obtained.
Ok, tell me how - we have 16 bit sample, we split them on two equal 8 bit parts, upper part is feed to first DAC directly, lower 8 bit part is feed to second DAC,second DAC has enabled attenuation of 36dB (6 bits), both DAC outputs are combined in analog domain.

Quote:
Originally Posted by meynaf View Post
This implies that a 16-bit DAC would play 8-bit samples (that have been zero-extended to 16 bits) with better quality than a 8-bit DAC.
Yes, with higher accuracy - if 16 bit DAC has better performance then analogue signal will be more accurate (same rule as using for calculation extended precision and later using only most significant bits).

Quote:
Originally Posted by meynaf View Post
Accurary is never perfect, it is always within a tolerance level.
Where one expects 1V output, it may be 1.00033V or 0.99995V.
If you have 44100Hz samples it will never be exactly that, for the simple reason no clock is perfect.
So "14 bit accuracy" is meaningless, or you have to define it.
Well, once again asking you to not use word perfection - it works as 'reductio ad absurdum' as perfection is indefinite .
Please don't involve clock as this lead us nowhere. It will only start discussion about jitter and all related time domain problems related to signal conversion.

14 bit accuracy is not meaningless and can be easily defined - simplest approach is SNR [dB]= 6.02N + 1.76 (where N is number of bits but this is very simplistic).
More accurate are accuracy (quality) descriptions like this http://www.analog.com/media/en/train...als/MT-003.pdf

Quote:
Originally Posted by meynaf View Post
You may pretend accuracy is below 14bit - i can't prove you wrong. Doing so would require precise measurements and i can't do that - i guess you can't either. Perhaps someone else should, settling the matter once and for all ('coz it's starting to be a bit long, in case you haven't noticed).
It is not pretending but curiosity and doubt which is foundation for every person having rational approach to problem.
I have very precision audio analyzer side to me https://www.ap.com/analyzers-accessories/apx555/ - all i need is Amiga with proper signals (as APx555 internal generator can't be used) - to have a proper signals i need to spend some time to prepare those signals (re-sampling to native Amiga sample rate) and need to have Amiga capable to play it (that's why i waiting for V500).
However ANY modern PC card can be used as audio analyzer as usually they parameters are way higher than Amiga audio system capabilities.

Quote:
Originally Posted by meynaf View Post
But if you pretend resolution drops below 14bit - it can be proven right or wrong. And you seem to do so, because you said "only 16383 are usable and after calibration there is even less".
I have this intention especially after such hot dispute.

This may help all of us in more formalized discussion in future:
http://www2.electron.frba.utn.edu.ar...easurement.pdf
pandy71 is offline  
Old 13 October 2016, 18:00   #53
meynaf
68k wisdom
meynaf's Avatar
 
Join Date: Nov 2007
Location: Lyon (France)
Age: 44
Posts: 2,362
Quote:
Originally Posted by pandy71 View Post
Exactly however you insisted to use AY nonuniform log scale DAC as comparable to Amiga and due internal AY design those DAC can be stacked (added) when Amiga you can introduce only fixed attenuation to lower part.
My point was just about adding several channels together to get sounds that can't be made in theory.


Quote:
Originally Posted by pandy71 View Post
Definitely not checking all codes as to do this you must provide stimulus signal that use all possible code values... all calibration doing is checking two overlapping bits (IMHO).
It's clear you don't have a clue on what calibration does


Quote:
Originally Posted by pandy71 View Post
It is 150 useful for full scale but nominally rarely sample values can utilize full scale so finally it is always less... and my claim covers all possible problems in analog path.
But do we really have problems in analog path ?


Quote:
Originally Posted by pandy71 View Post
It may have non-unique output levels but it doesn't be good (as without calibration overall error can be larger) - once again you extending analogies from arithmetic in ALU to DAC which is not justified.
But you do agree that the calibrated output must lead to less than 16384 different values, and this can be verified.


Quote:
Originally Posted by pandy71 View Post
So there is no lower 8 bit but only 6 - you have 2 overlapping bits - attenuator introduce 6 bit attenuation so in analog voltage scale you shifting 8 bit low by 6 bits.
It does not introduce "6 bit attenuation", just reduces the signal by 64. $40 with a volume of 1 is NOT the same as $01 with a volume of 64.


Quote:
Originally Posted by pandy71 View Post
Glad that everyone can express his opinion about perceived Amiga noise level.
Sure, but you didn't do the experiment.


Quote:
Originally Posted by pandy71 View Post
Same as above - i respect your opinion.
It's not an opinion. It's a fact that can be checked.


Quote:
Originally Posted by pandy71 View Post
Nope - i'm trying to cover whole analog path but you insist to speak about arithmetic operations on signal.
Apparently yes, else you wouldn't talk about -36dB coming out of the paula chip.


Quote:
Originally Posted by pandy71 View Post
Well so if they are nonlinear and you correcting 2 bits how the heck suddenly 14 bit accuracy was born?
This has nothing to do with "correcting 2 bits". It's about combining two 8-bit signals to get 14 bit.


Quote:
Originally Posted by pandy71 View Post
What kind of experience - any objective source of this claim?
Why wanting external sources ? It's an experience anyone here can do.
Try playing 14 bit on any PC (uncalibrated 14 bit running on an emulator should suffice), and compare with an Amiga running software such as Play16.


Quote:
Originally Posted by pandy71 View Post
I don't know and that's why i ask and i strongly consider to measure this (but waiting for Vampire 500).
You don't know, but according to what you said, it should be less than 16384.
I can do that measurement anytime, but first you have to be clear on what to expect. If, say, it's 17000, there's something wrong somewhere, don't you think ?


Quote:
Originally Posted by pandy71 View Post
We not talking about perfection only about objectively quantified parameters that precisely describe DAC conversion capabilities.
As said, i don't care about all that DAC theoretical stuff. What matters is the final output quality, and i know for having heard it that it's certainly not just 10 bit.


Quote:
Originally Posted by pandy71 View Post
Well, you can't expect 14 bit accuracy from a component that provide 8 bit accuracy (but due volume regulation design in Amiga with perfect 8 bit DAC it may work as Amiga for volume regulation use PWM and as such it is free from requantization issues - you can have 8 bit with AUDxVOL=1).
As i already said, the component that provides the 14 bit accuracy isn't Paula.
At the end the trick is exactly the same as if we were using 8-bit video D/A and do +1 half the time - at the end we get 9 bit (if we're fast enough).


Quote:
Originally Posted by pandy71 View Post
Ok, tell me how - we have 16 bit sample, we split them on two equal 8 bit parts, upper part is feed to first DAC directly, lower 8 bit part is feed to second DAC,second DAC has enabled attenuation of 36dB (6 bits), both DAC outputs are combined in analog domain.
The top 8 bits aren't used directly. That would be too easy


Quote:
Originally Posted by pandy71 View Post
Yes, with higher accuracy - if 16 bit DAC has better performance then analogue signal will be more accurate (same rule as using for calculation extended precision and later using only most significant bits).
But in real life a PC (16 bit DAC) won't play a 8 bit sample better than an Amiga (8 bit DAC), at least not without some upsampling.
This is a fact anyone can check.


Quote:
Originally Posted by pandy71 View Post
Well, once again asking you to not use word perfection - it works as 'reductio ad absurdum' as perfection is indefinite .
Please don't involve clock as this lead us nowhere. It will only start discussion about jitter and all related time domain problems related to signal conversion.

14 bit accuracy is not meaningless and can be easily defined - simplest approach is SNR [dB]= 6.02N + 1.76 (where N is number of bits but this is very simplistic).
More accurate are accuracy (quality) descriptions like this http://www.analog.com/media/en/train...als/MT-003.pdf
Again, just theory.


Quote:
Originally Posted by pandy71 View Post
It is not pretending but curiosity and doubt which is foundation for every person having rational approach to problem.
I have very precision audio analyzer side to me https://www.ap.com/analyzers-accessories/apx555/ - all i need is Amiga with proper signals (as APx555 internal generator can't be used) - to have a proper signals i need to spend some time to prepare those signals (re-sampling to native Amiga sample rate) and need to have Amiga capable to play it (that's why i waiting for V500).
However ANY modern PC card can be used as audio analyzer as usually they parameters are way higher than Amiga audio system capabilities.
Perhaps it's better to just wait until you have the Amiga...
If you need a program able to play any wave file, ask me (if you're not locked to old 68000).


Quote:
Originally Posted by pandy71 View Post
I have this intention especially after such hot dispute.

This may help all of us in more formalized discussion in future:
http://www2.electron.frba.utn.edu.ar...easurement.pdf
I don't want a formalized discussion, you trying to bring me to your land, huh ? It doesn't work this way.

You're not exactly answering the question...

So i will ask it again, in a different way.
Let's say we load a calibration file, then build the table giving the 65536 values. How many unique value pairs do you expect to find in the table ? (this can easily be verified ; if you don't know you can at least estimate a reasonable range)
meynaf is offline  
Old 13 October 2016, 20:44   #54
pandy71
Registered User
 
Join Date: Jun 2010
Location: PL
Posts: 1,526
Quote:
Originally Posted by meynaf View Post
My point was just about adding several channels together to get sounds that can't be made in theory.
It can be made and it is used in real life circuits - i already provided you multiple examples like PCM63 where there are two 19 bit DAC connected in a way to allow them process 20 bit samples (and you think that Burr Brown guys was so stupid that they use two 19 bit converters i.e. 28 bits to convert only 20? why they not used two 10 bit converters stacked as you propose?)

Quote:
Originally Posted by meynaf View Post
It's clear you don't have a clue on what calibration does
That's why i asked you as experienced software coder for explanation, i can understand if you don't know or if you not wish to share such knowledge but then say this openly.

Quote:
Originally Posted by meynaf View Post
But do we really have problems in analog path ?
IMHO this depends from you - as you don't care too much about facts if they not fit to your opinion.

Quote:
Originally Posted by meynaf View Post
But you do agree that the calibrated output must lead to less than 16384 different values, and this can be verified.
This is quite obvious as in combined two 8 bit Paula DAC's at best we may have 16384 different values and if DAC is not perfect then it means that less than 16384 values can be available.

Quote:
Originally Posted by meynaf View Post
It does not introduce "6 bit attenuation", just reduces the signal by 64. $40 with a volume of 1 is NOT the same as $01 with a volume of 64.
Please correct me but HRM provide this kind of values -36dB is exactly 6 bits.

Quote:
Originally Posted by meynaf View Post
Sure, but you didn't do the experiment.
As i said - i've tried long time ago 14 bit on Amiga and IMHO it was not 80dB+ SNR but less - i would locate his somewhere around 60 - 70dB.

Quote:
Originally Posted by meynaf View Post
It's not an opinion. It's a fact that can be checked.
Something can be fact for you and opinion for someone else.
If you write 'checked' you refereeing to subjective or objective method?

Quote:
Originally Posted by meynaf View Post
Apparently yes, else you wouldn't talk about -36dB coming out of the paula chip.
Well this almost direct value from HRM.

Quote:
Originally Posted by meynaf View Post
This has nothing to do with "correcting 2 bits". It's about combining two 8-bit signals to get 14 bit.
So somehow splitting 16 bit on upper and lower part leading us to 14 bit sample resolution... why not 10, 12 or 16 but 14? And why this is 2 bit difference?


Quote:
Originally Posted by meynaf View Post
Why wanting external sources ? It's an experience anyone here can do.
Try playing 14 bit on any PC (uncalibrated 14 bit running on an emulator should suffice), and compare with an Amiga running software such as Play16.
Nope as common audio DAC in PC has usually 16 bit and additionally it is usually single bit converter.
If you run calibration on emulated in PC Amiga what you expect to get?

Quote:
Originally Posted by meynaf View Post
You don't know, but according to what you said, it should be less than 16384.
I can do that measurement anytime, but first you have to be clear on what to expect. If, say, it's 17000, there's something wrong somewhere, don't you think ?
I expect to see measurement results not 15500 or 18000.
And yes i will be surprised to see for example SNR like 98dB but trust me if this will be for example 82dB (as this is very close to 14 bit) it will be more then very good for such circuitry like in Amiga and i will be very happy.


Quote:
Originally Posted by meynaf View Post
As said, i don't care about all that DAC theoretical stuff. What matters is the final output quality, and i know for having heard it that it's certainly not just 10 bit.
Did you ever have opportunity to hear 10 bit DAC?


Quote:
Originally Posted by meynaf View Post
As i already said, the component that provides the 14 bit accuracy isn't Paula.
At the end the trick is exactly the same as if we were using 8-bit video D/A and do +1 half the time - at the end we get 9 bit (if we're fast enough).
It is not clear to me - are referring to temporal dithering?


Quote:
Originally Posted by meynaf View Post
The top 8 bits aren't used directly. That would be too easy
Well - if they are not used directly then it means that DAC resolution suffer and overall SNR will be worse.

Quote:
Originally Posted by meynaf View Post
But in real life a PC (16 bit DAC) won't play a 8 bit sample better than an Amiga (8 bit DAC), at least not without some upsampling.
This is a fact anyone can check.
You refereeing to subjective or objective method?

If in real life you will feed analog signal from 8 bit DAC with 8 bit accuracy (assumption decent +-0.5LSB) then same samples played by 16 bit DAC with same accuracy (i.e. +-0.5LSB) will be objectively better - not sure if this can be hear but this is subjective area and everyone may have own opinion.


Quote:
Originally Posted by meynaf View Post
Again, just theory.
The one you didn't care...


Quote:
Originally Posted by meynaf View Post
Perhaps it's better to just wait until you have the Amiga...
If you need a program able to play any wave file, ask me (if you're not locked to old 68000).
As i said, waiting for V500 then i need to find player capable to play 14 bit with and without calibration from HDD.
And I have many Amiga computers and that's why i know that CDTV was worse than CD32 in terms of audio quality.

Quote:
Originally Posted by meynaf View Post
I don't want a formalized discussion, you trying to bring me to your land, huh ? It doesn't work this way.
I've realized - you don't care about objective data.

Quote:
Originally Posted by meynaf View Post
You're not exactly answering the question...
Well... not sure - what do you mean?

Quote:
Originally Posted by meynaf View Post
So i will ask it again, in a different way.
Let's say we load a calibration file, then build the table giving the 65536 values. How many unique value pairs do you expect to find in the table ? (this can easily be verified ; if you don't know you can at least estimate a reasonable range)
Well if LUT has 65356 entries and they need to be mapped to 16384 16 bit words then i expect at best 16384 unique values (i assume that LUT is to deal with slow processing capabilities of Amiga - sample value is address for LUT and value read from LUT is feed to DAC's)
pandy71 is offline  
Old 13 October 2016, 22:18   #55
meynaf
68k wisdom
meynaf's Avatar
 
Join Date: Nov 2007
Location: Lyon (France)
Age: 44
Posts: 2,362
Quote:
Originally Posted by pandy71 View Post
It can be made and it is used in real life circuits - i already provided you multiple examples like PCM63 where there are two 19 bit DAC connected in a way to allow them process 20 bit samples (and you think that Burr Brown guys was so stupid that they use two 19 bit converters i.e. 28 bits to convert only 20? why they not used two 10 bit converters stacked as you propose?)
Of course two 8-bit together can't do 14 bit "as is". I didn't pretend they could (remember a thing called PWM).


Quote:
Originally Posted by pandy71 View Post
That's why i asked you as experienced software coder for explanation, i can understand if you don't know or if you not wish to share such knowledge but then say this openly.
I know what the calibration program does. What i don't know is exactly how the table is made - the code is unclear.


Quote:
Originally Posted by pandy71 View Post
IMHO this depends from you - as you don't care too much about facts if they not fit to your opinion.
You didn't show any fact - just theoretical stuff.


Quote:
Originally Posted by pandy71 View Post
This is quite obvious as in combined two 8 bit Paula DAC's at best we may have 16384 different values and if DAC is not perfect then it means that less than 16384 values can be available.
But we get more (see below).


Quote:
Originally Posted by pandy71 View Post
Please correct me but HRM provide this kind of values -36dB is exactly 6 bits.
Try this example in reality, in phase opposition so that they cancel each other. In theory (or from the HRM) you get only silence, but in real life you'll get a small beep.
(And this is exactly how the calibration program does it !)


Quote:
Originally Posted by pandy71 View Post
As i said - i've tried long time ago 14 bit on Amiga and IMHO it was not 80dB+ SNR but less - i would locate his somewhere around 60 - 70dB.
With what software ?


Quote:
Originally Posted by pandy71 View Post
Something can be fact for you and opinion for someone else.
If you write 'checked' you refereeing to subjective or objective method?
You can just check by yourself, choose your method.


Quote:
Originally Posted by pandy71 View Post
Well this almost direct value from HRM.
The HRM doesn't have to be especially precise.


Quote:
Originally Posted by pandy71 View Post
So somehow splitting 16 bit on upper and lower part leading us to 14 bit sample resolution... why not 10, 12 or 16 but 14? And why this is 2 bit difference?
Once again, it's not about splitting 16 bit on upper and lower part.


Quote:
Originally Posted by pandy71 View Post
Nope as common audio DAC in PC has usually 16 bit and additionally it is usually single bit converter.
You can just cut off 2 bits by software.


Quote:
Originally Posted by pandy71 View Post
If you run calibration on emulated in PC Amiga what you expect to get?
On emulators you get a flat response, as if it were perfectly linear.


Quote:
Originally Posted by pandy71 View Post
I expect to see measurement results not 15500 or 18000.
And yes i will be surprised to see for example SNR like 98dB but trust me if this will be for example 82dB (as this is very close to 14 bit) it will be more then very good for such circuitry like in Amiga and i will be very happy.
Then be happy


Quote:
Originally Posted by pandy71 View Post
Did you ever have opportunity to hear 10 bit DAC?
No, but it's easy to simulate. Just cut off the last 6 bits.
I know how 8, 14, 16 bits sound, so it's a reasonable estimate to locate 10 bit somewhere between 8 and 14...


Quote:
Originally Posted by pandy71 View Post
It is not clear to me - are referring to temporal dithering?
Seems yes. Isn't it (by its principle) very similar to PWM ?


Quote:
Originally Posted by pandy71 View Post
Well - if they are not used directly then it means that DAC resolution suffer and overall SNR will be worse.
Remember, values are irregularly spaced. The LSB can fix that.


Quote:
Originally Posted by pandy71 View Post
You refereeing to subjective or objective method?

If in real life you will feed analog signal from 8 bit DAC with 8 bit accuracy (assumption decent +-0.5LSB) then same samples played by 16 bit DAC with same accuracy (i.e. +-0.5LSB) will be objectively better - not sure if this can be hear but this is subjective area and everyone may have own opinion.
I won't go in the objective vs subjective land with you, as you seem to consider yourself objective and the others subjective (especially me).

Note : your "objective" data forgets about any possibility of having a good hardware lowpass filter. You concentrate just too much on the DAC.


Quote:
Originally Posted by pandy71 View Post
The one you didn't care...
Why would i care ?


Quote:
Originally Posted by pandy71 View Post
I've realized - you don't care about objective data.
Did you present objective data ?


Quote:
Originally Posted by pandy71 View Post
Well... not sure - what do you mean?
I mean that you didn't answer the question. Never mind, seems i've got that answer now (quoted below).


Quote:
Originally Posted by pandy71 View Post
Well if LUT has 65356 entries and they need to be mapped to 16384 16 bit words then i expect at best 16384 unique values (i assume that LUT is to deal with slow processing capabilities of Amiga - sample value is address for LUT and value read from LUT is feed to DAC's)
So, you confirm that you expect at best 16384 unique values.
But the 65536 are not mapped to 16384.

I ran a program to count them in actual data. My A1200's calibration file gave 17400 values. The one from Christian Buchner (available in the Play16 package) gave even more.
meynaf is offline  
Old 14 October 2016, 12:41   #56
paul1981
Registered User

paul1981's Avatar
 
Join Date: Oct 2013
Location: England
Posts: 121
@pandy71

You said earlier 'let's not introduce time domain problems', yet shortly afterwards you talk of downsampling to Amiga compatible rate for testing, by this I presume you mean 28836 Hz. I thought this thread was all about the quality of Paula. There is no reason not to use 44.1KHz, just use a double scan mode (play16 will require DOUBLE tooltype though). If your A500 can't open a double scan screen, you might want to upgrade it to ECS). I just don't see how you can make comparisons with cd quality if downsampling.
paul1981 is offline  
Old 14 October 2016, 12:49   #57
paul1981
Registered User

paul1981's Avatar
 
Join Date: Oct 2013
Location: England
Posts: 121
@maynaf

I had no idea there were more than one version of Cybersound. Which one was the latest and best release to use? I only know of the one inside the Play16 archive.

Edit: Ahh...I see, you mean his calibration file and not a different program.
paul1981 is offline  
Old 14 October 2016, 13:22   #58
pandy71
Registered User
 
Join Date: Jun 2010
Location: PL
Posts: 1,526
Quote:
Originally Posted by meynaf View Post
.
.
.
I started replying but at some point realized that this is pointless at least at current stage of our knowledge - some data need to be collected and analysed to create some conclusions.
Sorry for your time, please All accept my apologies for this discussion - it was to early to start it - i had questions and didn't found answers yet.

Quote:
Originally Posted by paul1981 View Post
@pandy71

You said earlier 'let's not introduce time domain problems', yet shortly afterwards you talk of downsampling to Amiga compatible rate for testing, by this I presume you mean 28836 Hz. I thought this thread was all about the quality of Paula. There is no reason not to use 44.1KHz, just use a double scan mode (play16 will require DOUBLE tooltype though). If your A500 can't open a double scan screen, you might want to upgrade it to ECS). I just don't see how you can make comparisons with cd quality if downsampling.
Downsampling (or re-sampling) has nothing to do with time domain.
Yes i will use 28604Hz (sufficiently close to nominal 28603.9919Hz) sampling rate with 16 bit samples to avoid any additional conversion on Amiga (as even in productivity closest to 44100Hz frequency is incorrect - 44336.19Hz).

I will not use any higher than standard TV scan rates(i.e. no productivity or similar video mode).

My goal is to understand capabilities of the '14 bit' audio mode and collect objective data to describe overall audio quality on Amiga.

I'm not comparing Amiga to CD as this will be highly unfair for Amiga - my intention is to use whenever it is possible techniques used for digital/analog audio measurements.

Last edited by pandy71; 14 October 2016 at 13:41.
pandy71 is offline  
Old 14 October 2016, 19:18   #59
paul1981
Registered User

paul1981's Avatar
 
Join Date: Oct 2013
Location: England
Posts: 121
The A500 has a different filter circuit to the more modern Amigas, so bear this in mind when using the filter. And true, you can't compare it to cd or sound card quality as there is no oversampling with Paula for starters, you just get the pure output. So in reality you can't really compare it to anything.

With all this test gear you're going to throw at it, and with your downsampling, I suppose there's no chance of a listening test then? :-P
paul1981 is offline  
Old 15 October 2016, 01:03   #60
pandy71
Registered User
 
Join Date: Jun 2010
Location: PL
Posts: 1,526
Quote:
Originally Posted by paul1981 View Post
The A500 has a different filter circuit to the more modern Amigas, so bear this in mind when using the filter. And true, you can't compare it to cd or sound card quality as there is no oversampling with Paula for starters, you just get the pure output. So in reality you can't really compare it to anything.
It is not so different - different is in Amiga 1000 - there is high order lowpass filter there. Every generation of Amiga has simplified filtering.

Quote:
Originally Posted by paul1981 View Post
With all this test gear you're going to throw at it, and with your downsampling, I suppose there's no chance of a listening test then? :-P
I have test signals already - generated by AP software for 16 bit 32000 then resampled to 28604 in SoX - there are 3 sets of signals, 8 bit, 16 bit and 14 bit on 16 bit (16 bit audio files, volume reduced by 4 then increased by 4). All files are stereo. 8 bit are 8svx, remain are wav. In total there is over 150MB - need to check how to compress them lossless and need to upload somewhere so anyone interested can download them and play on Amiga - I assume that PC with free software may be used as audio analyzer.

And you can compare it as we comparing analog signal.
Listening tests... anyone may propose own method for subjective testing - if you have ideas on this please share it - thank you.
pandy71 is offline  
AdSense AdSense  
 


Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)
 
Thread Tools

Similar Threads
Thread Thread Starter Forum Replies Last Post
Questions about 14 bit audio playback xxxxx Coders. Asm / Hardware 16 22 December 2014 19:30
15 bit 44 khz audio idea. Thorham Coders. General 30 07 November 2013 20:37
High Quality reproduction of Audio on 8 bit. pandy71 Amiga scene 0 01 July 2013 15:08
Using two voices for 56K sample/second audio output rate -- OCS -- No CPU or Copper mc6809e Coders. Asm / Hardware 2 28 January 2012 00:29
Simple 14 bit audio question... Thorham Coders. General 7 06 June 2010 10:55

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT +2. The time now is 06:04.


Powered by vBulletin® Version 3.8.8 Beta 1
Copyright ©2000 - 2017, vBulletin Solutions, Inc.
Page generated in 0.43165 seconds with 12 queries