Queries on AGA Copper and 14-bit Audio

[meynaf]

Originally Posted by pandy71

Of course i need to be binary compatible - if i don't care then simply i can do things completely different.

Binary compatible doesn't mean you can't do things different. Only the programming interface needs to be the same.

Originally Posted by pandy71

If sample rate is higher than 44kHz then you can consider this as same replay freq.

By replay freq i meant sample rate, what else would that be ?
Aren't you calling sample rate what should be called bit rate ?

[paul1981]

Another thing to consider is the quality of the opamps in Amiga models. I'm quite sure they would have been chosen on price and not audio quality! There's only top notch sound cards that will have decent opamps in.

With the LED filter off, there's no filter at all is there? That'll be why it doesn't sound as good as it could (16-bit 44KHz listening tests), well that and no oversampling.

[meynaf]

Originally Posted by paul1981

With the LED filter off, there's no filter at all is there? That'll be why it doesn't sound as good as it could (16-bit 44KHz listening tests).

There is still some filtering done even with led filter off.

[paul1981]

I have an 80's cd player, Aiwa. I don't play cd's through it though because it sounds SO BAD. Not all digital gear is sparkly and perfect.
I have a clock port Prelude sound card in my A1200 and that sounds superb...I simply can't fault it.

[paul1981]

This chap knows what he's talking about regarding filtering:

http://www.iar-80.com/page25.html

[daxb]

I guess there is a big difference between processed samples from a given source and a recorded sample. You get better results when convert 16 to 8 bit or direct 8 bit calculation. There are some variables that can make 8 bit records sound bad (mostly noise added). But it depends on source quality. So I don`t see much sense in measure comparing because at the end you hear what you hear and that differs from person to person.

If you record the analog Amiga output that plays a 16 bit sample via 14 bit calibration. What do you think you can measure or what result will expected?

[pandy71]

Originally Posted by meynaf

Binary compatible doesn't mean you can't do things different. Only the programming interface needs to be the same.

Generally yes but if trying to mimic original behavior is quite important to understand some design aspects.
This is not only functionality.

Originally Posted by meynaf

By replay freq i meant sample rate, what else would that be ?
Aren't you calling sample rate what should be called bit rate ?

And it was my understanding - nope - bitrate or bandwidth is number of bits required to transport some information.

Originally Posted by paul1981

Another thing to consider is the quality of the opamps in Amiga models. I'm quite sure they would have been chosen on price and not audio quality! There's only top notch sound cards that will have decent opamps in.

With the LED filter off, there's no filter at all is there? That'll be why it doesn't sound as good as it could (16-bit 44KHz listening tests), well that and no oversampling.

This is not a big problem, more important is PCB routing and power quality.
Side to this you can easily replace older OPAMP in Amiga for something new like TLE2084 or similar.
And IMHO those OPAMP hype is a bit exaggerated - if you need HQ design then you need to go for discrete and higher voltage.

And Amiga have at least two filters - one that can't be switched of - call this filter reconstruction filter - it remove PWM from AUDxVOL.

Originally Posted by daxb

I guess there is a big difference between processed samples from a given source and a recorded sample. You get better results when convert 16 to 8 bit or direct 8 bit calculation. There are some variables that can make 8 bit records sound bad (mostly noise added). But it depends on source quality. So I don`t see much sense in measure comparing because at the end you hear what you hear and that differs from person to person.

If you record the analog Amiga output that plays a 16 bit sample via 14 bit calibration. What do you think you can measure or what result will expected?

Noise is not bad - depend on noise level - it is useful as it allow to linearize quantizer (or rather it spread quantization errors in a more uniform and not correlated with signal way).
Not sure what to expect - THD+N between 60 and 70dB maybe better...

[Stratosplat]

Originally Posted by pandy71

And IMHO those OPAMP hype is a bit exaggerated - if you need HQ design then you need to go for discrete and higher voltage...

Totally unnecessary and unlikely to outperform a high quality opamp even when running on low voltage supply rails without vast complexity and expense.

Not sure what to expect - THD+N between 60 and 70dB maybe better...

On that basis, upgrading the opamp is probably a complete waste of time as the distortion and noise it creates is most likely already at least an order of magnitude below what's present in the input signal.

What many don't seem to understand is that it's also important to ensure that an opamp's surrounding component values suit it unless it's clearly specced as a higher quality direct drop-in replacement for the original.

Not selecting a suitable opamp or not making the appropriate component value changes if necessary can often make matters far worse than just leaving it alone, although placebo effect will frequently tell someone otherwise.

Designing opamp circuits properly is a science, not a crapshoot.

[pandy71]

Originally Posted by Stratosplat

Totally unnecessary and unlikely to outperform a high quality opamp even when running on low voltage supply rails without vast complexity and expense.

Designing opamp circuits properly is a science, not a crapshoot.

True and as such most of situations OPAMP are overhyped marketing blahblah - OPAMP only make circuit simpler as they can be used almost like black box without knowing science by anyone, even by marketing people - they will not magically improve audio as usually modern OPAMP has sufficiently high parameters itself and they are frequently ruined by improper application (for example in Amiga reference voltage for DAC is created by dividing in resistor divider general digital supply voltage by half - this short way to inject power supply noise to output signal - for any analog circuitry you use separate clean power line and separate ground plane).
Amiga is also different when compared to consumer DAC (in Amiga time) as Paula feed samples with over 3MHz sample rate - this is very high sample rate and some slower OPAMP working as in Amiga case in first stage may distort signal (DAC produce series of pulses where level of pulses depends on sample value and length of pulses depends on volume set in AUDxVOL - for small values like 1 in case of 14 bit mode this may be important - also it is not clear what kind of DAC is used in Paula - current or voltage [but CD32 schematics suggest that it may be current type DAC] - current DAC need additional current to voltage conversion which can be very tricky itself with slow OPAMP).
All this may impact analog signal quality (like hearing FDD or HDD in analog audio line not uncommon).

[Stratosplat]

Yes. The worst thing about the audio output stages on an Amiga is everything else apart from the chosen opamp.

[pandy71]

Originally Posted by Stratosplat

Yes. The worst thing about the audio output stages on an Amiga is everything else apart from the chosen opamp.

Well TL074 is not so bad in terms of audio quality and as overall it could be worse...
http://www.diyaudio.com/forums/chip-...fferences.html
https://e2e.ti.com/support/amplifier...udio-equipment

But once again - yes, even if you trying to be sarcastic there are bigger issues in Amiga than OPAMP - replacing them by four high end modern ultra fancy marketing hypes OPAMP's will not brings Valhalla to your ears...

btw some of people like old uA739 sound...

-*-*-
Promised test files for Amiga - generated by AP tool (APx Waveform Generator Utility v4.2) as 32kHz, 16 bit, 2 channel, later converted by SoX (re-sampling to 28604, 8 bit, 16 bit and 14 bit, no dither applied, 8 bit version use native Amiga 8svx format).

https://ulozto.net/!KokppftoNZdU/8364-test-7z
archive 15.6MB large, after unpacking it should be around 154.6MB.

[Stratosplat]

I was actually agreeing with you. The entire audio circuitry is incredibly poorly implemented. It would be hard to have made it any worse on purpose let alone by accident. It seems to be a classic example of what happens when you trust digital design engineers to design analogue signal stages.

Preferring the sound of one opamp over another doesn't mean that it's better. The same applies with power amplifiers where many audiophiles prefer the sound of some ancient designs with clearly audible distortion over modern designs where distortion can be orders of magnitude too low to be audible.

I was talking about actually being better rather than just giving a warm fuzzy feeling of "I've changed it and it sounds different so it must be better". There's so much crap floating around on the voltage rails, reference rail and in the signal itself that the minor noise and distortion introduced by even a cheap opamp is going to be irrelevant by comparison.

[pandy71]

Originally Posted by Stratosplat

I was actually agreeing with you. The entire audio circuitry is incredibly poorly implemented. It would be hard to have made it any worse on purpose let alone by accident. It seems to be a classic example of what happens when you trust digital design engineers to design analogue signal stages.

Oh sorry, then i misunderstood emotional context - and i can't blame Paula (and/or Amiga) designers - we learned so many things about signal sampling and digital audio at the beginning of 90's - many of those things was unknown to Amiga creators and they can't be blamed as it was unknown also for science. I would say that understanding of the digital audio come to us around 1991/1992 so too late for Amiga.

Originally Posted by Stratosplat

Preferring the sound of one opamp over another doesn't mean that it's better. The same applies with power amplifiers where many audiophiles prefer the sound of some ancient designs with clearly audible distortion over modern designs where distortion can be orders of magnitude too low to be audible.

I was talking about actually being better rather than just giving a warm fuzzy feeling of "I've changed it and it sounds different so it must be better". There's so much crap floating around on the voltage rails, reference rail and in the signal itself that the minor noise and distortion introduced by even a cheap opamp is going to be irrelevant by comparison.

I don't blame people for trying to be exceptional with better than average ears but... but i like scientific studies showing how much judgment of such people is biased by their knowledge and psychological factors not related to perception of sound but for example way how things look or how big they are. Good text on this but not only: http://www.harman.com/sites/default/...dioScience.pdf so true

BLIND vs. SIGHTED TESTS – SEEING IS BELIEVING

And HQ audio whole system need to be HQ not only DAC and/or OPAMP - i can hear FDD or HDD activity in Amiga so how it can be 14 bit audio source but OK maybe i have different expectations.

[meynaf]

Originally Posted by pandy71

And HQ audio whole system need to be HQ not only DAC and/or OPAMP - i can hear FDD or HDD activity in Amiga so how it can be 14 bit audio source but OK maybe i have different expectations.

Of course you can hear FDD or HDD - they have mechanical parts
But on my a1200 with ide-cf i can't hear any disk activity, even with volume pushed.

[pandy71]

Originally Posted by meynaf

Of course you can hear FDD or HDD - they have mechanical parts
But on my a1200 with ide-cf i can't hear any disk activity, even with volume pushed.

Ha ha ha - great joke...

[vulture]

@pandy71

There must be something going on with your system to have audio interference from fdd and hdd accesses. Just for the heck of it, I plugged one of my A1200 (no RF shield, cf for hd, apollo 1230 accelerator) to my KRKs and I can't hear a thing, copying right now as I'm writing about 1gb of whdload stuff from internal cf to pcmcia. My ear to the woofer, nothing's there.

[pandy71]

Originally Posted by vulture

@pandy71

There must be something going on with your system to have audio interference from fdd and hdd accesses. Just for the heck of it, I plugged one of my A1200 (no RF shield, cf for hd, apollo 1230 accelerator) to my KRKs and I can't hear a thing, copying right now as I'm writing about 1gb of whdload stuff from internal cf to pcmcia. My ear to the woofer, nothing's there.

Connect Amiga audio RCA out to amplifer, don't play anything on audio, raise volume on amplifier, start using FDD or HDD - enjoy sound created by current ripple in speakers...

[meynaf]

Originally Posted by pandy71

Connect Amiga audio RCA out to amplifer, don't play anything on audio, raise volume on amplifier, start using FDD or HDD - enjoy sound created by current ripple in speakers...

No, nothing to be heard.
Perhaps your amiga simply needs recapping.

[pandy71]

Originally Posted by meynaf

No, nothing to be heard.
Perhaps your amiga simply needs recapping.

Can you provide FDD and/or HDD model used during test?

[jarp]

Motherboard revision may also affect. At least on some Amiga 600 motherboard revisions PCMCIA and IDE activity can be literally seen from the composite out picture. Quite annoyingly even. Some motherboard revision do have manually soldered ceramic cap to prevent this and later ones have that designed in.