Test Drive

[hitbyambulance]

So the pcm1773 might go away if it is not significantly better. The twl6040 will remain there anyways since it is the only chip which has an ADC for the microphone.


Something else you might want to know: the twl6040 is supported by Linux 4.0 out of the box while there is no driver for the pcm1773.

Please don't underestimate the importance of having rock solid audio and amplification on the Pyra. The Pandora blows away every cell phone and portable media player I've ever tried. It would be disappointing if the audio quality (and quantity) in the Pyra were to be in any way a step backwards from the Pandora.

Well much of how the Pandora sounds is how the audio circuitry is designed... the expensive tantalum capacitors hns mentioned most likely has more of an impact than the DAC did to the Pandora sound.

And those capacitors would be eliminated in the simple/cheaper/easier solution...?

according to the post, yes

 

[ImmaculateChimp]

So the pcm1773 might go away if it is not significantly better. The twl6040 will remain there anyways since it is the only chip which has an ADC for the microphone.


Something else you might want to know: the twl6040 is supported by Linux 4.0 out of the box while there is no driver for the pcm1773.

Please don't underestimate the importance of having rock solid audio and amplification on the Pyra. The Pandora blows away every cell phone and portable media player I've ever tried. It would be disappointing if the audio quality (and quantity) in the Pyra were to be in any way a step backwards from the Pandora.

Well much of how the Pandora sounds is how the audio circuitry is designed... the expensive tantalum capacitors hns mentioned most likely has more of an impact than the DAC did to the Pandora sound.

And those capacitors would be eliminated in the simple/cheaper/easier solution...?

according to the post, yes

How much more expensive would the pcm1773+tantalum capacitors option actually be? High quality audio isn't something you should just toss out because it's a bit more expensive. I would greatfully pay more to ensure that my Pyra has the highest quality audio possible.

 

[Binky]

I would greatfully pay more to ensure that my Pyra has the highest quality audio possible.

Yep. I'd be very supprised if the extra cost was greater than several quid - and I'd happily pay more than that for beter audio quality.

 

[codifies]

+1,000,001 !! a few quid is easily a small price to pay for good audio quality... Audio design is a black art, especially where you have copper tracks nearby going at silly frequencies, the lower harmonics of which can be in the audio range... so its much more than just a few quid, but real good design and testing...

 

[hns]

I would greatfully pay more to ensure that my Pyra has the highest quality audio possible.

Yep. I'd be very supprised if the extra cost was greater than several quid - and I'd happily pay more than that for beter audio quality.

Well, first tests did show that the pcm1773 is not better. Or in other words, the twl6040 is equally good according to data sheets and first (coarse) tests, but significantly cheaper.
Currently, we are not yet in a stage where we can test that intensively (i.e. for 1-2 weeks with several units handed out to testers who try things we don't even think about).

And, as "codifies" says, EMI reducing design is more important than a specific chip. BTW, there are less copper traces with the twl6040 where sensitive signals are exposed because the pcm1773 needs an external amplifier for the weak DAC output. So the risk for design issues is higher with the pcm1773.

But since we (like you) don't know how good it really is, we have designed the board to use both options and do a real life comparison so that you can get the best audio quality we can achieve.

 

[Letalis Sonus]

I'd really like to know what MWeston thinks about the twl6040, he did put quite some effort and audiophile knowledge into fine-tuning the whole audio circuity of the Pandora after all.

For the sake of it, here's a post in which he describes some details about what he did and what's so special about his choice of components: http://boards.openpandora.org/topic/1762-a-review-of-the-pandoras-sound-and-music-capability/#entry32907

 

[fusion_power]

Somehow, my old classic CC Pandora doesn't even sound that great. Maybe I have bad headphones (they tend to overtune high pitch sounds like ZZZZ and SSSS ) or I have  not the best music quality files, who knows. However, the internal Speakers of the Pandora are crap for sure, every cheap mobile phone sounds much better and "fuller" than these.

Every improvement there is welcome.

 

[Grench]

I would greatfully pay more to ensure that my Pyra has the highest quality audio possible.

Yep. I'd be very supprised if the extra cost was greater than several quid - and I'd happily pay more than that for beter audio quality.

Well, first tests did show that the pcm1773 is not better. Or in other words, the twl6040 is equally good according to data sheets and first (coarse) tests, but significantly cheaper.

Currently, we are not yet in a stage where we can test that intensively (i.e. for 1-2 weeks with several units handed out to testers who try things we don't even think about).


And, as "codifies" says, EMI reducing design is more important than a specific chip. BTW, there are less copper traces with the twl6040 where sensitive signals are exposed because the pcm1773 needs an external amplifier for the weak DAC output. So the risk for design issues is higher with the pcm1773.


But since we (like you) don't know how good it really is, we have designed the board to use both options and do a real life comparison so that you can get the best audio quality we can achieve.

Is the twl6040 capable of outputing the power/levels of noise-free amplification that the pcm1773 does?  That is something has always been very impressive on the Pandora - it could drive large headsets with big drivers loudly without distortion.

Here's a thought - put together three test platforms.  

1.  The Pyra board with twl6040 only

2.  The Pyra board twl6040 with the pcm1773

3.  A 1Ghz Pandora

Enclose them in a cardboard box with a 3 way switch and one headphone jack - so that the test becomes blind.

Call for volunteers in the Germany area to visit and bring in their own headphones.  Have someone on the back of the box available to adjust volume and track played.

It would be a quick way to put lots of different ears and lots of different headphones to work evaluating the sound quality while putting minimal bias into their opinions.

Yeah, that would be a lot of work and may be time consuming and 'not worth it' if you already have a solid solution.

Still, the sound quality and quantity on the Pandora is VERY impressive.  It sets a very high bar for the Pyra.  Don't consider our concern as a criticism of your design capabilities.  We're all simply very anxious for the Pyra to be exceptional in every aspect.

 

[Grench]

Somehow, my old classic CC Pandora doesn't even sound that great. Maybe I have bad headphones (they tend to overtune high pitch sounds like ZZZZ and SSSS ) or I have  not the best music quality files, who knows. However, the internal Speakers of the Pandora are crap for sure, every cheap mobile phone sounds much better and "fuller" than these.

Every improvement there is welcome.

In 2010 the external speakers on the Pandora, designed in 2008, were far better than any cell phone of it's day.  External phone speakers have simply gotten MUCH better in the last 5 years.

If you're not happy with the sound coming through your headphone port on a Pandora, you likely have bad headphones.  If you're not a 'audiophile' but want to try out very good headphones on a budget - i.e. entry level audiophile - these are a good start.

Open style - more bass, but you'll annoy your neighbors if in a cubicle environment.  $50.

http://www.amazon.com/Sennheiser-100-II-Ear-Miniheadphone-Black/dp/B002VPDOH8

Closed style - a bit less bass response, but you'll be the only on listening unless you're at concert volumes.

http://www.amazon.com/Sennheiser-PX-200-II-Headphones/dp/B002VPDOHS $60.

As for your audio track quality, use Asunder on a CD you own and record a variable bit rate MP3 with the max bit rate maxed out and see if that helps?  That or buy a high bit rate MP3 track from Amazon?  Try Audacious for playback.

The net is - if you don't like the sound you're getting out of the Pandora's headphone jack, something is wrong.

 

[Binky]

As for your audio track quality, use Asunder on a CD you own and record a variable bit rate MP3 with the max bit rate maxed out and see if that helps?

Why use mp3?

If audio quality is a priority, lossy compression is bad.

If you're a FOSS geek who buys wierd palmtop computers, patented encodings are bad.

FLAC is nice.

 

[fusion_power]

I guess my ears are to sensitive (son of a musician) so I hear "bad" sounds earlier than others. Plus, my (old) Music collection are mainly 128kbit mp3's (and some newer  196kbit ogg), not a good combination.  I used Audacious but it had isssues with umlauts in file names so I switched to DeadBeef, which also comes with a handy equalizer including presets (rare!).    Oh, I use in ear headphones by the way, these (old) ones here:

http://www.amazon.com/Creative-Labs-51MZ0085AA030-EP-630-Earphones/dp/B002EL4MXY/ref=pd_sim_e_2?ie=UTF8&refRID=01XHKYW5VTBJK1Y5A606

They tend to make "cord noises" from the cables so I really should try to get better ones. I had better audio with my fathers Sennheiser HiFi headphones but of course you can't carry these outside in public without looking silly.  So I still prefer in ears.

 

[TrashyMG]

The TWL6040 and PCM1793 are geared for different things, the TWL6040 is an all in one solution for cell phone and media devices and the PCM1793 is pretty much just a straight up DAC. the TWL6040 has a built in 8ohm 1.5W stereo speaker driver and a regular line level stereo output, Since that speaker driver is directly on the silicon I have a feeling it may be a little on the weak side compared to having a beefier external amplifier. I'm not a designer, only a hobbyist at best, so I take Nikolaus word over my own. 

I do love that the Pandora can drive my expensive headphones properly that my cell phone and electronic devices can't drive without a separate headphone amplifier. I hope the Pyra is up for the task as well.

 

[Exophase]

Pandora's PMIC (TPS65950) also has a built-in audio codec. If it was at all competitive to the current solution MWeston would have used it. He has stressed that datasheet data is nowhere near enough to evaluate audio properties, and the supporting analog circuitry makes a big difference.

TWL6040 may improve enough on TPS65950 to close the gap that was there, but I'd still be hesitant to rely on its integrated amp. Is it possible to use TWL6040 with the Pandora-style audio circuitry? Would require a board modification I'm sure, but I assume getting rid of PCM1773 (not PCM1793, that's actually a superior part) would too.

Here's the datasheet of the TWL6040: http://pdf1.alldatasheet.com/datasheet-pdf/view/424448/TI1/TWL6040.html

And Pandora's audio amp chip, TPA6110: http://www.ti.com/lit/ds/symlink/tpa6110a2.pdf

If I'm reading it correctly, TWL6040's amp actually has better THD+N than TPA6110. And its output power is actually higher as well.

TWL6040 is hardly cutting edge, it was used way back in 2011's Samsung Galaxy Nexus, but of course Pandora's PWM1773 is much older than that too. Maybe someone could do a test with Galaxy Nexus to see how it sounds. But maybe there could still be some circuit decisions that make a difference in how TWL6040 sounds even when using the integrated headphone amp.. Pandora uses an isolated power supply for example, and has taken noise into consideration with the audio circuit layout..

Would be great to get MWeston's take on all this. Going for anything even a little worse than Pandora's audio would be out of the question IMO, especially after Pyra's been advertised as having audiophile class audio.

 

[Grench]

As for your audio track quality, use Asunder on a CD you own and record a variable bit rate MP3 with the max bit rate maxed out and see if that helps?

Why use mp3?


If audio quality is a priority, lossy compression is bad.


If you're a FOSS geek who buys wierd palmtop computers, patented encodings are bad.


FLAC is nice.

I use MP3 because it's ubiquitous.  It works on anything.  The way I record in MP3, though, is essentially lossless.  I set Assunder to VBR MP3 and max out the bit rate - I can't remember off hand what that max setting is and I'm not in front of my Linux workstation right now.  The playback bitrate varies between 100 and 380.  Essentially the only thing that is getting compressed is silence.

FLAC and OOG are nice, but in recording my family CD collection (100s of CDs) I wanted the best possible play back that would work on -any- device from Linux PCs, Pandora, Windows PCs, Android phones, Android tablets, Roku, Nintendo Wii, in-vehicle USB based audio platforms, etc...  The generic portability of MP3 won on that regard - so I made an effort to make sure that it was the best quality MP3 file that it could be.  Not really worried about disk space.

I like Exophase's ideas of adding additional amplification power or cleaner amplification to the main DAC - it's a good possibility to explore anyway.  Standard cell phone amplifier - even in the Nexus - is pretty lame next to a Pandora.

 

[levi]

FWIW, MP3 won't work on Debian PCs as installed (although it's trivial to install the not-good codecs or whatever they are and get them).  I've ripped all my music to FLAC.  From there I retranscode it to ogg/mp3/whatever best codec I can find that works.

So the Pandora uses a separate DAC and amp arrangement.  It has up to three audio levels out - line, headphone and speaker (though I suspect headphone and speaker are actually the same level).  Does it have an amp for each level?  Or does the DAC actually produce line level, which is then amped up to headphone (and speaker) level?

In my experience the line level output has a significantly better s/n ratio than the further amplified headphone output.  That's not surprising - more amplification means more noise, but the fact I can easily hear noise between tracks through the headphone socket but not the line level output is noteworthy.

 

[notaz]

If I'm reading it correctly, TWL6040's amp actually has better THD+N than TPA6110. And its output power is actually higher as well.

On OMAP5 uEVM TWL6040's output is very weak (compared to pandora) and sounds bad through my headphones on my listening tests, even though ED's tests (there is an old thread somewhere) showed it's better IIRC. Something might be misconfigured in my setup though.

Pyra's been advertised as having audiophile class audio.

It has?

 

[slaeshjag]

http://pyra-handheld.com/specs.html

Under "Audiophile sound" would be one example.

 

[rSl]

if i remember correctly,


this first audio test done in the


old thread used the (default) enabled alsa dmix/softvol


conversion plugin, so the sound generated by the raw


audio devices was not tested properly.


so there is a good chance the pcm1773/amp will win


in a new shootout.

 

[matzesu]

May whe get a updatet list somethime, what you can make whit the Pyra out of the box??


I know, there a a lot things to make which whe know from the Pandora, but whit all these Chips..

 

[Exophase]

On OMAP5 uEVM TWL6040's output is very weak (compared to pandora) and sounds bad through my headphones on my listening tests, even though ED's tests (there is an old thread somewhere) showed it's better IIRC. Something might be misconfigured in my setup though.

Would be nice if someone could test TWL6040 with an external amp, but I don't know if that's possible on the uEVM.