[Q] audio ... again

[mr72]

Yes, I've read (or skimmed) most, or all, of the threads on this topic.

My NC audio is...

1. too bright, sibilant
2. lacking in bottom end
3. harsh/hissy in the top end

not talking about playback over the built-in speaker. talking about listening with high-end headphones (in my case, some Shure SE315s and also ME Electronics M6).

I'm not talking about the audio jack behaving differently according to how you insert the plug. Not talking about trying to get bluetooth audio. Not talking about Pandora sounding bad, or a lack of an equalizer on the Google Music player, etc.

Has anyone found a solution that will just fix the actual audio quality (as in, tone balance) issues?

edit:
BTW, I did find that using DSPManager (on CM7.1) it gets much better with a custom EQ setting (each band):

+1.0 +0.5 0.0 -0.5 -1.0 -1.5

The top end is still harsh sounding, but at least it's not overbalanced in this case.

This makes me think this is likely a hardware issue... too small or the wrong type of coupling cap somewhere.

 

[mr72]

Perhaps you guys are not trying to listen to music on your Nook Color, thus the lack of discussion on this thread.

After comparing to my laptop and my HTC phone, I have discovered that at 100 Hz the NC is at least -10dB compared with my phone. This is extreme! I will try to do some FFTs of this over the weekend.

I think it's a hardware problem.

For the time being I have "bass boost" on DSP Manager turned on, along with some rather extreme EQ settings. It still sounds bad, but at least it's not ear-bleedingly bad.

 

[NRGZ28]

I've noticed this too with other "lower end" devices, or devices where the engineers aren't too overly-concerned with its audio quality. They usually don't sound the best. I'm sure it's not just your imagination on this. And to think I was actually considering getting a nook as a cheap in-car entertainment device... ugh

 

[mr72]

I've noticed this too with other "lower end" devices, or devices where the engineers aren't too overly-concerned with its audio quality. They usually don't sound the best. I'm sure it's not just your imagination on this. And to think I was actually considering getting a nook as a cheap in-car entertainment device... ugh

I don't think it's because it's "low-end".

I think one of a couple of things may be at play here.

1. the NC is intended to play only through the internal speaker and/or only play audiobooks (spoken word only). So they intentionally rolled off the bottom end (and most of the midrange as well).

2. audio was not even on the minds of the engineers of the NC, so they didn't bother to check what value coupling capacitors or filter components will work correctly

3. the software/driver is programming the audio device with an extreme HPF intentionally

The reason I don't think it's just because it's "cheap" is because a $20 MP3 player will get all of this right. In fact, it's almost trivial to get this more than good enough. The rest of the NC hardware is way more well-done than this. This screen would not be on a "cheap" device where they severely cut corners on audio by oversight.

So my suspicion is that the audio behavior of the NC is quite intentional. I don't know if it's HW or SW. I seriously hope it's SW.

Glancing at the datasheet for the audio chip (http://www.ti.com/lit/ds/slas671/slas671.pdf) I notice that the audio device has automatic headset detection. This does not seem to be implemented on the NC. Maybe that's the problem. It also has two audio inputs. The more I look at this, the more I want to crack open my NC. The datasheet suggests AC coupling the headphone outputs so there is some capacitor likely in series with each of these lines. IMHO this is the most likely culprit to the NC's crappy sound, if it is a hardware problem. In this case it's only two components to swap, should be a snap. But I don't really want to solder on my NC.

The DRC (dynamic range control) does have a HPF that might have the ability to cause this crummy sound in software, but I doubt it. However, it would not surprise me for the NC developers to turn on some DRC all the time.

I guess I really need to get at the source code of whatever the driver is for the audio device, or get a busted NC so I can trace the output circuit and determine if it's a hardware issue.

Or maybe I'll just not use this to listen to music and watch movies.

 

[mr72]

OK. This is quite strange.

For a quick experiment, I did a FFT on pink noise playback from my Nook Color, my HTC Aria phone, and my computer's sound device (in this case, Tascam US-144).

Now, of course I am not playing this into HEADPHONES. In fact, the input of my US-144 "Line In" that I used to test this is AC-coupled and about 50K ohms.

Into that impedance, the output of the NC looks pretty freakin' flat. Moreso than my phone, in fact. So this is good news for you guys wanting to use a NC to plug into the Line In on some device. There is nothing wrong with the audio. It looks great.

However, since my ears, with my earphones, told me that the NC's output was anything BUT flat, I decided to do another test. This time I clipped in an 8 ohm load across the output, to see what the NC's frequency response looks like into an 8 ohm load. I did this also for my phone for comparison.

The NC's output is about -25dB at 100Hz going into an 8 ohm load (typical headphones are maybe more like 16 ohms). As you can see, this is hugely different from the 25K load output.

The verdict? The NC's output is very reactive according to what load you attach to it. Plugging in most headphones will result in a huge roll-off of the bottom end. You could plug it into a buffer or external headphone amp and it would probably sound just fine.

And for comparison, I did the same test with my HTC Aria, and the 8 ohm output looks pretty much exactly like the no-load output. The Aria is a teeny bit lower in the sub-100Hz range than the NC into no-load, ant it is pretty much exactly the same into an 8 ohm load.

After making this discovery, I am more inclined to think this is a hardware design problem, with the output coupling caps being too small.

Armed with the frequency response curves, I experimented with DSPManager EQ settings to try and come up with an EQ that gets it close to right into an 8 ohm load. It is impossible to do it.

However, with DPSManager's "bass boost" turned on, and "Moderate", and EQ settings of +10, 0, -10, -10, -10, -10 it gets to "flat" but not "pink". I am guessing it's distortion showing up at the top end.

In all, this kind of sucks.

 

[TonyBigs]

Thank you for the info. I would look at the coupling capacitors as well. If they're Tantalum I would replace them with Aluminum Oxide and ensure the capacity is at least 220ufd. The problem would be finding the room. In some experiments I did years ago repairing a portable CD player, Tantalum did awful things to high frequency reproduction.

EDIT: here's a picture of the audio circuit on the motherboard (both from ZDnet):

NOOK COLOR

I cannot identify what I would think are the two headphone coupling capacitors. I suspect they are very small value. They *look* like ceramic (color, no polarity markings) and if so, they would be marginally suitable for a line out and sonically and electrically useless for headphones for music listening.

To contrast:

NOOK TABLET:

I am assuming the large two black capacitors side by side are the headphone coupling capacitors. I do believe these to be tantalum electrolytics (color, polarity markings, shape).

S 7 j e

Trying to interpret that code.

My best determination is 47uf @ 16v -- the "j" may be a tolerance code.

If that is the case then 47uf is way too small of a value for any strong bass output at low impedances (16-32ohm). Minimally acceptable at best.

Color and Tablet:

Both circuits seem to have a very small (pf) capacitor after each coupling cap as a high frequency roll off filter. Not unusual for portable equipment if the capacitor value isn't too large causing excessive rolloff.

 

[zires]

You sir are a star! So an amplified set of headphones is the key?

 

[TonyBigs]

Possibly. I still think there would be issues with Nook's coupling capacitor type and value but the input impedance of the headphone amp might overcome the frequency response issue. Sound quality and noise may be another problem that the headphone amp might not address. Tantalums are not good for sound. That is, garbage in, garbage out.

There may be other problems concerning the configuration of the sound chip in the firmware. The TI chip is capable of a lot of sonic "tricks" but based upon your graphs maybe the sound isn't too compromised in the firmware.

 

[mr72]

Hard to tell if those are the headphone output coupling caps without tracing the circuit or at least being able to see the alignment dot on the component package of the audio chip. However, if those are the headphone output caps, then they are followed by a ferrite bead and parallel caps most likely to trim off HF noise as well, evidenced by just looking at the pic.

The Nook Tablet pic definitely has bigger caps, probably tantalum, and who knows the value but that's not a 7343 package, so it can't be more than maybe 100uF.

Looking at the FFT, it looks like about a 1KHz Fc into my 8 ohm load, so I am guessing the caps in a NC are more like 22uF. 220uF still gives like 90Hz Fc into 8 ohms, down to 45Hz into 16 ohms. To get full bandwidth into 16 ohms is going to require about 470uF. Maybe compromise at 330uF and hope you have bigger impedance cans.

Those look like 1206 packages on the NC and 1812 package on the NT to me. Even a tantalum at 470uF is going to be a whole lot better, if this is in fact like 22uF. However, finding an 1812 cap at 220uF or 470uF may be extremely difficult. I am not going to argue about putting a different material cap in here because it's just way too small. A brief check of Mouser shows 7343 is the smallest 470uF caps even at 2.5V in tantalum. They have 6032s in 330uF. But those are way too big.

How hard is it to get the NC apart to the point at which this picture was taken? Do you have to unsolder any components to get there?

I wonder if there is any DC on those outputs, or whether these caps can just be jumpered.

The datasheet says the headphone outputs are rated to drive a 16 ohm load. I guess I was overloading them a little into 8 ohms.

After scanning caps, I think it's pretty much impossible to find a reasonable value cap that will fit into a 1206 or 1210 SMD space. Even the 1812s on the NT are not nearly big enough size-wise. So the only way to solve this with hardware mods is to either put caps somewhere in the wiring between the jack and the board, and jumper the caps on the board, or just jumper the caps on the board and hope for no DC on the outputs. Looks like a good time to use an external headphone amp, or noise-canceling headphones (have a built-in amp).

 

[mr72]

BTW, here's the cheapest, easiest solution to this problem:

http://www.amazon.com/Fiio-E3-BK-E3-Headphone-Amp/dp/B001MPWMDA

And if I were to crack open my NC, rather than doing any surgery or trying to cram different caps in there, I would do this: take the whole PCB from an E3 and shoehorn it into the NC in some open space, wire it between the headphone jack and the headphone output on the NC PCB, and pick up some power from the NC somewhere to run the E3.

 

[hwong96]

Have you installed Dalingren's Nook Tweaks from the market? You can adjust audio settings using this app, specifically increase DAC, adlust headphone gain, speaker gain. Warning! Increasing DAC too high will damage your speaker. I have my DAC setting about 5/8 inch from the left under the a and s in the word "increasing".

 

[mr72]

Have you installed Dalingren's Nook Tweaks from the market? You can adjust audio settings using this app, specifically increase DAC, adlust headphone gain, speaker gain. Warning! Increasing DAC too high will damage your speaker. I have my DAC setting about 5/8 inch from the left under the a and s in the word "increasing".

Yes, this is not a volume problem. It's a frequency response problem, related to the impedance of the device connected. Please see the plots.

 

[22jjones]

I posted earlier about my solution to this. By no means cost effective or portable. FiiO E7 USB DAC and CM7. I stream and download 320 bit from MOG. Great for me working hours at desk. Download is unlimited and can listen offline. The audio quality with this little rig is superb. As in unbelievable. I also use good headphones.

Sent from my NookColor using Tapatalk

 

[mr72]

I posted earlier about my solution to this. By no means cost effective or portable. FiiO E7 USB DAC and CM7.

You using this with USB Host Mode? Is it drawing power from the USB?

Certainly this will sound good but like you say, it's hardly portable. Now, if it were bluetooth rather than USB...

I think a FiiO E3 or E5 headphone amp will mostly fix the main problems, while not improving the DAC or noise performance any. It will definitely fix the reactivity of the output stage. Plus an E5 is cheap + rechargeable and an E3 extremely cheap and extremely compact, barely noticeable and perfectly portable.

I plan to use my NC to read, listen to music and watch movies while on long plane rides, which are all too frequent with my job. Long battery life is key (so drawing power on USB host mode may be bad) and noise performance is not critical. For this kind of thing, I think a small external headphone buffer amp is the ideal solution.

 

[22jjones]

No power drawn from USB, the FiiO has it's own rechargeable battery, one charge will last much longer than the Nook. You have to turn charging off on the FiiO while plugged in to Nook. It is using USB host mode. Add Sennheiser CXC 700 ear buds, and the rig is entering into the audiophile zone.

I think you're right, a small headphone amp, and some EQ will be good. I just did my setup as kind of a quest on how far I could go with Nook audio!

 

[bluespire]

BTW, here's the cheapest, easiest solution to this problem:

http://www.amazon.com/Fiio-E3-BK-E3-Headphone-Amp/dp/B001MPWMDA

And if I were to crack open my NC, rather than doing any surgery or trying to cram different caps in there, I would do this: take the whole PCB from an E3 and shoehorn it into the NC in some open space, wire it between the headphone jack and the headphone output on the NC PCB, and pick up some power from the NC somewhere to run the E3.

That E3 also notes it's for 16+ ohm devices. How does that fix it if you are using your 8 ohm headphones?

 

[mr72]

The two pair of earphones I will be using are 16 ohms and 23 ohms. I tested the NC's audio output into an 8 ohm load simply because that's the practical lower limit for headphones. They are much more commonly 16 ohms or more.

However, to answer your question specifically, it probably doesn't matter, but it's not worth getting into here. The LM4917 amplifier used in the FiiO is rated at 95mW into a 16 ohm minimum load. Running into a smaller load will just limit the maximum voltage swing or increase distortion when the amplifier is running on max.

FWIW, I think I could probably dead-bug a LM4917 into a NC easier than cramming big output coupling caps in there. That'd fix it as well.