Rockbox: Clip+

[dfkt]

Well, I did not change the two conditions if (vol_* <= 0x16) - you say I should change that level too?

No, I only tested the two higher settings in RMAA, not by ear. I will try it now.

[CrowdAds]

[Timar]

For those of you who want to get more juice out of the Clip+, here's a build I made with mix volume set to 0x1B. Greater values produce some amount of clipping with a 0dbFS sine wave at absolute volume levels => mix volume. This one is free from clipping however. It's based on r26054, so the SD-fix is included. Please remember: don't ask the developers for support on this build.

http://www.mediafire.com/?emytdtcjjju

EDIT: It's actually 0x1B, not 0x1A

[dfkt]

Here's some RMAA tests with Timar's 0x1b and 0x1e builds, and the current stock 0x16 build.

Clip+ - Rockbox Volume 0x16 - Official build 16 Ohm
Clip+ - Rockbox Volume 0x1B - Timar build 16 Ohm
Clip+ - Rockbox Volume 0x1E - Timar build 16 Ohm

+++++

I've measured a 0dB white noise file played into my -10dBV weighted sound card, here's how loud everything is (using a 16 Ohm load):

OF max		-5.3dB

RB 0x16 0db	-15.9dB
RB 0x16 6dB	-10.1dB

RB 0x1b 0db	-10.1dB
RB 0x1b 6dB	-4.9dB

RB 0x1d 0db	-10.1dB
RB 0x1d 6dB	-4.9dB

RB 0x1e 0db	-9.5dB
RB 0x1e 6dB	-4.5dB

0x1b and 0x1d are equally loud for some reason, and 0x1e is only marginally louder. Seems there's something else limiting the volume level?

[Adub]

Originally Posted by ultrauber

I know it's old new but hey it's r/w now. Too bad it's so unstable...keep up the work.

Hehe unstable, i did 5 strait battery bench and the clip never crashed, i wouldn't can that unstable at all.

[Timar]

Originally Posted by dfkt

Here's some RMAA tests with Timar's 0x1b and 0x1e builds, and the current stock 0x16 build.

Thanks for running these tests. I think 0x1b yields very acceptable results (very acceptable... it's strange that the THD values are much lower than with the equally-loud 0x1d you've tested before). I wonder, however, why the 0x1e build did so well in the RMAA. It clipped horribly even at volumes below 0db. Maybe the test signal isn't normalized to full scale?

It'S interesting that there's no difference between 0x1b and 0x1d. There seems to be some limitation reached. Did you measure the volume with 0x1e? From my impression, it was much louder.

[dfkt]

I've added the 0x1e results above. They're just marginally louder. Very weird. And they're still quieter at 0dB than the stock firmware.

[Timar]

I think it can be explained this way: according to the AS3543 datasheet, the volume range of the mixer is 32 steps @ 1.5db, where the maximum is +6db. To mix a signal to the highest possible volume before clipping (0db), you need to set it to 0x1f - 4 = 0x1b, which is indeed the highest volume that worked without clipping in my tests.

So we have the mixer at the optimal volume with that value. The headphone amp, hoewever, is fixed to vol - 0x16 for these builds. Thus values above the saturation point of the mixer don't give much more volume...

[spartacus]

Originally Posted by Adub

Hehe unstable, i did 5 strait battery bench and the clip never crashed, i wouldn't can that unstable at all.

Think I'm going to have to take the plunge into Rockbox waters - which build are you using?

[saratoga]

Careful, this is how you get hooked and end up a developer

[Adub]

Bootloader from dfkt here from a page back and i am currently using r26025 becuase my clip+ is finally 99% (there's always a chance hehe) stable so I'm not touching another version yet.

[setmenu]

That volume patched build of Timars seems to be working fine thus far.[not listened for any length of time yet]
Does the job nicely.[thanks Timar!]

saratoga , could such a patch find its way onto future builds?

[dfkt]

It's not a patch, it's just a different volume setting. The current one is obviously not optimal, that's why we were trying around a bit.

[setmenu]

Apologies for using the wrong term.
I am very appreciative of everyone's work, Rockbox has transformed my clip/phones combo.

[dfkt]

Here's a new build with 0x1b volume mix level - http://www.mediafire.com/?zd5o1ltazuj

[Kenaneu]

Originally Posted by dfkt

Here's a new build with 0x1b volume mix level - http://www.mediafire.com/?zd5o1ltazuj

Tried it. Volume seems totally adequate to me compared to earlier builds. Also very stable for me. No probs using the database and no crashes during several hours of use. Fantastic progress!

[Adub]

Wow dfkt that improved my enjoyment using cans on the clip+ by a very large amount, works great thanks.

[HiDeF]

Yeah, I think the latest build is really stable for now, another wait is for usb charging >67% & files transfer. :-) Thank again dfkt.

[dfkt]

Just checking - none of you are hearing any distortions with that volume level?

[HiDeF]

Mine, the sound just perfect than any previous build before. And moderate volume level are -6 to -2 is really enough for normal listening.

[jsmaye]

Originally Posted by sanka

1.2% doesn't sound like a big number, but in terms of pitch tuning, it's a pretty big difference.

I like to use the Clip v2 when practicing guitar. I can play along, on an in-tune guitar, with a track that plays in-tune in Play or in iTunes. It sounds fine: the guitar and the track playback are in tune. If I try that with my Clip running Rockbox, the Clip plays noticeably out-of-tune with the guitar. It plays about 1/5 of a semitone flat. 1/5 of a semitone flat is difference of about -1.12%. This seems like a small number, but if you heard two instruments that were tuned -1.12% different from one another, you would definitely be able to hear that they were out-of-tune with one another...

That's why I said "most people" - since my Clip is the last device I'd play along with, I may or may not notice +/- .2 semi-tone without another tone or instrument as reference.