600D Audio Controls?

Started by testaevisual, June 17, 2012, 06:16:10 AM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

miyake

Remained our TODO:
-check High pass filter
-Ext balanced
-currently not working with Beep(Disable it?) (double check please
-monitor    OK
-auto int/ext (But just added codes, check my repo)  Confirmed just now.
-digital gain fluctuations (checking now) check OK

I think, analog gain and digital gain are not same width on each gain settings.
It will be confusing to end users , I guess. And Digital gain values distance are little bit big. If we shoot car race, sound is really big. So we need to use digital gain. But the values distance is big now. So I think end user can't set best value for it.

Just remember it, So I didn't use db values.
What do you think?





miyake

Quote from: 1% on August 06, 2012, 08:21:51 PM
Does balance even work the same way? I think we're just switching the polarity of the mics. "Balanced" audio is a joke anyway without a real interface/preamp. Iif you're doing long runs you should be using better gear than 1 long XLR to stereo.

1%
sorry for my poor English, I can't understand these sentence.
my understanding
-lapis chip support differential in.
-If we can convert XLR's HOT and COLD to stereo LR, lapis chip can handle to those HOT/COLD phase like a balanced audio amp.
-We can use long XLR cable and short XLR2stereo cable, and it will provide a low cable noise. such as real balanced audio.

Is this a miss understanding ?

scrax

Quote from: miyake on August 07, 2012, 07:57:57 AM
Remained our TODO:
-check High pass filter
-Ext balanced
-currently not working with Beep(Disable it?) (double check please
-monitor    OK
-auto int/ext (But just added codes, check my repo)  Confirmed just now.
-digital gain fluctuations (checking now) check OK

I think, analog gain and digital gain are not same width on each gain settings.
It will be confusing to end users , I guess. And Digital gain values distance are little bit big. If we shoot car race, sound is really big. So we need to use digital gain. But the values distance is big now. So I think end user can't set best value for it.

Just remember it, So I didn't use db values.
What do you think?

digital gain can have a lot more value, but if you uncomment the part I've commented and try it you will see that when set to 0x70, or 0x71 it is not low, I hear a lot of noise, and also at -35dB there is a strange volume change.
maybe i made some mistake
I'm using ML2.3 for photography with:
EOS 600DML | EOS 400Dplus | EOS 5D MLbeta5- EF 100mm f/2.8 USM Macro  - EF-S 17-85mm f4-5.6 IS USM - EF 70-200mm f/4 L USM - 580EXII - OsX, PS, LR, RawTherapee, LightZone -no video experience-

miyake

Ah, OK i see. I will check it.

MSG added
Checked all (0xff-0x70) for recdgain.
Valid is 0xff to (0xff- 127). It's really loud after 128.
And my old code is mask value is miss taken. Correct mask is 0x7f but I was set 0x70. So it's strange.

Will try to solve the codes.

MSG added 2
Newest code added. Which one do you like?

scrax

Miyake,
Volumes control are fine now, i've just made some little changes to help string in menu (not yet finished),

Remained our TODO:
-monitor   Now internal speaker is enabled when monitor is on and no usb-cable connected, this will make a lot of larsen effect with internal mic selected and medium rec volumes, so sadly it's not useful, if you know how to revert to only headphones monitor I think it's better
-Ext balanced (for now we can keep this last I think)
-check High pass filter
-currently not working with Beep(Disable it?) confirmed fworking now for me too
-auto int/ext Confirmed for mee too.
-digital gain fluctuations Confirmed too.

what is maybe possible to (try TO) ADD
from DSP
-3 band notch filter and audio enhancemnet for microphone at recording?
-5 band notch filter/equalizer (Notch/EQ)?
-Loudness
other
-playback limiter (ALC?)

I'm checking in the pdf if we are making all the right changes, or if we can improve something.
I've noticed less pops than before, but not tested a lot with the last code so far.

For example
VMIDCON Reference Power p42 and p58
QuoteVMIDCON[1:0]
These bits control power ON and OFF of the VMID generation circuit. Power up time can be reduced by using high speed power up mode. VMID generation circuit should be changed to normal mode after power up is completed.
VMIDCON[1:0]        Description
     00                     VMID generation circuit power off
     01                     VMID generation circuit power on high speed power up mode
     10                     VMID generation circuit power on normal mode
     11                     Write prohibit*
* Normal operation is not guarantied at this setting.


Are we using 01?

and also on p43 there is something about pop noise to investigate more maybe, i'm still reading...:
It is easy to generate pop noise when speker amplifier is power up or down, therefore, the LSI provides two method (functions) to reduce pop noise.
First of all, powerup pop noise can be reduced by the following sequence. When power up speaker amplifire, pre-amplifier power on first, then output amplifier on.These are controled by SPCON bits of Speaker Amplifier Power Management Register. For example, when power is on, write SPCON bits 00h?13h?1Fh in order. When the power down, power off pop noise can be redused by output amplifier off first.
Secondaly, pop noise can be reduced by setting speaker amplifier volume to mute and then execute powerup or powerdown. Amplefier Volume Control Enable Register is provided for this purpose and it has volume mute function and fade function. To write AVMUTE bit to "1", speaker amplifier volume becomes mute setting by compulsion. In this case, if AVFADE bit is set to "1" also, the fader function becomes on and then, volume change process(mute transition) is executed with fader function.
I'm using ML2.3 for photography with:
EOS 600DML | EOS 400Dplus | EOS 5D MLbeta5- EF 100mm f/2.8 USM Macro  - EF-S 17-85mm f4-5.6 IS USM - EF 70-200mm f/4 L USM - 580EXII - OsX, PS, LR, RawTherapee, LightZone -no video experience-

miyake

Good to hear monitor speaker issue. My first considering, current code will enable speaker when we disconnect headphones. But I don't know why that is SPK is not enabled. cleanup was correct work for us. So I will add SPK disable code today.

I think extra codes (notch filter and so on) will add after finishing current basic codes. Now I re-confirm lapis chip is really noisy. Alex is also said to us a , "The best thing is we can get cleanest audio. If you want to effect audio, you should do post.".  I', strongly agree.
I guess, If we implement those extra functions, but almost user doesn't use it. Because lapis chip is noisy.
One more thing. As I told you ,currently too much difference with main repo. it's really hard to understand merge.
We added more codes, Alex need a much time to understand it.
So we add basic function first. Then add extra code after finishing merge.

will check your codes.


---
Remained our TODO:
-monitor   Now internal speaker is enabled when monitor is on and no usb-cable connected, this will make a lot of larsen effect with internal mic selected and medium rec volumes, so sadly it's not useful, if you know how to revert to only headphones monitor I think it's better
-Ext balanced (for now we can keep this last I think)
-pop noise reduction
-check High pass filter
-currently not working with Beep(Disable it?) confirmed fworking now for me too
-auto int/ext Confirmed for mee too.
-digital gain fluctuations Confirmed too.
-Delete debug code when we finish editting

1%

Quote-If we can convert XLR's HOT and COLD to stereo LR, lapis chip can handle to those HOT/COLD phase like a balanced audio amp.

This is what I asked. I don't think it really works that way on the other IC (AK). Balanced audio on those stereo cables involved recording 1 channel and then combining it with inverted second channel to make "balanced" audio. If its really putting + on L and - on R then it might work depending on how good ground is. I only have a preamp, I can build a cable but don't see a point when the preamp is 10x better and takes 2 mics.

QuoteNow I re-confirm lapis chip is really noisy

Noisy or just fighting with canon audio? I get pretty decent sound at rest, before I start recording. I'll have to see what happens with latest changes.

QuoteMy first considering, current code will enable speaker when we disconnect headphones.

I think you have to set the speaker as an option. It shouldn't be used with internal mic... only with a further away external mic otherwise feedback is pretty likely. Also, monitor was coming on with audio disabled which isn't good when you're recording without sound and audio IC is using power anyway.

QuoteIf you want to effect audio, you should do post.

While this is 100% true some things might be helpful to be done in camera. The filters are already pretty decent and the EQs may help you cut something you don't want to remove in post later or will be unable to once its mixed with everything else. Noise that is never written is better than trying to filter it on your only copy. The reverse could also be true and the digital effects could be really bad, we haven't tried yet.

Quote
and also on p43 there is something about pop noise to investigate more maybe, i'm still reading...:

This is why I set the monitor the way I did. If you don't follow the order you get sudden changes and pops or other weird effects. All this is in the data sheet. You don't get pops in canon audio because thats how they do it. If you look at the original functions  they use the orders in the datasheet and sometimes mute things when changing power, etc.

miyake

QuoteThis is what I asked. I don't think it really works that way on the other IC (AK). Balanced audio on those stereo cables involved recording 1 channel and then combining it with inverted second channel to make "balanced" audio. If its really putting + on L and - on R then it might work depending on how good ground is. I only have a preamp, I can build a cable but don't see a point when the preamp is 10x better and takes 2 mics.

I see. And I checked block diagram in PDF. baranced and un-baranced is different cable connection with mic and chip-pins.So I think , we can't input + on L and - on R..

QuoteNoisy or just fighting with canon audio? I get pretty decent sound at rest, before I start recording. I'll have to see what happens with latest changes.
I mean hiss noise.

QuoteI think you have to set the speaker as an option. It shouldn't be used with internal mic... only with a further away external mic otherwise feedback is pretty likely. Also, monitor was coming on with audio disabled which isn't good when you're recording without sound and audio IC is using power anyway.
I also think so too. Basically disable speaker while we using HP monitor.
Anyway, current code doesn't disable speaker . But actually disabled speaker. <= I think this is strange thing.
Might be my poor English is reason for this miss communication ....

QuoteWhile this is 100% true some things might be helpful to be done in camera. The filters are already pretty decent and the EQs may help you cut something you don't want to remove in post later or will be unable to once its mixed with everything else. Noise that is never written is better than trying to filter it on your only copy. The reverse could also be true and the digital effects could be really bad, we haven't tried yet.
I see.  But You need to add this function in this time? Can you accept to add this function after finishing basic function?
And it has a lot of settings , did you think menu layout?

miyake

scrax:

I can't reproduce SPK issue......
If you still have problems, check audio_ic_set_lineout_onoff().
> if(audio_monitoring){
had a lot of regsiter values setting.but
>    }else{
has few things.
So you need to try to add under follows in }else{ field.
-head phone amp off on ML_PW_REF_PW_MNG   
or
-Digi->Ana(DAC) off on ML_PW_DAC_PW_MNG
or
-Speaker off on ML_PW_SPAMP_PW_MNG

MSG added
Added mute on/off functions. But the mute on/off make a pop noise(lol.  How can I do that.......................
The reason of pop noise is RECPLAY state to AUTOREC bit on.   But it need a 0.5-1.5secs for switching on. So If we completely disable pop noise, we need to add msleep(1500) before mute_off() . But it's too long. So I added 500 now. for testing.

1%

QuoteCan you accept to add this function after finishing basic function?

Yea, without the basics done we're kinda stuck.

QuoteI mean hiss noise.

You get hiss no matter what, I've gotten hiss on $1000 mics. The way you set in volume + boost may have something to do with it. You're always boosting the highest input volume setting when they really go together. If you look at the block diagram it goes input->PGA->boost->ADC. There should be the option to set them interdependently. You might want X input volume with X boost and it might sound better than really high input volume and no boost. It think boost is power or analog gain, need to research what it actually does. Input volume may just be PGA sensitivity? Its a tough call, they don't go into much detail here.

miyake

1%:
I already have a JuicedLink preamp and RODE video mics. So basically understood those how-to.
My miss understanding is
a lot of hiss noise = we can't hope a lot of things by lapis chip.

We can use another function(Eq, limitter, and so on) with no-boost and small volumes. And input big volume source.
right?

Anyway, you also have HP monitor and SPK power on issue?

1%

I'll check it. But another thing to test: is the monitor audio we hear the same as what is being recorded.


scrax

some things are getting solved. but i have problems in recoding video, strange hangs with battery pull needed.
I'm using ML2.3 for photography with:
EOS 600DML | EOS 400Dplus | EOS 5D MLbeta5- EF 100mm f/2.8 USM Macro  - EF-S 17-85mm f4-5.6 IS USM - EF 70-200mm f/4 L USM - 580EXII - OsX, PS, LR, RawTherapee, LightZone -no video experience-

1%

DC on/off is switched, I think.

miyake

Quote from: scrax on August 09, 2012, 12:15:40 AM
some things are getting solved. but i have problems in recoding video, strange hangs with battery pull needed.

I guess, same as
https://bitbucket.org/miyake_t/magic-lantern600daudio/changeset/dda0a1ddd6587c71a097141bc5f7f5f33d5f247e

Will try to solve it.

miyake


miyake

To 600D-audio early testers: souce code is already opened. If you find a problems, please let me know the detail. on my issue tracker.
https://bitbucket.org/miyake_t/magic-lantern600daudio/issues?status=new&status=open

and source is here:
https://bitbucket.org/miyake_t/magic-lantern600daudio

Ash McKenzie

Anychance of a precomplied autoexec.bin I'm super keen to test and i have balanced mics but I don't know how to complie

miyake

Maybe, we can publish 600D audio autoexec.bin after 2.3 is out by free(13rd). (We need to confirm core team)

One more thing , 600D's lapis-chip has no differential in capability.(chip is supported but internal mic connection is different.)
Disabled balanced now.

johndb

I am very keen on testing the 600D audio as well. I also cannot compile and would prefer the Autoexec.bin.

Thanks for all your hard work.
600D ML | 450D | EF-S 17-55mm f/2.8 IS USM, EF-S 60mm f/2.8 Macro USM, EF 70-300mm f/4-5.6 IS USM, Speedlite 580EX II, Azden SMX-10 Microphone; Sony Vegas HD Platinum, DxO Optics Pro, DxO Filmpack, Oloneo HDR Photoengine

1%

Wait the 3 days, trust. Its still not perfect with canon messing up our volumes and such.

juantrueno

This is great! Thanks to all developers, contributors and any one who worked on this project.  Fixing audio on 600D is a great step.
Personally is huge for me and my work!

Thanks again
Canon 600D. Canon 18-135mm f 4/5.6 / 50mm f 1.8 / Yashica (Análogos) 28mm f 2.8 / 50mm f 1.4 / 135mm f 2.8
Vimeo.com/juantrueno youtube.com/juakinyan @JuanTrueno

miyake

Now I'm confirming the release thing of 600D audio beta autoexec.bin to core team through Alex.
If I got replay, I will act up to core team decision.

ilguercio

Nice to know we have some skilled developers apart from A1ex :D
Canon EOS 6D, 60D, 50D.
Sigma 70-200 EX OS HSM, Sigma 70-200 Apo EX HSM, Samyang 14 2.8, Samyang 35 1.4, Samyang 85 1.4.
Proud supporter of Magic Lantern.