Author Topic: Analysis of VRC6 volume  (Read 13809 times)

0 Members and 1 Guest are viewing this topic.

Offline aldocx

  • Newbie
  • *
  • Posts: 12
  • Karma: +0/-0
    • View Profile
Re: Analysis of VRC6 volume
« Reply #60 on: May 17, 2018, 02:22 PM »
As requested, here's Intro+Epitaph+Prayer+Beginning from Akumajou Densetsu recorded from a Famicom Family Famicom:
https://mega.nz/#!4NZliYLY!rj655cRqSyUJ6bkfL1O2a2kzcekYw7S3ogtZZxEL39A
« Last Edit: May 17, 2018, 02:27 PM by aldocx »

Online foldor

  • Newbie
  • *
  • Posts: 11
  • Karma: +0/-0
    • View Profile
Re: Analysis of VRC6 volume
« Reply #61 on: May 17, 2018, 03:56 PM »
So, for those of us with NA front loader NES consoles, is 47k ohms still the recommended resistor value (Knowing there's a ~10% tolerance)? If not, what would be?

Offline James-F

  • Full Member
  • ***
  • Posts: 179
  • Karma: +17/-0
    • View Profile
Re: Analysis of VRC6 volume
« Reply #62 on: May 17, 2018, 07:29 PM »
As requested, here's Intro+Epitaph+Prayer+Beginning from Akumajou Densetsu recorded from a Famicom Family Famicom:

Got to say that I consider this the most balanced Akumajo Densetsu recording I've heard.
This is a "FF"/"VCI" HVC-GPM Famicom with original AD cartridge.

Analyzing the balance of this recording reveals that it closely matches the VRC6 mapper with 35 volume for N8 with AV Famicom.
The "even more original" non-FF Famicom HVC-CPU had its internal volume slightly lower than FF Famicom from what I hear.
OR, it may be that Famicoms and Akumajo Densetsu carts vary in volume. :)

@folder
Balancing a modified US NES is even more difficult for obvious reasons.
« Last Edit: May 17, 2018, 07:59 PM by James-F »
Mega Everdrive x5, Everdrive 64 v3, Everdrive N8, SD2SNES, Joyzz.

Offline Slamy

  • Newbie
  • *
  • Posts: 36
  • Karma: +4/-0
    • View Profile
Re: Analysis of VRC6 volume
« Reply #63 on: May 17, 2018, 09:48 PM »
A 100_100 version was requested.
A 100_100 version shall be delivered.
I hope this suits the need for normal Famicom users.

Offline James-F

  • Full Member
  • ***
  • Posts: 179
  • Karma: +17/-0
    • View Profile
Re: Analysis of VRC6 volume
« Reply #64 on: May 17, 2018, 09:50 PM »
Thanks!

Now there is FOUR identical 100 levels, since the LO setting is no longer part of N8 OS.  :)
Gee, I wonder who asked for it... maybe someone still using v13?

I shall repost this image for people who did not see it yet:


@Slamy
Did Krikzz contact you about improving FDS audio?
« Last Edit: May 18, 2018, 04:13 AM by James-F »
Mega Everdrive x5, Everdrive 64 v3, Everdrive N8, SD2SNES, Joyzz.

Offline Arnold101

  • Sr. Member
  • ****
  • Posts: 309
  • Karma: +4/-1
    • View Profile
Re: Analysis of VRC6 volume
« Reply #65 on: May 18, 2018, 01:00 AM »
A 100_100 version was requested.
A 100_100 version shall be delivered.
I hope this suits the need for normal Famicom users.
;) thanks

since the LO setting is no longer part of N8 OS.  :)
 
wrong, the v20 have LO by default...not HI

Online brianvgplayer

  • Full Member
  • ***
  • Posts: 143
  • Karma: +3/-0
    • View Profile
Re: Analysis of VRC6 volume
« Reply #66 on: May 18, 2018, 01:08 AM »
Thanks for the 100_100 version!

Offline James-F

  • Full Member
  • ***
  • Posts: 179
  • Karma: +17/-0
    • View Profile
Re: Analysis of VRC6 volume
« Reply #67 on: May 18, 2018, 04:10 AM »
To avoid confusing future readers, here's the official change log:
https://krikzz.com/pub/support/everdrive-n8/OS/changelist.txt

Quote
nesos-v14 28.10.2016
1)Fixed MMC4 (glitches in Fire Emblem)
2)New mappers support. 31, 137, 138, 139, 141, 147, 168
3)Removed "audio mix vol" option. Now it always in "Hi" state
4)Removed "Mapper info" menu due the lack of memory space
5)Minor bugfix

edit:
I have tested all four 100_XX RBF files, there is zero difference between them in v20.
« Last Edit: May 18, 2018, 01:54 PM by James-F »
Mega Everdrive x5, Everdrive 64 v3, Everdrive N8, SD2SNES, Joyzz.

Offline Adsixnine

  • Newbie
  • *
  • Posts: 2
  • Karma: +0/-0
    • View Profile
Re: Analysis of VRC6 volume
« Reply #68 on: May 18, 2018, 01:32 PM »
Thank you very much Slamy!

Offline Arnold101

  • Sr. Member
  • ****
  • Posts: 309
  • Karma: +4/-1
    • View Profile
Re: Analysis of VRC6 volume
« Reply #69 on: May 18, 2018, 08:09 PM »
To avoid confusing future readers, here's the official change log:
https://krikzz.com/pub/support/everdrive-n8/OS/changelist.txt

Quote
nesos-v14 28.10.2016
1)Fixed MMC4 (glitches in Fire Emblem)
2)New mappers support. 31, 137, 138, 139, 141, 147, 168
3)Removed "audio mix vol" option. Now it always in "Hi" state
4)Removed "Mapper info" menu due the lack of memory space
5)Minor bugfix

edit:
I have tested all four 100_XX RBF files, there is zero difference between them in v20.
is impossible that is on hi...all the exp games on  the v20 have very low sound now, wich i haven't when i was using HI on previous updates
« Last Edit: May 18, 2018, 08:11 PM by Arnold101 »

Offline Sarge

  • Full Member
  • ***
  • Posts: 106
  • Karma: +7/-0
    • View Profile
Re: Analysis of VRC6 volume
« Reply #70 on: May 19, 2018, 07:42 AM »
I'm seeing the same behavior.  I'm running a modded NES front-loader (47K resistor).  The VRC6 audio in V20 beta 1 sounded fine, but in V20 beta 2, it's very low.

EDIT:  James-F, replacing with Slamy's 100_100 and your maprout.bin file, the audio is back to normal.  Not sure what is going on with the official version right now.
« Last Edit: May 19, 2018, 07:53 AM by Sarge »
Hardcore Retro Gaming - Discuss games ten years old or older!

http://hardcoreretrogaming.boards.net/

Offline James-F

  • Full Member
  • ***
  • Posts: 179
  • Karma: +17/-0
    • View Profile
Re: Analysis of VRC6 volume
« Reply #71 on: May 19, 2018, 08:17 AM »
OS v20 beta2 has volume 35 for VRC6, to match with VRC7 and Sunsoft 5B.

The previous OS v16 has krikzz's VRC6 mapper which is nowhere near as accurate as Slamy's, and yes it is louder and very imbalanced internally.
Since Krikzz updated VRC7 and Sunsoft 5B,,, Slamy's VRC6 with volume 35 was chosen to match their volumes.
The worst thing is different mapper volumes on the same console.

As to when and/or if krikzz is going to provide a solution for different consoles, I have no idea, so please stop asking me as if I have anything to do with it.
« Last Edit: May 19, 2018, 09:05 AM by James-F »
Mega Everdrive x5, Everdrive 64 v3, Everdrive N8, SD2SNES, Joyzz.

Offline Slamy

  • Newbie
  • *
  • Posts: 36
  • Karma: +4/-0
    • View Profile
Re: Analysis of VRC6 volume
« Reply #72 on: May 19, 2018, 02:34 PM »
@Slamy
Did Krikzz contact you about improving FDS audio?

I wasn't contacted. I assume he is still trying to do it by himself or for a later version as the current beta already has many changes.
Krikzz seems to be more like an engineer. It's important for him that logical stuff like the basic functionality of the Everdrive being a working flash cart is provided.
This whole sound stuff is sometimes purely subjective and people can argue what is right or what is wrong.
From the perspective of an engineer (I'm also one and I know how that feels) this is extremly frustrating as there is no "right solution".

As I've given my mapper sources to him I reevaluated the source code so that it's clean for him to read.
I've even found a bug inside that could cause sound glitches in the range of a few nanoseconds dependend on signal latencies on the FPGA.
But as no one is even noticing this a fix is maybe not needed.

This is what I'm talking about. When engineering meets audio there is nothing that can be exact as also our ear is far from being exact. If something is wrong you can only guess it if you are a musician.

Why am I even writing this. I don't know...
If he asks me for assistance I will most likely help but maybe I'm not the best one for this job.
The VRC6 fix was developed because of personal reasons as I've bought the N8 especially for this job to play Castlevania III. I know how it must sound like and I wanted it.
With the FDS I'm currently not that motivated so I won't be the active one this time to ask him.

Offline Jegriva

  • Jr. Member
  • **
  • Posts: 73
  • Karma: +1/-0
    • View Profile
Re: Analysis of VRC6 volume
« Reply #73 on: May 20, 2018, 02:31 PM »
So, for those of us with NA front loader NES consoles, is 47k ohms still the recommended resistor value (Knowing there's a ~10% tolerance)? If not, what would be?

If you don’t aim to the perfect musicophile volume level... yes, that should generally do.
But bear in min that it is no perfect solution.

From Nerdly Pleasures:

Quote
It was the received wisdom for a long time that if you inserted expansion audio into the NES audio path using a particular mixer value, 47K for the Front Loader and 1.2K for the Top Loader, you would be all set for Famicom Expansion Audio.  But this one-size fits all approach will not give you accurate audio levels with every real cartridge that uses expansion audio.  Each type of cartridge used different mixing methods, and the Namco 163 carts used varying mixing resistors.  You should really use a potentiometer that can encompass those ranges and more and adjust based off samples taken from a real Famicom.
« Last Edit: May 20, 2018, 02:33 PM by Jegriva »

Offline goldenband

  • Newbie
  • *
  • Posts: 16
  • Karma: +1/-0
    • View Profile
Re: Analysis of VRC6 volume
« Reply #74 on: May 21, 2018, 08:26 PM »
As I've given my mapper sources to him I reevaluated the source code so that it's clean for him to read.
I've even found a bug inside that could cause sound glitches in the range of a few nanoseconds dependend on signal latencies on the FPGA.
But as no one is even noticing this a fix is maybe not needed.

This is what I'm talking about. When engineering meets audio there is nothing that can be exact as also our ear is far from being exact. If something is wrong you can only guess it if you are a musician.

I haven't modded my NES yet to enable expansion audio, but is this representative of the issues heard playing FDS games on the N8?

https://www.youtube.com/watch?v=n0EY_k40IEk

I'd say those are pretty striking issues. The entry of the wavetable channel is extremely flat in Kid Icarus (something like 3% -- it could be measured more precisely if the other channels were silenced). But even when it seems to come back into line for the main tune, it's still very slightly flat, maybe 0.5%.

Same thing for the game (whatever it is) that comes in at 1:50, where the FDS channel is both delayed/laggy and flat by a small but noticeable amount. Problem is, it's also flat on real hardware too! :D But I think the N8 may lose an extra -0.1% or -0.2% above that; it'd be easier to tell if the FDS channel were isolated.

And then what I assume is a Jajamaru-Kun game that starts at 3:05...check out what happens at 3:23...wow! :o Sounds like a sample-and-hold effect on an old analog synth...what's happening there?

All of these are definitely more conclusive than the "Which combination of hardware provides the right reference volume level for FDS sound?" debate, which is an interesting question but far less objective, in terms of what's 100% right or wrong, than the "blEARGHarGHARgh" of that bit at 3:23.
Have: EverDrive 64, EverDrive N8 (Famicom), Turbo EverDrive, EverDrive MD, Super EverDrive, EverDrive-GB, EverDrive-GG, Master EverDrive, EverDrive-GBA X5

Other flashcarts & multicarts: Harmony (Atari 2600), Cuttle Cart 3 (Intellivision), 233-in-1 (Odyssey^2), 72-in-1 (Vectrex)