Analysis of VRC6 volume

[Slamy]

Hey there.

I recently bought the Everdrive N8 from the EverDrive Shop in Germany. After modding an US NES for the audio i was confused about the audio quality and decided to further analyse it.

I've made a YouTube video which plays only the VRC6 channel of akumajou densetsu recorded from the real NES (expansion audio to line in of soundcard) and from AudioOverload (NSF player on PC). It should show the issues.
The audio is not that bad but the sawtooth wave is much too quite.

Please have a look:
https://www.youtube.com/watch?v=nHKLq6YUFmg

What confuses me is a very old recording uploaded here which shows that it was correct in the past.
Can somebody explain?

[goombakid]

Did you switch the audio mix volume option to Hi?

[Slamy]

Did you switch the audio mix volume option to Hi?

Yes i did. But switching to Hi makes not only the Sawtooth louder but also the pulse waves. I can't hear the APU any more by doing so.

[Slamy]

Am i really the only one having problems here? Am i just to picky?

To be sure that it's not just a faulty Everdrive N8 (who knows) i've searched for more evidence.

Here is a guy playing an original Akumajou Densetsu on a real famicom with the framemeister:
https://youtu.be/1ViykEF7sgg?t=141
Notice the baseline.

Here is some other footage of a guy playing this game with the everdrive:
https://youtu.be/8LE9krbHe2s?t=138
Now notice the quite baseline.

[pld]

Well yeah it is known that the expanded audio is not played 100% correctly.

[Slamy]

Well yeah it is known that the expanded audio is not played 100% correctly.

It is ?
Then why is this in the official feature list of the Everdrive N8? In the german and american shop it says that expansion audio is supported.
This is the equivalent of saying that the execution of windows software under linux is supported. Well it is but kinda sorta and not 100% accurate. You get the idea.

If the shop would tell that it has issues than that would be fine. There should be a list of games that are known to not work well.
I'm just angry that I've wasted hours of soldering and changing the resitors value and the active low pass in the hope that there is an issue with the mixing circuit in the NES to only find out that the generation itself is faulty to begin with.

[XiTaU]

Its listed because expansion audio is supported.
If it was not supported u couldnt hear the WIP mappers outputing expansion audio

is it 100% accurate no but it is supported.

[Slamy]

I've found another issue. I was playing Metroid when i noticed a low pitch buzzing sound when paused and the game is supposed to be silent. It seems that the FPGA is outputting noise if the expansion audio is NOT simulated. This static noise is non existent when pausing in akumajou densetsu. Maybe the pin is floating with all other mappers not supposed having that output. A pull down might solve this but i'm not sure if this could have a bad effect on the expansion audio.

[KRIKzz]

I've found another issue. I was playing Metroid when i noticed a low pitch buzzing sound when paused and the game is supposed to be silent. It seems that the FPGA is outputting noise if the expansion audio is NOT simulated. This static noise is non existent when pausing in akumajou densetsu. Maybe the pin is floating with all other mappers not supposed having that output. A pull down might solve this but i'm not sure if this could have a bad effect on the expansion audio.

If N8 turn on expansion audio, it reduce overall volume of sound, so, i guess, you just not heard this noise

[Slamy]

First of all. Thank you for supplying me with the source code.
I've managed to improove the VRC6 implementation with informations from nesdev.

Please have a look at my blog where i write something about this issue:
http://slamyslab.blogspot.de/

If you want to test the mod you can download it here:
https://sites.google.com/site/slamyselectronicslab/024.RBF?attredirects=0&d=1

Just replace 024.RBF with the one supplied. Installation at your own risk.

I archived quite good results with 77kOhm between pin 3 and pin 9.
This is my first shot at this so the volume setting is ignored and should be roughly at HI.
But as the volumes are now different you need to change the resistor value anyway.

Please tell me how you like it and what you think about the accuracy. The next days i will make further improovements if i find some time.

NOW WHIP THOSE VAMPIRES!!!

[darcagn]

I've found another issue. I was playing Metroid when i noticed a low pitch buzzing sound when paused and the game is supposed to be silent. It seems that the FPGA is outputting noise if the expansion audio is NOT simulated. This static noise is non existent when pausing in akumajou densetsu. Maybe the pin is floating with all other mappers not supposed having that output. A pull down might solve this but i'm not sure if this could have a bad effect on the expansion audio.

Indeed this is a problem, the expansion audio pin of the N8 is very noisy. On my cartridge I cut the trace of the pin for the expansion audio and installed an on/off switch for the expansion audio on the cartridge. I keep it turned off unless I'm playing a game with expansion audio.

[goombakid]

Think this will improve it with the N8 Fami on a Famicom?

[Slamy]

Think this will improve it with the N8 Fami on a Famicom?

If i click on the OS Update on the Everdrive N8 Famicom it gets me to the same Dropbox folder as with the NES version. So i assume the design is identical. But I fear that there might be volume differences between VRC6 and APU. I don't own a Famicom so i can't test it.
What the options need is an analog way to change the volume. LO and HI is too digital. Looking at the code of the FPGA side it should be capable of that. One thing what I can try tomorrow is to upload multiple versions of my mod. Each with a different volume setting and you can try some of them.

[Ryoandr]

Note that there are differences even within red/white famicom revisions. Add sharp-made consoles with their inverted resistors, and AV famicom and it's crappy IC that doesn't deliver the right level, and you have a complete mess... If you can you should check your console and restore the sound circuit. Sometimes it's as simple as changing a resitor.

I have an AV Famicom with a rebuilt sound circuit to make it identical to original famicom (inverter gate and 100k feedback resistor, bypassing that custom IC). Will test your mapper soon.

Also, while I don't have Akumajo Densetsu, I do have a Madara which also uses VRC6, so I can do comparisons.

[Slamy]

It's kinda sad they didn't got it consistent. That kinda sounds awful.. literally. You've bought the game but got the wrong console to play it on.

The mapper is now configurable and has support for HI and LO.
If you download the zip file here....
https://sites.google.com/site/slamyselectronicslab/24VRC6Mod.zip?attredirects=0&d=1

.... you get multiple files that are named with the following scheme:
24Mod_<HI>_<LO>.RBF.
The numbers represent the total output volume in percent. 100% is the maximum the FPGA can drive and offers the best signal to noise ratio.
This way you can alternate the volume by swapping the mapper or maybe check two different ones without swapping.

I did make myself some notes while testing.

Code: [Select]

Output Volume of Everdrive N8 inside an US NES by Ear - Akumajou densetsu

100% Prefer this for best signal quality. Use poti or normal resistors for correct volume.
22%  Seems usable. More emphasis on the baseline than the whip (done on APU).
20%  Seems usable. Only volume I tested with headphones to get a more clear sound.
18%  Seems usable. Maybe balanced.
14%  VRC6 is still to quite. Melody in "Mad Forest" not intensive enough.
12%  Harmonics are clearly louder than melody. VRC6 is too quite. On the other hand the whip has a decent volume.

One stupid thing that i don't get yet is the volume of the whip. It's feels to quite. It's actually played by the noise channel of the APU. But if the whip is loud enough the harmonics which are mostly performed on the internal pulse waves overshadow the melody.

Happy testing.