So, as I said in this blog post, the Nintendo Entertainment System lacked the 2 pins necessary to use the sound expansion chips included in many of the Japanese Famicom games. From time to time, I’m going to post tracks that lost a lot of fidelity and sound because of this issue.
Akumajou Densetsu (or as we know it in the US, Castlevania III: Dracula’s Curse) is one of the most famous examples of loss of fidelity due to the lack of sound expansion chip. Konami was forced to use the the MMC5 Nintendo sound expansion chip, which operated differently than the VRC6 sound chip that the original game used.
Here’s “Riddle” from the Akumajou Densetsu Original Soundtrack, using the Konami VRC6 sound expansion chip:
(credit : explod2A03)
Now, here’s “Riddle” from the Castlevania III: Dracula’s Curse Original Soundtrack, which had to be remixed to use the Nintendo MMC5:
(credit : explod2A03)
The main difference is the lack of sawtooth wave. Sawtooth is EXCELLENT as a synth bass. Most of that “crunchy” almost “thudding” noise in the VRC6′s bassline is the sawtooth. It makes a huge difference. Unfortunately, as I discussed before, the MMC5 track bassline is based off the volume of the triangle wave and the triangle wave on the stock 2A03 (and MMc5) had no way to control the volume. The added sawtooth gave composers a way to control the volume of the basslines and add more depth. Not having the sawtooth was a huge limitation that many composers had to overcome. What’s even more obnoxious is the MMC5 was a VERY expensive chip compared to the Konami VRC6, which was cheaper for Konami to use since they manufactured it.
Note that in the NES version, because of sound/space limitations, the A section only plays once before going to the B section instead of the two times it plays in the VRC6 version.
You’ll also see there’s no true square waves – as evidenced in the running arpeggios in the background. Instead of sounding bold and strong, they sound tacky. All the MMC5 does is add another two pulse waves. The VRC6 adds two square waves and a sawtooth which, when paired with the stock A203′s two pulse waves, triangle wave, noise channel, and DPCM, gave the composer a total of 8 different instrument choices and 6 different instrument sounds. Clearly, the MMC5 is an inferior chip. It’s sad that this is the version of the music we heard. I’m curious as to what the composers thought. At any rate, I hope this was interesting. Any comments are welcome.
February 28th, 2011 at 11:05 am
I have plenty of questions to ask, but one more immediate than the others. Why, both in layman’s and technical terms, do the arpeggios sound *so* bad in the MMC5 version of this song? In my five seconds of research, it appears that square waves and pulse waves are similar in nature, besides pulse waves not necessarily having symmetry. From your field of interest, does this slight dissimilarity cause the entire lack of fullness or ’roundness’ of the instrument voice, or are there other limiting factors of the hardware?
February 28th, 2011 at 4:46 pm
I’m assuming that the original composer(s) intended the arpeggios to be generated by the VRC6 chip and tweaked the arpeggios to sound the way they wanted with a VRC6 set up. By comparison, I would say the MMC5 does not maintain the original character and idea behind the VRC6′s versions of the arpeggios. Whether the composer(s) would be upset or satistified by the results on the MMC5 could only be answered by the composers.
And you’re estute in saying it has something to do with the shape of the physical sound itself. I’m not an electrical engineer but I’ll do my best to explain it. I’ve been messing around the with Nintendo NDS sound chip and you can physically generate pulse and square waves and I think I have a fairly decent understanding of them. It has a 128-point waveform that you can manipulate. If you manipulate it in accordance with the rules of making a square wave, it will sound like a square wave. If you manipulated to the specs of a sawtooth wave, it will sound like a sawtooth wave. Really fun to work with. Here’s what I was using to tweak the Nintendo NDS.
Square waves, it seems, are perfectly symmetrical as demonstrated by this image:
(credit : http://www.thedawstudio.com/Tips/Soundwaves.html)
Where as a pulse wave or “rectanglar” wave looks like this:
As you can see, the “duty cycle” is different – the square wave has a more even “form” (probably not the technical word but I’ll take it) compared to the pulse wave. I’m assuming this causes the major disparity in sounds. I’m not sure exactly how or why or what technical terms I should use to describe the difference. Maybe “roundness” isn’t the correct word. This is a great idea for another post. I’ll take a look this evening.
March 1st, 2011 at 5:40 pm
I’ve been working on designing a simple audio comparison between PWM and Square-waves. You’d be surprised how difficult it is to find a plain PWM VST plugin.
March 1st, 2011 at 6:18 pm
Have you considered using FamiTracker ? I know you can save the music you create as .wav. Then, you could load those into whatever program you’re working with. I find it useful.
March 2nd, 2011 at 10:34 am
[...] focusing on the VRC6 vs. the MMC5, I decided maybe we should highlight the Nintendo FDS vs. the Nintendo MMC1. The beloved game [...]
March 2nd, 2011 at 10:53 pm
[...] I was JUST talking about the complications with the VRC6 to MMC5 transition of Akumajou Dentetsu/Castlevania III: Dracula’s Curse. And it turns out, Jeremy Parish, [...]
March 23rd, 2012 at 6:46 pm
[...] I was JUST talking about the complications with the VRC6 to MMC5 transition of Akumajou Densetsu/Castlevania III: Dracula’s Curse. And it turns out, Jeremy Parish, [...]
January 23rd, 2013 at 2:39 pm
[...] I’ve pointed out in a previous post, the US-released Castlevania III is missing the VRC6 audio mapper. The VRC6 adds an extra 3 [...]
January 24th, 2013 at 9:34 am
[...] I’ve pointed out in a previous post on my blog, the US-released Castlevania III is missing the VRC6 audio mapper. The VRC6 adds an [...]