Talk:Yamaha YM3812

Audio clip
Can anyone provide an audio clip? I'd like to hear what the “characteristic sound” of the chip is like. ::Travis Evans 05:40, 4 December 2007 (UTC) http://www.youtube.com/watch?v=Rx4BRd5mUEc is a recording of one of the demo applications that came with the Adlib sound card.


 * Unfortunately haven't used one of these in ages, but I know I have recorded SB16 MIDI output to WAV (possibly even CD) in the past. Problems with this however are twofold - 1/ finding it, 2/ copyright issues... There is however an old webpage floating around out there with old computer soundchip samples, if you search you may find it. Had not only OPL but a number of other old chips like the YM2149 and Tandy 3-voice, etc. Very old though. Samples in archaic .AU format, but most audio players e.g. Winamp should be fine with it. 82.46.180.56 (talk) 13:29, 21 January 2008 (UTC)


 * There you go, I added an audio sample. I think it is pretty characteristic. Olli Niemitalo (talk) 16:30, 12 November 2008 (UTC)


 * That's pretty nice ... I know it's been nearly 9 years, but, any chance of a source on that? Your own composition, or some copyright-free demo piece? Or, um... something that might need more comprehensive tagging to not eventually get nuked by some overzealous copyright-bot? 51.7.16.187 (talk) 13:28, 12 October 2017 (UTC)


 * The player has an "enlarge window" kind of symbol that takes you to the file info page. Because I made the song myself, I didn't want to be pushy and advertise on the page itself. Olli Niemitalo (talk) 07:49, 2 December 2017 (UTC)


 * Wow! I totally forgot about this request I made three years ago. Then I somehow stumbled back onto this article, not realizing I had been here before, and once again wondered what it sounded like and came to this discussion page to look for an audio sample (overlooking the one that was added to the main article because I'm great at missing things :-P) only to rediscover my original request and some responses. :-D Anyway, thanks you two for the samples. It does sound characteristic; it kind of reminds me of how the Yamaha synthesis chip in the Sega Genesis sounded, sort of. ::Travis Evans (talk) 01:35, 7 February 2011 (UTC)


 * Well, it should do, as they're made by the same company and use largely the same method of sound generation. However AFAIK, the Genesis chip (YM2612/OPN2) had 4-operator voices, but ones based wholly on full sinewaves - same as the rest of the OPNx series - vs the 2-op ones with four different sine-derived waveforms as used by the OPL2. It also had the option of using squarewaves and pseudo-whitenoise from the onboard PSG (something that was again common to the other OPNs, but in their cases built-in to the chip rather than being separate). 51.7.16.187 (talk) 13:28, 12 October 2017 (UTC)

A more comprehensive list of devices that contained the OPL2 chip? Surely there had to be some other customers of Yamaha buying this chip before Adlib came around. --Baldur (talk) 23:09, 24 October 2008 (UTC)


 * Depends how long it had been on the market, but it seems a few arcade machines used it at least. Other than Adlib and Soundblaster/SB2.0/SBPro, however, and some of their clones either on PC or nominally PC-compatible rivals, it doesn't seem to have been massively popular, and the great majority of production probably ended up soldered to one of those four boards (the SB16 and Adlib Gold getting OPL3s instead), or inside a Yamaha branded synth keyboard. Generally it seems the OPL, OPLL, and OPN-series chips (as well as the plain PSG) were rather more popular as choices for ground-up system building, as they incorporated more functions into a single chip, and could make a wider range of sounds, meaning they were much better value despite being a bit more expensive in terms of raw unit cost and individual integration requirements, and so meant a considerable saving overall through reduced component count and system simplification. The OPLs seem to mainly have appeared as add-on parts, whether in PC sound cards, external units, game cartridges, or plug-in audio upgrades (the Master System FM module, an FM cart for the Atari ST whose name I forget, etc), as a cheap and basic-function alternative to more sophisticated and rather less affordable mainstream options like the Roland MT32 or Sound Canvas... which, funnily enough, then themselves were often positioned as upgrade paths FROM either an OPLx or OPNx machine. 51.7.16.187 (talk) 13:28, 12 October 2017 (UTC)

more samples for the yamaha ym3812 can be found here http://queststudios.com/smf/index.php?topic=2885.0 Stewievader2 (talk) 18:19, 11 February 2014 (UTC)

Sampling frequency details
While the YM3812 chip can be used with a master clock between 2.0MHz and 4.0MHz, PC sound cards like Adlib or Sound Blaster use NTSC subcarrier frequency of 3579545.4545... Hz as it is easily available by dividing the ISA bus oscillator signal by four. The chip uses the master clock divided by 72 as the sampling frequency, so the sampling frequency is 49715.9090... Hz. If an integer approximation is needed, it should be rounded to 49716 Hz. —Preceding unsigned comment added by 91.153.27.229 (talk) 08:36, 21 January 2010 (UTC)


 * That's very useful to know, but as not all uses of the chip are in those cards - and, indeed, as some systems didn't run their ISAs at exactly the correct frequency (for one thing, I've certainly seem 8mhz and even 10 and 12.5mhz as BIOS options in later immediately pre-PCI computers) - might it be worth still noting Yamaha's own figures for the output frequency range, with their fairly common 55,556hz max at one end and 27,778hz at the other? Though I suspect you could probably push it to 56000hz without too much drama, and hardly anyone would ever have run it slow enough to produce less than 30khz output... 51.7.16.187 (talk) 13:59, 12 October 2017 (UTC)

Register amount is incorrect
The register count is taken from a document that supposedly describes the FM portion of specific Sound Blaster sound card versions, where there are two YM3812 chips, one for left and one for right channel. A single YM3812 has 9 channels consisting of two operators per channel. A single operator has five registers while a single channel has three registers, plus six global registers, which equals 123 write-only registers, plus of course the single read-only status register. Anyone can verify this from the linked YM3812 datasheet, or from the YMF262 (OPL3) datasheet which has a more detailed description. —Preceding unsigned comment added by 91.153.27.229 (talk) 12:57, 7 June 2010 (UTC)


 * Ouch, now my head is spinning (well OK part of that is because I'm not very well at the moment anyway). Has this been properly corrected in the article now? It should be fairly easy to verify vs, for example, Yamaha's own datasheets, or even looking up programming guides for the chip or pulling apart some old .ROL adlib files and seeing how many different parts they address. Or indeed loading up Adlib Composer on an old computer and seeing what parameters it allows you to change and tallying them all up.


 * If we work this out in another way...
 * * nine channels x 2 ops = 18 operators.
 * * 18 x 5 = 90
 * * 9 x 3 = 27
 * * global registers, a mere 6 (funny, thought there'd be more than that; the OPNs seem to have quite a lot?)
 * * status register, just 1
 * ((90 + 27 + 6 + 1) = 117 + 7) = 124 ... seems legit.


 * Funnily enough I was only on this talk page to try and clear up whether the chip was genuinely only 2-op without even any option of combining some of the channels together into 4-op ones (an otherwise common Yamaha feature, at least where the hardware isn't inherently 4-op anyway, in which case there's usually some backdoor method to fake having additional lower-complexity channels either as intervals of the base note or completely independent)... after all the current feature list is a bit hard to read and it doesn't seem to be entirely clear. I guess the multiple types of waveform are an attempt to compensate for the otherwise very limited modulation range and introduce additional harmonics that would otherwise be difficult or impossible to achieve in their own right, and not at all available for modulation purposes? 51.7.16.187 (talk) 13:53, 12 October 2017 (UTC)

Any parallels with the ESS688?
Just wondering, as there's comparison with the ESS1688 (and its OPL3/SB16 compatibility mode) on the OPL3 page... we had a 688 card and the general FM quality seemed somewhere between OPL2 and OPL3 when played using its own drivers in Windows plus the few DOS games/apps that supported it directly. Its own compatibility mode claimed to be that of the SB Pro, which was OPL2 (x2 ??) as far as I know, and it seemed to work flawlessly to my ears. Anyone know anything about how the two might relate to each other? What there is of an actual ESS Audiodrive article at the moment isn't particularly useful... 51.7.16.187 (talk) 15:28, 12 October 2017 (UTC)