What are the differences between the "AY-3-8910" chip and the "YM2149" chip? Why does it matter?
Talk about whatever you want. General forum rules still apply.
Moderator: Staff
- dmbnation41 Offline
- Posts: 19
- Joined: 2024-01-02, 1:49:28
- Contact:
What are the differences between the "AY-3-8910" chip and the "YM2149" chip? Why does it matter?
I was told that the "YM2149" chip is a clone of the "AY-3-8910" chip. The "YM2149" chip can even make samples, but I am not sure if the "AY-3-8910" chip can do the same thing too. So, that's why I am questioning this.
Re: What are the differences between the "AY-3-8910" chip and the "YM2149" chip? Why does it matter?
YM2149 is Yamaha's re-implementation of AY-3-8910 chip.
Of more noticeable sound changes:
P.S. None of these chips can play digital sounds natively, however if you have CPU time, or dedicated hardware, it is possible to make chip playback samples in a rather cheap and noisy way
P.P.S. There is 1 or 2 I/O ports that convert register value into a series of parallel outputs, with some tinkering it is possible to put together a serial resistor digital to analog converter and get some more low quality digital sounds at the cost of high CPU usage, or a dedicated timer — it's the same principle as behind COVOX sound device.
Of more noticeable sound changes:
- YM has 5-bit hardware volume envelope compared to 4-bit on AY, so tracks that use it for volume get more smooth sounding volume slides, and tracks that hack envelope for producing sounds sound more "soft", with less pronounced harmonics.
- YM chip uses different volume table: when you decrease channel volume on AY chip it has somewhat smaller step between loud sounds, and bigger steps on quiet sounds, on YM chip these are perceived as more regular intervals.
P.S. None of these chips can play digital sounds natively, however if you have CPU time, or dedicated hardware, it is possible to make chip playback samples in a rather cheap and noisy way
![Devious :devious:](./images/smilies/devious.gif)
P.P.S. There is 1 or 2 I/O ports that convert register value into a series of parallel outputs, with some tinkering it is possible to put together a serial resistor digital to analog converter and get some more low quality digital sounds at the cost of high CPU usage, or a dedicated timer — it's the same principle as behind COVOX sound device.
- dmbnation41 Offline
- Posts: 19
- Joined: 2024-01-02, 1:49:28
- Contact:
Re: What are the differences between the "AY-3-8910" chip and the "YM2149" chip? Why does it matter?
[Replying to: tails_]
Finally, I was waiting for someone to reply!
I did realized AFTER the post was in, that the "AY-3-8910" chip can use samples... I just forgot to mention that earlier.
Finally, I was waiting for someone to reply!
I did realized AFTER the post was in, that the "AY-3-8910" chip can use samples... I just forgot to mention that earlier.
Re: What are the differences between the "AY-3-8910" chip and the "YM2149" chip? Why does it matter?
There is a YM SSG variant with a dedicated DAC out there. I believe it's the YMZ285.
The above post is pretty much on point, except for some other notable things:
a) GI's design was entirely made by hand. If you look at the output envelope or measure its volume output, the GI 891x chips have a very broken, vaguely logarithmic, volume curve. I could have sworn the 8910 datasheet mentions this, but those assume the output is perfect. The AY8930 datasheet shows this discrepancy, though, but only in 8930 expanded mode (5-bit output). Which is rather funny, since it also shows how much they broke the expanded mode's volume curve/envelope output as well.
b) Yamaha's design was entirely remade in some sort of CAD software, hence the fixed volume curve. I've heard they licensed the design from GI, but I don't have a source for this. At any rate, they changed the envelope's resolution to 5-bit. and they re-used this design for and inside of everything. But basically the result is that the output is a lot more bassier and doesn't have overtones that are as audible as the 8910 envelope. And there's the clock divider pin as well.
-
Because both lines were super popular, each one got a boatload of variants. GI made a bunch of "reduced pin count" ones that effectively just internally disconnected the I/O buses from an external set of pins, and also made one for the Intellivision with a different pinout and the envelope has... adjustable amplitude. And then there's also the later AY8930, which was formed when GI spun off their microelectronics division as a standalone subsidary.
Yamaha on the other hand, licensed out the SSG core to companies interested in integrating it into other products (eg: Sunsoft 5B ASIC or some MSX-Engine ASICs), or made even smaller low pin count variants. Yamaha also made a CMOS variant (YM3439) and a bunch of other oddity ones in the mid 90s that were essentially the same chip, sometimes with extra features. Supposedly the Sega VDP PSG is also based on the YM2149 design, with aspects from it forming the basis of the VDP PSG (such as the altered noise LFSR from the TI SN76489, or the differently implemented square wave generation), but I haven't gotten much of a source on this.
To be quite honest, the rabbit hole for this chip design is weirdly deep and I kind of find it surprising there's this much to it, honestly.
The above post is pretty much on point, except for some other notable things:
a) GI's design was entirely made by hand. If you look at the output envelope or measure its volume output, the GI 891x chips have a very broken, vaguely logarithmic, volume curve. I could have sworn the 8910 datasheet mentions this, but those assume the output is perfect. The AY8930 datasheet shows this discrepancy, though, but only in 8930 expanded mode (5-bit output). Which is rather funny, since it also shows how much they broke the expanded mode's volume curve/envelope output as well.
b) Yamaha's design was entirely remade in some sort of CAD software, hence the fixed volume curve. I've heard they licensed the design from GI, but I don't have a source for this. At any rate, they changed the envelope's resolution to 5-bit. and they re-used this design for and inside of everything. But basically the result is that the output is a lot more bassier and doesn't have overtones that are as audible as the 8910 envelope. And there's the clock divider pin as well.
-
Because both lines were super popular, each one got a boatload of variants. GI made a bunch of "reduced pin count" ones that effectively just internally disconnected the I/O buses from an external set of pins, and also made one for the Intellivision with a different pinout and the envelope has... adjustable amplitude. And then there's also the later AY8930, which was formed when GI spun off their microelectronics division as a standalone subsidary.
Yamaha on the other hand, licensed out the SSG core to companies interested in integrating it into other products (eg: Sunsoft 5B ASIC or some MSX-Engine ASICs), or made even smaller low pin count variants. Yamaha also made a CMOS variant (YM3439) and a bunch of other oddity ones in the mid 90s that were essentially the same chip, sometimes with extra features. Supposedly the Sega VDP PSG is also based on the YM2149 design, with aspects from it forming the basis of the VDP PSG (such as the altered noise LFSR from the TI SN76489, or the differently implemented square wave generation), but I haven't gotten much of a source on this.
To be quite honest, the rabbit hole for this chip design is weirdly deep and I kind of find it surprising there's this much to it, honestly.