Fourays: A Tribute to the AY-3-8910 | Hacker News Reader

Fourays: A Tribute to the AY-3-8910

doug.lon.dev

44 pointswdfx26 comments

Sharing the introductory post for a chip-tune synthesizer I wish to build.

The ZX Spectrum has this sound chip (or almost, for the pedants) and there's a lot of interesting music composed for it. Mostly from Eastern Europe apparently? Here's a good selection: https://zxart.ee/eng/music/top-100/

(I had a bunch of time last year and learned a lot about some retro sound chips, including this one, and while at it converted two tunes I liked to MSX ROMs, as I have an MSX and the sound chip is the same: https://blog.qiqitori.com/2023/05/playing-psg-tunes-or-even-...)

This was the 128K spectrum. The original ones (16K and 48K) had a 1-bit CPU driven speaker, worse than the PC’s original speaker (which could do background square waves thanks to a programmable timer).

Even the 48K had interesting music, when programmers figured out how to do pwm with up to 5 channels. But not a lot of it.

I think the AY sound chip is overrated, nothing but pure 50% square waves. You move up to the next step, and you get the NES sound chip, with 12.5% and 25% duty cycles. Then you move up another step, and you have the Turbo Grafx sound chip, with short custom waveforms.

Looking at this list of sound chips[0] is an interesting tour through past encounters with computer and video-game systems. Even today music synthesizers are still trying to come up with new techniques for generation and filtering/modification.

[0] https://en.wikipedia.org/wiki/List_of_sound_chips

OTH it's impressive what you can do with multiple AYs when they have a dedicated CPU to drive them. For instance the Bomb Jack arcade machine has a separate sound board with 3x AY and a 3 MHz Z80 CPU and it sounds pretty nice:

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

Here's Jochen Hippel's(Mad Max in the Atari ST demoscene) use of the AY's on the Guryss hardware(x3 AY). It's a quite lovely tune. They use timers to make "buzzers" and "SID-sound".

https://www.youtube.com/watch?v=K5TxsZNJpBA

Multiple AYs is a different story.

You can approximate 12.5% and 25% duty cycle square waves by playing multiple tones at different octaves and a different volume level.

For sure the 50% square wave is a limitation, but sometimes such limitations feed creativity. That's also why I intend to couple the oscillators to filters, so there's another dimension to play with.

Amazing work. One random thought; Would giving each AY its own clock make stack/unison much richer?

That hadn't really occurred to me actually. I'll keep it in mind as I start mixing multiple chips together.

I am limited in the current design for clock sources but I intend to implement pitchbend so it'll be possible to detune two voices that would otherwise be sounding in unison.

What a fantastic project, I hope you get it all to work and it is definitely inspirational material.

Thank you. I definitely intend to get this together, there's a lot of satisfaction to be had using a device you designed and built yourself.

Any YouTube videos of it making audio?

I'm not quite at that stage, perhaps in the next couple of weeks and for the next article.

FYI, that is NOT what genuine Microchip DIP markings look like. Compare, for example, https://www.lcsc.com/product-detail/Analog-To-Digital-Conver... .

Also there is zero chance that some of them have bottom-marked lot codes and some don't, given the same top date/facility code. Dead zero. And the cavity IDs on the DIP molds should be the same style, not randomly mismatched.... I'll stop here, but this is classic gray market remarking stuff. These are random old chips that have been relabelled.

They may well still be AY-3-8910s in there -- the gray market is very creative -- and may well be usable for you, but be cautious. If they look like they're misbehaving, the chances are much higher than usual that it's not your fault.

I really don't understand why chip resellers re-mark chips like this. It's always obvious and it gives me the idea that they have something to hide. I've paid more for obviously used chips that I can tell are genuine than "new"/"refurbished" examples like the ones in this article.

I love the AY-3-8910 and I've bought chips that were remarked in this fashion, fortunately they all worked ok and sound like the real deal.

There were such chips sold under the Microchip brand. I may still have one somewhere in a box.

Yeah it's strange, they've clearly been relabeled and maybe only the top two I tested are real. Even if that is true I still have a couple of new toys to play with. I'll test the rest tomorrow.

I finally got around to testing them all. It turns out that of the 2 I already picked off the top of the stack, one works perfectly and the other works mostly but is slightly intermittent.

The remaining 8 are all either dead or not AY chips at all.

I WAS WRONG: https://doug.lon.dev/blog/2024/fourays/chip-testing/

My testing was flawed, in fact all but one of the chips work. Pretty happy about that :)

can you explain what "cavity IDs on the DIP molds" means?

Things that are mass-produced in very high volumes by injection molding are never molded just one at a time: the molds have multiple cavities to make many parts at a time in a single molding shot. Very often, each cavity will have a marking to identify it: sometimes just a number, sometimes alphanumeric, sometimes dots. So you can tell that if everything coming out of cavity B12 is having quality problems, you probably need to inspect that cavity.

Now take a look at the picture in the article of the undersides of the DIP-40 parts. There are two circles, which are either ejector pin locations or just things that look like them. (Ejector pins are the things that kick finished parts out of the mold once they're done.) Inside a few of those circles you find those cavity markings. I'm not sure exactly why the ejector pins and cavity markings (appear to?) coincide, but it's pretty common out there in the wild. You'd need someone who's more of a mold specialist and less of an EE to answer that one.

I've looked at Lego bricks more than I care for and I've noticed similar mismatches over the many years that a particular brick was produced, but they were all still original Lego. It could simply be that at some point a completely new mold was made and it has its markings in a different spot.

There are also so called modular molds with interchangeable inserts.

Yes, mismatches over the years are completely normal... but not with the same date code and site code.

I love how well everything is explained and how clear he is regarding every step.

Reply on news.ycombinator.com