R.I.P. AY
Vložit
- čas přidán 5. 03. 2021
- Get your brand new ZX Spectrum hardware at: zxspectrum.shop/ !
The current version of the AVR AY does not have I/O, but only takes care of the sound part.
So for a replacement on a computer mainboard that uses the I/O ports, an original AY-3-8912x replacement chip is required.
But for Harlequin kits and other sound devices, it's a perfect choice.
You might notice a slightly offtune sound comparing the AVR AY to the AY-3-8912, which will be fixed in the final version that ByteDelight will be producing.
It's caused by the clock frequency on this AVR AY, which is 1.75MHz for Pentagon machines - the ZX Spectrum version will be 1.77MHz. - Zábava
I love the look of vintage logic boards. All those beautiful chips all lined up in rows.
Wow, incredible work. Thanks so much for tackling this problem!
The original AY seems to have a "fatter bass"... but the AVR AY is pretty close. The YM2149 often sounds thinner because it's waveforms run at 2x the speed of the AY
alleluja
Not the speed but rather the shape. YM has 2x the envelope resolution but the mapping of volume values to actual amplitudes is different.
Awesome work. A bit like the Nano Swinsid. Sounds great. BTW I just built one of your 128D kits and it was a pleasure. Really well laid out and documented. Thanks.
Smooth video editing skills man! 😃👍🏼
And yeah, when I wasn't looking, I couldn't hear the difference. I did hear the different sound of the third chip.
Fantastic sound, great find to make a replacement for the hard to find AY-8912
The AVR sounds perfect, when the YM came on though, I swollowed thinking my ears had popped, the left channel was just wrong somehow.
Super Ben, excellent 👍🏼
The difference is quite small with the AY-3-8912 but however I find that the sound is shallower, less dynamic with the AY-AVR.
There may be a few resistor / capacitor values to modify at the output to achieve the same result as the AY-3-8912.
Anyway, I'm really glad you found this alternative with the AY-AVR !
I look forward to the single-chip AVR version coming out of your lab 😃
The king is dead. Long live the king! I cant find any diferencie between them. Is cheaper and sounds perfectly. Great choice Ben
Use headphones. The differences between the 3 are considerable.
Ben, you are a Legend.
brilliant alternative so much better than the YM2149, you never cease to amaze me with your work
I had six 8912's with a PC interface (ISA) long ago. Nice work on emulating most of it. The I/O I'm sure could be done as well.
I’m not at all surprised that genuine working AY-3-891x chips are getting hard to find. They have been obsolete for many, many years. Production stopped in the late 1980s after the relevant General Instrument division was spin off as Microchip. Microchip only made them for a couple of years and then production stopped [any with date codes after 1990 are likely remarked older chips, are remarked Yamaha chips, or are fakes]. So we’ve been running on new/old stock that has been sitting around in warehouses and ‘pulls’ (from old boards) ever since.
But still the AY-8912 has a more full sound compared to the AVR-AY, but hardly noticeble
Could also be the volume level of the 8912 was slightly higher, but the lows of it do sound a bit better. And I need to test the latest firmware, maybe it's also a bit improved (I wonder if it can get any better though).
I found the AY sounded considerably more full than either of the others. The AVR AY sounded close but with much less bass. The YM doesn't even sound like the same chip.
I'd say the AVR AY is a passable alternative but I'd use an original AY if I had the option.
You could add a programmable clock generator and add a register to to ay firmware to control it from the spectrum. I've seen Aiden Lawrence do something similar on his vgm player boards. Great job!
I'm going to stick my neck out here and say I actually prefer the sound of the AVR-AY, at least compared to the 8910's sound - I started searching because I was messing around with the idea of using the broken toy keyboard I have to enter notes into an AY driven by an Arduino like I did with my '1943' project except without the musical competence :). I used the 8910 for that project and will probably end up re-using it for the new one if I ever get chance to do it. Form-factor/Pinout won't be an issue for me because I'm not building a Speccy clone (this time :) ) but a pin-for-pin replacement with a sensible cost in relation to the original is cool either way and would make DIY Speccy/AY interfaces possible again for nutcases like me.
I don't mind when someone has another opinion than me, I respect that, so don't worry ;)
I will do a new video soon about a comparison of new clone chips, and some other original AY-3-8910 clones.
More material to listen to ;)
The difference between the original and AVR-AY sound might be due to different internal output impedances, which in turn might affect the sound balance in the RC network that follows the output chain in Harlequin diagram. Decoupling AVR-AY outputs with capacitors might help.
I'm in my favorite professional recording studio and I can ear all the differences between them perfectly. The original AY3-8912 have the right channel balance L C R (Left Center Right), the AVR-AY is more clean and pleasant but lost the C channel signal a lot, at least 2149 sound L and R inverted, less resolution, very bad compared to the AY3-8912. These are not a real alternative for me :-)
IT depends how the AY is used. In the Spectrum, all 3 channels are commoned together into a mono signal. In the later "black" +2A/+3 the audio circuitry has some severe deficiencies that distort the sound.
Other systems connected the audio outputs differently. Some spectrums were modified for ACB (left, common, right) or ABC sound.
@@jaycee1980 my judgement is about what sounds in the video, I don't know if is possible to create a combination with other zx spectrums, but if we are in the same zx I'm expecting the same conditions, this is what the "clone" meaning or not. Sounds better than the original one? Definitely not. :-)
None of these have left, centre or right channels. The AY-3-891x and the Yamaha YM2149 have three tone generator channels, each of which is brought out to a separate pin. In most home computers, all three channel outputs were connected together, to produce a single mono output. Yes, it is possible to use the three output pins separately and then mix to two “stereo” channels. But as with a lot of other things in life, there is no single standard. So how they sound does depend on the hardware configuration and on how the composer programmed the chip...
@@Mark1024MAK this is what I listen in front of speakers. Even the theory behind the chip, the sound on the video is very clear :-)
@@Mark1024MAK what is so hard to understand? I'm judging the sound on the video that's so simple. The video show how these chips sounds and there is a left center and right. You can program the chip as you want, but in this comparison the sounds speak easy, we can discuss infinitely how to use the chip, but nothing will change the audio on this video. They sounds same? NO, The new one sounds better? NO. That's it mate.
Hi there, I listened to this comparison carefully and I must say it was a good choice to choose this tune as it contains lots of sound tricks and stuff. I must say the AVR AY emulation sounds near perfect. The only differences that I heard were that the AVR AY sound was quite a little bit higher in pitch and flat. I guess the presence of flatness is a feature of the emulation, as this same effect appears on C64 SID versus the ARMSID replacement chip. Anyway if the user isn't an audiophile, the difference between the original AY and AVR AY is almost inaudible. Thanks for the video!
The slightly different pitch is caused by the AVR AY used for this video, had a clock of 1.75MHz set for Pentagon machines.
The version I will be using for products will have a clock of 1.77MHz.
Dat is een tijd geleden, dat ik een ZX Spectrum + had. Ben nu volop bezig met de C64’s en de AMIGA 500/1200’s. Zag zojuist je website en helaas was er veel out of stock.
Zal mooi zijn om een HARLEQUIN 128K REV 2D ASSEMBLED + ROM met een mooie COMPLETE ORIGINAL PLUS CASE FOR ZX SPECTRUM INCL. NEW MEMBRANE. En daarop een BRAND NEW DIVMMC ENJOY! PRO ONE WITH CASE, INCL. PREINSTALLED SD CARD. Hopelijk kan ik het op je website kopen. Wat anders, dan jaren lang klooien met mij C64/A500/A1200’s.
De ZX-AY lijkt me geweldig voor Demos.
Got one of the AY’s as well as some of Texas sound chips, but having them is one thing, but getting sounds out of them something quit different.
Jus recently obtained (a bl0ddy $100] an SGS-ATES M110 - a DCO (digitally controlled oscillator). But many of cool chips for making sounds do belong to the past 😢.
And don’t forget that those AY chips could also generate 1bit pcm audio by combining all 4 soundchannels together and then quickly alternate the volume of them to create a waveform😁
the chip which emulates frogs' croaking really well 😁
Sounds great. I do have a question though. Why are you working on a single chip version when it is cheap enough to sell the dual AY version? TurboSound will now just be a stock feature if this chip is used for all hardware. Will anybody want a 3 channel option?
In order of best to least.... I do not own any of these machines so my ears are not fan boy biased....
1. AY-3-8912..... warmest sounding of them all.
2. AY-AVR ....... not as warm but has better high end definition.
3. YM2149..... weakest of them all
Over all ... they all work fine and it's really minor the differences. The AY-3-8192 is definitely the warmest of them all. I'd be fine with any of them. That's what equalizers are for. :D
The question is, which mode is the Yamaha YM2149 in during this test?
This is the second time in recent weeks I have seen a "replacement" chip that uses its own crystal rather than rely on an external clock like the original. That's a strike against in my book.
You're right and wrong.
The microcontroller requires its own external clock signal, so that's mandatory.
But it seems that the limitations of the selected solution, doesn't allow space for an external clock signal for the AY chip to take into account, so instead it's just 'calculated' and generated internally.
I received another solution which I think uses an FPGA (need to check what exact chip is used there, but I couldn't get that clone to work yet, so haven't spent more time on it) which uses the external clock, but obviously that will be a much more expensive solution.
I think the 8912 sounds more blocky in the ADSR department in contrast to the AVR Y, where the echoes seem to sound more overlapped.. Does the 2419 seem a bit lighter sounding or was that mixing levels when recording lol ?
AY-8912 vs AVR not mutch difference but original for me better sound
11:48 We Are Alive The Outro Credits. Oh, the famous part and tune!
Does AVR AY emulate IO port too, or is it useful only for sound? Which is amazing achievement, but i am using often RS232 compatible with ZX Spectrum 128k too and it needs one 8 bit IO port.
mind = blown
the original ay sounds a lot more dynamic and open than the others, the avr ay seems sorta compressed? if that makes sense. i dont know much about these chips and exactly how they work, but i think ive seen some modulate up or down in steps really fast to produce a sound rather than doing it smoothly? id probably think the avr ay has more resolution step wise which may be causing some of it? im not sure.
Either way, Would love to know more about these chips and how they work! i've seen *some* videos on these kinda chips but ive since forgotten any specifics and im not confident enough to talk much more about them since i don't know enough how they work to do so without feeling like an idiot.
to those reading this, i hope you have a wonderful week :)
Let's face it, the Sega master system is more what the spectrum 128k should have been like (but, with 16+ sprites per line) or a Sinclair QL like graphics, but with pallet. From memory the QL hardware was originally to incorporate a z80. I had a spectrum and QL. I communicated with Sir Clive. The QL was great, but needed to have a pallet and better drive system.
But, an interesting thing O found out, was the z80 had undocumented instructions in the free space around the outside of the chip on the wafer. This was truely legendary stuff, as you can set up a protected mode with it and a special chipset. It's like the 8 bit version of the x8086, but with a protected mode that means it could have done a primitive alternative to the 68008, but with 64KB spaces. This means that a QL based z80 spectrum could have been done, while waiting for the 16 bit z8000 to replace the z80. They had trouble getting out the z8000, but like with the early 16 bit 6502, they could have done essentially, a 16 bit widened version of the original z80 circuit (with undocumented instructions documented, and a few more). It all makes me appreciate what they were trying to achieve compare to the 6502. We were so close. What happened, cheapness, no game philosophy to the dating QL. However, imagine protected mode CPM OS, GEM window GUI, and enhanced graphics. So close. But, the QL, like the electron suffered due to memory handling performance issues, I found out, which slowed things down. I would buy such a machine back then, and Clive would get his business machine. Put the business model in a small square box to put under the counter. Excellent for mum and pop stores, POS Terminals, and some office machines at a fraction of the PC price at first. If it was me, I would have looked at enhancing graphics and maybe Sega technology, for a joint project. They do games, we sell the machines into the computer and business market, game home and business version. Extend the Sega/MSX architecture to 512+ pixel width, and more sprites. The game machine gets the standard resolution, the business version gets the enhanced resolution and memory. In a year or two, the home/games version also gets the high resolution features. Eventually, with increased 16 bit processor, it competes in the space the mega drive competed in, instead.
But, what really happened. It seems that the Commodore Amiga chipset technology might have been licensed to Motorola, as they had early plans for a similar featured chipset, which disappeared, maybe after the Commodore / Atari Amiga fiasco. It has been announced, and people were planning machines based on it. The Tandy Coco 3 was originally to come out earlier with the chipset, but they scrapped and did the present design instead. Could this be something that influenced the 68008 decision (the chipset had variable bus width) as a future upgrade? The in development timing of the technology, is interesting. The flare team actually based their chip on the idea of a stripped down Amiga equivalent, removing redundant features that did simar things. They would have known about the Amiga like Motorola chipset. That chipset should have been the 128k and 256k etc during Amstrad. They went on to develop the Atari Jaguar, which was the most powerful at release, but regrettably not for Sinclair.
Over at Commodore, the designer of the Commodore 128 was to use an Motorola chip I think was this chip. But was talked into using an chip Commodore had, which was supposed to be equivalent. It wasn't. There was compromises in the 128 aswell as QL and Electron. It is dumb founding they didn't just do a Vic 3 expanded version, in house. Latter, they did a suspiciously enhanced chil for the c65. Me, I would have twerked and palleted the Vic 2 to get more sprites per line and more sprite functionality and better graphics modes and tile handling. Compared to the more sprites and higher bandwidth, the enhancements would be relatively minor. A 128 business system, could have a seperate 16 bit or 8 bit bus for graphics, with a bridge write chip inbetween. Integrating the chips together on a long dip, to reduce cost. Such a system could have multiple chips in the dip. Including memory.
Please not what the c65 sounds a bit like.
All, so close, but so far away.
Compared to eh.. silence I take the AVR any day!
Maybe should you add a lowpass filter to the AVR AY output
I'd like to see someone build a computer based on the CP1600.
@ByteDelight. I suppose that the idea is to sell the TurboSound option, as it's cheap enough to add some value to the fact that in truth you are not using a TRUE AY8912 chip and loses something in the way... Additionally, it will be possible to upgrade somehow the firmware of these circuits once they are installed? It's planned ?
I will start with producing a drop-in replacement of the AY chip, without I/O, so for Harlequin kits and ZX-AY.
Later I will work on more features, but someone else is already doing that, so not sure if I need to.
Anyway, for now I need something to be able to keep offering hardware with AY sound.
Sound like YM chip was connected differently or improperly so 2 channels were swapped and 3rd channel almost muted.
It was a bit harder to fully compare the YM2149 because you swapped the stereo (I was checking on headphones)
yeah and muted middle channel too much.
Ever listen to the song afternoon delight?
Was interested in the avr-ay as a vectrex replacement however the vectrex use I/O as controller inputs
Just found out from someone in the community, that am AVR-AY version with I/O has been developed recently (2021) by someone else.
Will look into it.
@@ByteDelight cool very interesting ! I have made converter boards to fit a AY8910 in the vectrex a bit tight for space but they work
How to get it?
Try "Across the Edge" demo on this device... And cry... :(
AY is richer one, two, bleeds bitfully to have an upright character.
Cheap altrnative? No prices? I think i can still solder out an AY from my broken spectrum.
Original AY and YM chips are both O.K. Original old ZX games are writen for AY chip and some russian ZX music (for russian ZX clones) is writed on real YM2149 then logicaly only on YM chip sounds as original.
I think an FPGA would probably do a better job at emulating the AY-3-891x then a micro-controller, though with enough time and effort on the software side the AVR AY could get even closer then it is now.
I agree, but an FPGA is completely different, so would require a totally new piece of software programming. The ATMega chips have the benefit that they are cheap, have 16 bit timers (for PWM). But they're not perfect no.
The ideal solution would be following the exact circuit of the original AY chips, but I don't think that information is available. Maybe we could get there by reverse engineering. But knowing exact values of the analogue parts would be a must.
So I am happy with getting close.
And in the meantime other clones have been developed, so I need to dig into what currently is the best one for the bucks, to bundle with e.g. ZX Spectrum clone kits, or act as a substitute for an original faulty one on boards.
Avr is also scarce
Ingenuity Unlimited.
The original chip sounds best
The SoundginZ will be able to duplicate this chip.
Nay.
Just run "scroller" demo for AAA - you will hear half of the instruments are not there.