VOGONS


Picking a sound card for a low-end 486

Topic actions

Reply 40 of 91, by derSammler

User metadata
Rank l33t
Rank
l33t

The story is rather simple: at some point, Yamaha stopped selling the OPL3 chip to OEMs. That's why big companies like Creative and others created their own version of an OPL3 compatible FM synthesis (CQM, ESFM, etc.).

Many Asian companies did not care however and just manufactured the OPL3 on their own, disguising it with funny names. Replicating a chip wasn't a big thing in the 90s. I don't even think the blue prints were a huge secret.

Reply 41 of 91, by Marmes

User metadata
Rank Member
Rank
Member
appiah4 wrote:

I really don't think the CS4289 is a rebranded OPL3, CS4232 can be accompanied by genuine OPL3 or clone OPL3 chips but CS4289 is actually a Crystal FM implementation that I don't mind but think is inferior to CS4236..

Well I think it is, because there is no datasheet for cs4289 therefore it's highly possible that they are rebranded or 100%clones of ymf289, and it's pin compatible with ymf289. I find no difference in sound.
I tested my card with a genuine ymf289. Works great!

Reply 42 of 91, by appiah4

User metadata
Rank l33t++
Rank
l33t++

This story gets weirder and weirder by the day. Third day of testing results:

Variable changed: WS set to 1

YMF719: SetYMF continued to produce white noise in WSS testing but now a tone was actually audible behind the noise.
ES1868: Card was not detected in one ISA slot, but detected in an ISA/VLB slot. DOS installation went smooth, however ESS drivers in Windows resulted in garbled PCM audio. Removing ESS drivers and replaced them with SB1.5 drivers resulted in a working card but only FM Synth driver available was Windows default Adlib, which sucks.
OPTI929: Community driver set the card up beautifully and I managed to install official Windows 3.1 drivers. Everything worked perfectly. Until I went back to DOS and ran a game. Strangest thing: FM audio AND the game itself ran about 15% faster when I used it for FM music. 😵
SB16: Everything works perfectly.

So the SB16 is back in, and I have a theory. But first, here's a list of crystals on each card:

YMF719: 24.576MHZ, 33.8688MHz
ES1868: None
OPTI929: 24.576MHz, 16.9347MHz
SB16: 14.318MHz 24.0000MHz, 46.61512MHz

Yeah, I think that 14.318MHz is what's making the SB16 work problem free. For whatever reason, the ISA bus on this board is probably running out of spec and (well?) over 8MHz. Why, I do not know. FSB is set to 33MHz and the Bus divider is set for <=33MHz.There is, however, one header right above the 14.3MHz crystal on the motherboard market JS3 that I have no idea what it is.

ECS-486-UL-P101.jpg

Images of this board with 33MHz CPUs have that jumpered as 2-3 (here, hereand here) while mine is open (and came with a 40MHz CPU originally). So my LAST bet is to jumper that and hope stability improves and the cards actually work when the ISA frequency is resolved..

This is also probably why the VLB IO controller didn't work as well, though I'm really not keen on trying it again.

Anyway, some new theories to test tomorrow. For tonight, it's sleepy time.

Retronautics: A digital gallery of my retro computers, hardware and projects.

Reply 43 of 91, by gdjacobs

User metadata
Rank l33t++
Rank
l33t++
Marmes wrote:
appiah4 wrote:

I really don't think the CS4289 is a rebranded OPL3, CS4232 can be accompanied by genuine OPL3 or clone OPL3 chips but CS4289 is actually a Crystal FM implementation that I don't mind but think is inferior to CS4236..

Well I think it is, because there is no datasheet for cs4289 therefore it's highly possible that they are rebranded or 100%clones of ymf289, and it's pin compatible with ymf289. I find no difference in sound.
I tested my card with a genuine ymf289. Works great!

It is absolutely not a clone of the YMF289. The CS4289 incorporates Crystal Semi's own variant of OPL3 along with some spatial audio effects processing. I find the FM sound of this chip to be very different from genuine OPL3 and, to my ears at least, very harsh and objectionable.

All hail the Great Capacitor Brand Finder

Reply 44 of 91, by Marmes

User metadata
Rank Member
Rank
Member

As I told, ymf289 is a direct dropin ic, no spacial stuff there, it sounds exactly the same as ymf289.

gdjacobs wrote:
Marmes wrote:
appiah4 wrote:

I really don't think the CS4289 is a rebranded OPL3, CS4232 can be accompanied by genuine OPL3 or clone OPL3 chips but CS4289 is actually a Crystal FM implementation that I don't mind but think is inferior to CS4236..

Well I think it is, because there is no datasheet for cs4289 therefore it's highly possible that they are rebranded or 100%clones of ymf289, and it's pin compatible with ymf289. I find no difference in sound.
I tested my card with a genuine ymf289. Works great!

It is absolutely not a clone of the YMF289. The CS4289 incorporates Crystal Semi's own variant of OPL3 along with some spatial audio effects processing. I find the FM sound of this chip to be very different from genuine OPL3 and, to my ears at least, very harsh and objectionable.

Reply 45 of 91, by appiah4

User metadata
Rank l33t++
Rank
l33t++

Don't make me break out this card and re-test it.. The last time I checked it sounded like CS4236's CrystalFM and not Yamaha OPL3.. I can test again if you guys like.

Zoltrix_Audio_Plus_6400_3_D_Pn_P_AV309_CS4232_CS4289.jpg

By the way, using a jumper to close JS3 on this motherboard as I noted above appears to have done a whole lot of nothing. I don't like this board. As much as I want an all UMC system, I want a functional fast and reliable system more, so I will replace the motherboard with a Chicony CH-471A SiS motherboard and hopefully be able to use a VLB IO card with it too. If that works out, I will swap the SB16 out (yet again!) for the Yamaha card and give it another go.

Retronautics: A digital gallery of my retro computers, hardware and projects.

Reply 46 of 91, by Marmes

User metadata
Rank Member
Rank
Member

I have exactly that card, but without 3D. I removed the CS4289 and soldered an original YMF289, no diference in sound, same FM. As creative licensed some CT chips from yamaha, I believe that crystal did the same. Why not ymf262 you may ask? Well Crystal makes some of the finest DACs, so ymf289 is the best and cheapest choice. That's my point of view. There are no datasheets for CS4289 and I really doubt that a company like crystal would make a clone of that kind of chip. If you see CS4232 datasheet, you see it metion yamaha opl3, not other chip.
I don't know if 3d there is constantely enabled, but I don't see diference in sound.

appiah4 wrote:
Don't make me break out this card and re-test it.. The last time I checked it sounded like CS4236's CrystalFM and not Yamaha OP […]
Show full quote

Don't make me break out this card and re-test it.. The last time I checked it sounded like CS4236's CrystalFM and not Yamaha OPL3.. I can test again if you guys like.

Zoltrix_Audio_Plus_6400_3_D_Pn_P_AV309_CS4232_CS4289.jpg

By the way, using a jumper to close JS3 on this motherboard as I noted above appears to have done a whole lot of nothing. I don't like this board. As much as I want an all UMC system, I want a functional fast and reliable system more, so I will replace the motherboard with a Chicony CH-471A SiS motherboard and hopefully be able to use a VLB IO card with it too. If that works out, I will swap the SB16 out (yet again!) for the Yamaha card and give it another go.

Reply 47 of 91, by appiah4

User metadata
Rank l33t++
Rank
l33t++

If what you sah is true then that makes CS4232+CS4289 one of the most desirable ISA sound cards: Silent DAC, SB Pro/WSS, Bug-Free MPU-401, Wavetable Header and OPL3. I will test it out, but I don't remember it sounding like OPL3.

Last edited by appiah4 on 2019-08-21, 13:48. Edited 1 time in total.

Retronautics: A digital gallery of my retro computers, hardware and projects.

Reply 48 of 91, by Marmes

User metadata
Rank Member
Rank
Member

If it sounds diferent to you, you can allways buy a ymf289 and replace, I will take pics of my cards and post here. I have 2 different cs4232 Cards.
1 with ymf262 and one with ymf289.

appiah4 wrote:

If what you sah is true then that makes CS4232+CS4239 one of the most desirable ISA sound cards: Silent DAC, SB Pro/WSS, Bug-Free MPU-401, Wavetable Header and OPL3. I will test it out, but I don't remember it sounding like OPL3.

Reply 49 of 91, by appiah4

User metadata
Rank l33t++
Rank
l33t++

I may actually try this mod out.. But like I said I'm almost certain CS4289 does not sound like a genuine OPL3 core but maybe the differences I heard were normal differences between YMF262 and YMF289 (as now that I research them I realize they don't sound the same either: YMF289 has a different pitch, it also sounds like it almost has some distortion/clipping/artifacting and sounds more squeaky/metallic overall: https://www.wikizero.com/en/Yamaha_YMF262). So yeah, it may indeed be a YMF289 rebrand or just a CrystalFM chip pin compatible with YMF289.

On another note, do check out that Wikipedia page.. ESS ESFM sounds much better than either YMF262 or YMF289 to me.

Last edited by appiah4 on 2019-08-21, 14:03. Edited 6 times in total.

Retronautics: A digital gallery of my retro computers, hardware and projects.

Reply 50 of 91, by derSammler

User metadata
Rank l33t
Rank
l33t
Marmes wrote:

I removed the CS4289 and soldered an original YMF289, no diference in sound, same FM.

The question is what did you use to come to the conclusion: "no diference in sound, same FM"? ESFM and CQM can also sound identical to OPL3, depending on what you are listening to. You need to listen to OPL3 music that really make use of the OPL3, then you'll hear a difference, guaranteed!

Reply 51 of 91, by Marmes

User metadata
Rank Member
Rank
Member

Well I'm not a big fan of opl3, for me it sounds nice, and I hear no difference, but for purists, they may find something different. These cards, may have changed it's charateristics over the years, filters , etc, so I really don't know what a real opl3 sounds like.
I also think if it was a pin compatible made by crystal, there would have been a datasheet. Or you can use the yamaha ymf289 datasheet 😉

Here they are:

1.jpg
Filename
1.jpg
File size
1.62 MiB
Views
696 views
File license
Fair use/fair dealing exception
2.jpg
Filename
2.jpg
File size
1.12 MiB
Views
696 views
File license
Fair use/fair dealing exception
3.jpg
Filename
3.jpg
File size
1.72 MiB
Views
696 views
File license
Fair use/fair dealing exception

Reply 52 of 91, by gdjacobs

User metadata
Rank l33t++
Rank
l33t++

Compare for yourself. Even the intro should convince you that they sound different.
https://www.youtube.com/watch?v=p6IITanvS6Y

All hail the Great Capacitor Brand Finder

Reply 53 of 91, by Tiido

User metadata
Rank l33t
Rank
l33t

That video shows only the horribly broken CS4236 all-in-one chip which came later out of Crystal stuff. Other crystal all-in-one chips in the same line (5, 7, 8, 9) are nowhere near as broken, and they are still not the stand alone FM chip that is being discussed here.

Crystal I think licensed CS4231 all-in-one chip to Yamaha to base YMF71x design off (and it is 99% compatible), and I think Crystal in return got to do actual OPL3 in form of the chip that's pin compatible to YMF289.

T-04YBSC, a new YMF71x based sound card & Official VOGONS thread about it
Newly made 4MB 60ns 30pin SIMMs ~
mida sa loed ? nagunii aru ei saa 😜

Reply 54 of 91, by appiah4

User metadata
Rank l33t++
Rank
l33t++

Isn't CS4235 the one that is actually broken? CS4236 (IIRC) just has a very high sampling frequency so the usual FM artifacting we are used to (and some of the OPL music rely on) is absent on it. In games it actually sounds good.

But if you are also insistent that the CS4289 is actually a chip licensed back from Yamaha, then I will have to test it out for sure. Expect recordings in a few days as I sort out my 486; I just swapped out the ECS motherboard for a Chicony CH-471, and since I no longer need the disk overlay software I will need to reformat the drive and do a clean install.

Retronautics: A digital gallery of my retro computers, hardware and projects.

Reply 55 of 91, by Tiido

User metadata
Rank l33t
Rank
l33t

Could be 4235, I remember one of them was outright broken while others were ok. I don't have a CS4289 card to test stuff with but I would think it makes sense that it is same as original YMF289 given the same pinout and that there was some sort of deal between Crystal and Yamaha as far as CS4231 and YMF71x chips go.

I'm curious to hear any of the results about CS4289 ~

T-04YBSC, a new YMF71x based sound card & Official VOGONS thread about it
Newly made 4MB 60ns 30pin SIMMs ~
mida sa loed ? nagunii aru ei saa 😜

Reply 56 of 91, by Marmes

User metadata
Rank Member
Rank
Member

Well, that doesn't show anything, there are many videos that show Crystal chips are nice and SB pro ,SB16 are bad. Those cheap cards usually have different output circuits from what it's recommended by manufacturer. You should show me a video comparing cs4289 with ymf289 because that's what we're talking about. Further more, Crystal never intended to replicate the yamaha opl chip, if you see datasheets, in any chip from CS4236 to CS4239 you can install an original Yamaha OPL3 chip for retrocompatibility, the implemented FM on those chips are crystal, CS4235 was a budget chip without the options of the others. CS4232 uses original Yamaha chip, or rebranded, it can also use crystal FM when paired with something like cs9233. So in my opinion any good quality card with CS4232 and OPL3 and I mean CS4289, YMF262 or YMF289 is the way to go.
Actually I do have a soundcard with both ymf289 and crystal FM, I will do a comparison. I just need to know what uses the opl in a way that we can see differences.

gdjacobs wrote:

Compare for yourself. Even the intro should convince you that they sound different.
https://www.youtube.com/watch?v=p6IITanvS6Y

Reply 57 of 91, by gdjacobs

User metadata
Rank l33t++
Rank
l33t++
Tiido wrote:

Crystal I think licensed CS4231 all-in-one chip to Yamaha to base YMF71x design off (and it is 99% compatible), and I think Crystal in return got to do actual OPL3 in form of the chip that's pin compatible to YMF289.

The CS4231 isn't an all in one chip. It's a codec. Yamaha had no reason to license tech from Crystal as they have lots of knowhow from elsewhere in their organization.

CrystalFM doesn't use the same operations as the YMF262 or YMF289. There are many examples where CrystalFM sounds very different from the Yamaha variants which indicates substantial differences in the synthesis method. This is not a function of aliasing as the YMF289 operates on the same sampling frequency as the CS4289 (which I didn't know about prior to this thread) yet delivers virtually indistinguishable output compared to a YMF262.

All hail the Great Capacitor Brand Finder

Reply 58 of 91, by Tiido

User metadata
Rank l33t
Rank
l33t

Perhaps not an all in one part then, but the WSS part of the YMF71x is near identical clone of CS4231, the datashit even says so and I have personally verified it to be the case. Even the ID bytes are same and a bug of CS4231 is there also (upper and lower bytes get swapped in a WSS 16bit DMA transaction sometimes). Only difference is that the DAC calibration procedure is slightly simpler, as few things have been removed. And it is still a guess that there's some deal but given the available info it isn't all that unlikely.

It is not verified here right now that CS4289 is a CrystalFM part, not many even know about this chip, much less have it. The stuff found in Crystal's later chips certainly is their own stuff and there's no doubts about that. I have deep experience in actually emulating the Yamaha style FM and know exactly how things work and what can produce any of the differences observed in other implementations so I'm very curious to hear some recordings of a CS4289. Anything with noise type sounds in it are going to reveal a lot, if something is different the spectrum produced will also differ in a measurable way. Th CrystalFM parts I have heard seem to differ in the way envelopes progress and there's differences in power to linear conversion process which will result in modulation indexes to differ in higher feedback cases leading to differences in spectrum and tonality which can be easily seen.

T-04YBSC, a new YMF71x based sound card & Official VOGONS thread about it
Newly made 4MB 60ns 30pin SIMMs ~
mida sa loed ? nagunii aru ei saa 😜

Reply 59 of 91, by gdjacobs

User metadata
Rank l33t++
Rank
l33t++
Tiido wrote:

Perhaps not an all in one part then, but the WSS part of the YMF71x is near identical clone of CS4231, the datashit even says so and I have personally verified it to be the case. Even the ID bytes are same and a bug of CS4231 is there also (upper and lower bytes get swapped in a WSS 16bit DMA transaction sometimes). Only difference is that the DAC calibration procedure is slightly simpler, as few things have been removed. And it is still a guess that there's some deal but given the available info it isn't all that unlikely.

Interesting. Can you contrast with the AD1848 WSS codec?

Tiido wrote:

It is not verified here right now that CS4289 is a CrystalFM part, not many even know about this chip, much less have it. The stuff found in Crystal's later chips certainly is their own stuff and there's no doubts about that. I have deep experience in actually emulating the Yamaha style FM and know exactly how things work and what can produce any of the differences observed in other implementations so I'm very curious to hear some recordings of a CS4289. Anything with noise type sounds in it are going to reveal a lot, if something is different the spectrum produced will also differ in a measurable way. Th CrystalFM parts I have heard seem to differ in the way envelopes progress and there's differences in power to linear conversion process which will result in modulation indexes to differ in higher feedback cases leading to differences in spectrum and tonality which can be easily seen.

It'd be interesting to have a standard test pack for comparing different FM synth clone implementations.

All hail the Great Capacitor Brand Finder