Interesting...
there are a couple of people here that have broked XR-385 cards IIRC, maybe it's just the DAC... but why does it fail?

That's what I'd like to know. All voltages on the other pins seem okay. Maybe it's caused by ESD damage. Anyway, pin 8 should be a high impedance input pin, not a voltage source.

I don't think connecting this pin permanently to GND would be a good solution. The DAC requires a short pulse on this pin to start it's calibration cycle. It would also short (and possibly damage) pin 15 on the CPU, which can only source 2mA.

Yep, replacing is the only option here IMHO
Have you found a source for the DACs? utsource seems to have them

Well there's the solution then 😀
They do call it a "3V 16BIT ADC&DAC WITH BUILT-IN PGA" though, maybe they confused it with another AK chip?

edit: Yes they did, but the part number is correct so I placed an inquiry anyway

DACs are coming my way, can take a week or two though.

Now that I'm back into experimenting with these cards, I noticed something VERY unusual....
The XR385 I've always been using had a weird problem with Duke3D: near the end of the title song, you could hear one spurious/hanging guitar note, and when the track loops, the bassdrum disappears and doesn't come back until you restart the game. Now I never thought much of it, I thought it was fairly normal as it has always done this, until I tried another XR385... which didn't have this problem. Last night I've been trying out all the cards I have and only the first one had this problem.

Now I just tried to record it to show it to you guys, and guess what... The first card now behaves like it should!
I can't think of anything that could cause this, other than perhaps the extension cable, which may have some capacitive coupling between adjacent wires, or act like an antenna (picking up noise). But that wouldn't make any sense, because if that was the problem, the other cards would do the same thing. And if noise was the problem, why does it only happen on the Duke3D title song, consistently showing the same behaviour?

I can't get my head around this. Anyone else ever noticed this with their cards?

I had an issue with DB50XG: some of its instruments sounded like noise or were otherwise garbled. Turned out to be a hard scratch that abridged few ROM tracks together. Was tracked down and fixed by a specialist. Seems previous owner treated it like a piece of junk 🙁

Can somebody tell me whats wrong with the NEC XR385 when it doesnt work anymore?? The first card came loose when playing.. Now it doesnt work anymore.. I like to know which component on the card could be dead. Dont want to throw the card away but want to fix it..

Check the voltage between pin 8 of the AK4510 and ground. It should be 0V exactly. There's a via (hole) right in front of the chip where it's easy to measure.
Here's a picture to explain:
http://img22.imageshack.us/img22/6272/6qce.jpg

Another way to check if it still does something, is to touch the DC blocking caps with one finger, and poke around the circuit board with another finger. If you can hear something resembling music around the XQ731B0 chip it might still be working. (Yes, that's how I discovered my card was not completely broken. Please don't laugh 🤣)

bristlehog wrote:

I had an issue with DB50XG: some of its instruments sounded like noise or were otherwise garbled. Turned out to be a hard scratch that abridged few ROM tracks together. Was tracked down and fixed by a specialist. Seems previous owner treated it like a piece of junk 🙁

Of the four cards I have, this one is actually the least scratched. I did modify the output circuit on this card, though that's all analog and shouldn't influence the digital part at all.

So.. DACs still haven't arrived. But why even bother with those things if you just received some nice... S/PDIF transmitters?

Just tap into the I²S signals...

And you've got noise-free digital bits from the XR385/DB60XG! (with a little jitter, though..)

Rest in pieces!

Here's a recording for comparison:

Analog output, modified for flat frequency response and zero distortion: https://soundcloud.com/oplawaai3/blood-cblood7-db60xg-custom
Digital S/PDIF ouput: https://soundcloud.com/oplawaai3/blood-cblood7-custom-bank-s
Both sound more or less identical, but note the complete absence of noise in the quiet parts, around 1:20...

Anyone interested in a schematic?

so it's the dac after all... awesome work there jwt27!
(don't have a digital receiver and never used one btw..)

This is also the first time I ever used the S/PDIF connectors on my X-fi card... I don't have any other digital audio stuff. Analog is usually good enough, except when the noise level is really high. Like in the XR385.

Next step... S/PDIF output on my turntable 🤣

Here's the schematic: http://img545.imageshack.us/img545/6862/8x1t.png

You can use a DIT4096 or 4192, there's no difference except for the maximum supported sample rate. For the transformer, the datasheet says you'll need to use a 2:1 transformer. I found a 1:1 transformer works just fine with 100 and 330 ohm resistors in series. I got mine from a 3COM Etherlink III network card.

The same schematic can be used for just about any other soundcard or daughterboard with Philips I²S audio signals. Even the OPL3, I think...

And here I thought Asahi Kasei DACs were quality... well the bizarre part is that S-YXG50 has the exact same problems with floor noise.

What does bank 0x29 mean?

d1stortion wrote:

And here I thought Asahi Kasei DACs were quality... well the bizarre part is that S-YXG50 has the exact same problems with floor noise.

What does bank 0x29 mean?

The DAC itself is not that bad actually. Its SNR is 84dB minimum (A-weighted). This S/PDIF mod raises that to 96dB, the maximum dynamic range of a 16-bit signal. I think the main problem is that it's mounted inside a PC, with noisy power supply, EMI all over the place, etc.

Bank 0x29 means I've set the bank MSB (controller 0) to 0x29 on all channels before starting the game. I think it sounds better on this particular track.

edit: Here's another comparison. On the first 0.2 seconds from that Jimi Hendrix demo midi, you can hear a guitar fret noise if you turn up your volume to 11 (it's at -69dB..)
http://imageshack.us/a/img703/3549/y9t2.png

Never noticed that before? Neither did I. Probably because the analog recording looks like this: http://imageshack.us/a/img191/9602/rmku.png

edit 2: Here's the same sample on the S-YXG50 softsynth: http://img12.imageshack.us/img12/6541/1q2e.png

edit 3: S-YXG50 is actually clipping across the whole track, even worse than the stock XR385...

I have one of my MU10's S/PDIF modified and i used the same chip like you (DIT4096), it only works on MU10, DB60XG/XR385 though as all other XG devices use 18 bit wich DIT4096 do not support in hardware setup 😒

Stefan_L wrote:

I have one of my MU10's S/PDIF modified and i used the same chip like you (DIT4096), it only works on MU10, DB60XG/XR385 though as all other XG devices use 18 bit wich DIT4096 do not support in hardware setup 😒

Cool 😀 Does it improve the signal as much as in the XR385? I'd imagine a standalone synth would have less noise problems to start with. Just thinking about it, a daughterboard does not only have to deal with a noisy PC environment, the signal also has to pass through some noisy ISA card. The YMF719 for example, which I would consider "good", only has 78 and 84dB minimum SNR on the ADC/DAC..

What DAC does the MU10 use, anyway? Or is it just a DB60XG in a box...?

I also received the AK4510s today, think I'll replace it tomorrow. I still need the analog input... Digital input would be nice, too, but I think that would be a bit harder to do. An S/PDIF receiver will want to generate its own clock signals, which the XR385 already provides. And you'll need a sample rate converter in order to use it with the OPL3, for example.

Today I soldered another DIT4096 on a breakout board and hardwired it for use on the XR385:

Other side: http://img96.imageshack.us/img96/2100/8vod.jpg

EDIT:

Stefan_L wrote:

all other XG devices use 18 bit wich DIT4096 do not support in hardware setup 😒

Just thinking out loud here... If the audio format is right-justified 18-bit, MSB-first, and SDATA is held low until the actual audio data starts, wouldn't you be able to use the 24-bit right-justified mode?

jwt27 wrote:

Cool 😀 Does it improve the signal as much as in the XR385? I'd imagine a standalone synth would have less noise problems to start with.

What DAC does the MU10 use, anyway? Or is it just a DB60XG in a box...?

Just thinking out loud here... If the audio format is right-justified 18-bit, MSB-first, and SDATA is held low until the actual audio data starts, wouldn't you be able to use the 24-bit right-justified mode?

The MU10 already has good sound so S/PDIFing it is not a necessity 😀

Yes... it's just a DB60XG in a box.

And sorry i don't have any knowledge in S/PDIF... it was several years ago i did this circuit and i never know much back then either, it was more trial and error since i got no help with it.

Stefan_L wrote:

The MU10 already has good sound so S/PDIFing it is not a necessity :-) […]
Show full quote

jwt27 wrote:

Cool 😀 Does it improve the signal as much as in the XR385? I'd imagine a standalone synth would have less noise problems to start with.

What DAC does the MU10 use, anyway? Or is it just a DB60XG in a box...?

Just thinking out loud here... If the audio format is right-justified 18-bit, MSB-first, and SDATA is held low until the actual audio data starts, wouldn't you be able to use the 24-bit right-justified mode?

The MU10 already has good sound so S/PDIFing it is not a necessity 😀

Yes... it's just a DB60XG in a box.

And sorry i don't have any knowledge in S/PDIF... it was several years ago i did this circuit and i never know much back then either, it was more trial and error since i got no help with it.

You could try it, it might just work. I'm willing to help you out (not that I know a lot about S/PDIF or digital audio, just getting into this stuff...)

Of course, if the analog signal is already noise-free then there's not much to be gained from a digital connection.

Oh and I have some good news, just replaced the AK4510 on the broken card and that fixed it 😀
Robin4, if you need one just let me know.

Here's the final S/PDIF mod:

Transformer will be mounted on a separate stripboard and connected to the pin header on top.

For those looking to eliminate clipping from the analog outputs, see the µPC4570 op-amp to the left of the AK4510. This is the easiest way to do it and can be easily removed if you don't like it: Two drops of solder on pins 1-2 and 6-7 will provide a flat frequency response without clipping. This will attenuate the output level by about 16dB, though, so the noise level will effectively increase.

If you only want to eliminate the clipping, but want to keep the non-flat frequency response and high gain, replace the 4.7K resistors (R9 and R10) with 10 or 12K.