Looks like my Audiotrix 3D/XG has an issue after all. When running OPL3-SA generic drivers under windows 98 SE, the line-in is wonky . By the way, on this card the Waveblaster duaghterboard is connected to line-in.

Whenever I select line-in for recording, I have to set it to 0 to not severely clip, but even then, I still have some clipping . OK, so the daughterboard (Monster MIDI) is overdriving the input, I tell myself.

BUT, then I realize that line-in playback mixer control actually affects the recording level ! Just to re-iterate, the line-in selected for recording, not stereo-mix .

I thought that maybe the generic drivers do not suit the Audiotrix 3D/XG, so I try to force install the Audiotrix 3D/XG driver that I was using before the EEPROM flash. The driver install completes , but crashes Windows so severely that even deleting Audiotrix 3D/XG entries from device manager in Windows does not fix it . I had to restore the registry with scanreg. I tried twice with the sane result, so this not a fluke .

I think I really need that EEPROM , unless someone can confirm that this line-in level thing is normal on all OPL3-SA cards .

Monster MIDI board is indeed very loud and the line in can only take up to 5Vpp (and some boards like DB60XG have 8Vpp output that clips any existing sound card out there). 0db point on input mixer is at the very bottom and on playback mixer at first notch on most inputs as seen here :

Driver side crashing is most peculiar though...

Tiido wrote on 2020-04-24, 09:07:

Monster MIDI board is indeed very loud and the line in can only take up to 5Vpp (and some boards like DB60XG have 8Vpp output that clips any existing sound card out there). 0db point on input mixer is at the very bottom and on playback mixer at first notch on most inputs as seen here :
OptimalMixer.png

Driver side crashing is most peculiar though...

First, I would like rectify that changing Line-in Playback volume mixer does not change recording level . How I came to that conclusion, I do not know ( must have been really tired) .

I have also installed Audacity (Sound Recorder wasn't cutting it) so I can visualize the clipping directly while recording .

When I try the settings you suggested, I hear almost nothing of the daugtherboard playback through line-out, but recording line-in still clips (playback volume control does not control recording volume, which is normal), so it would seem the Audiotrix 3D/XG is not quit like other OPL3-SA cards .

Also strange is that this card was designed (or at least meant ) for the DB60XG which, as you say, is also very loud, so how do they prevent from clipping with that daughterboard ? If I understand correctly the OPL3-SA output is routed through the daughterboard so effects can be applied to PCM and FM synth and the audio of daughterboard is routed to Line-in of the OPL3-SA but also to Line-out and speaker out of Audiotrix card, which would mean the output of the DB60XG would have to be split, which would provide attenuation (3.5db ?) and this is would be how clipping is avoided with DB60XG .

Other daughterboards such as Monster MIDI do not have have line input, so the routing has to be different (changed with jumpers on Audiotrix 3D/XG card) and all the output power of the daughterboard is sent into line-in of OPL3-SA, which clips .

Also correct me if I am wrong, but I would guess that the line-input gain control in the recording mixer is digital gain after the ADC of the OPL3-SA, otherwise either there would have to be a separate signal path in the analogue mixer for line-in playback and for line-in recording, or line-in record level would affect playback volume, which is not the case .

EDIT : corrected a typing mistakes

In other words, this is likely another good reason not to buy a Monster MIDI and get a Dreamblaster instead, which has a tool to set output volume level in flash .

All difference on the 3D/XG card is in analog domain, but as I don't have that card I cannot really say more than that. EEPROM etc. will not have any effect on that aspect

Normally the clipping isn't a problem since in stock config YMF71x line out goes to line in of a WB card and the output of it goes to final sound output, where any excessive levels no longer matter. But you have to use it in alternate mode where YMF71x line out goes to final sound output and WB card output goes to line-in of YMF71x. Since clipping is happening there's no attenuation method used on the analog level... that is my understanding of how the card works.

All the gains are analog on YMF71x cards, all done by the analog mixer, there are no gain controls in digital domain at all on the card. ADC input has gain control from 0db to +22db in 1.5db steps prior to a mux that chooses between the 4 input sources seen in recording control. Even if it had negative gain possibility, it will still clip anything exceeding voltage rails and only such stuff will ever clip, therefore the designers of the chip didn't see the need for that.

MIDI devices have a Master Volume that can be changed via a Sysex or some other method. I actually made a tool for SAM9733/9704 based things to control it not so long ago since with CleanWave set there's actual cracking sounds as it digitally overflows rather than softclips like the GS set on many louder tracks which prompted me to make it in the first place. (firmware is stored on the sample ROM on SAM9733 stuff)
You can find it on SETYMF page at utils section : http://www.tmeeco.eu/TKAYBSC/DREAMFIX.EXE

Tiido wrote on 2020-04-25, 07:32:

All difference on the 3D/XG card is in analog domain, but as I don't have that card I cannot really say more than that. EEPROM e […]
Show full quote

All difference on the 3D/XG card is in analog domain, but as I don't have that card I cannot really say more than that. EEPROM etc. will not have any effect on that aspect

Normally the clipping isn't a problem since in stock config YMF71x line out goes to line in of a WB card and the output of it goes to final sound output, where any excessive levels no longer matter. But you have to use it in alternate mode where YMF71x line out goes to final sound output and WB card output goes to line-in of YMF71x. Since clipping is happening there's no attenuation method used on the analog level... that is my understanding of how the card works.

All the gains are analog on YMF71x cards, all done by the analog mixer, there are no gain controls in digital domain at all on the card. ADC input has gain control from 0db to +22db in 1.5db steps prior to a mux that chooses between the 4 input sources seen in recording control. Even if it had negative gain possibility, it will still clip anything exceeding voltage rails and only such stuff will ever clip, therefore the designers of the chip didn't see the need for that.

MIDI devices have a Master Volume that can be changed via a Sysex or some other method. I actually made a tool for SAM9733/9704 based things to control it not so long ago since with CleanWave set there's actual cracking sounds as it digitally overflows rather than softclips like the GS set on many louder tracks which prompted me to make it in the first place. (firmware is stored on the sample ROM on SAM9733 stuff)
You can find it on SETYMF page at utils section : http://www.tmeeco.eu/TKAYBSC/DREAMFIX.EXE

Thanks again for educating me and correcting my misconceptions, as well as for your your tool for SAM9733 volume . Will setting the volume this way survive a GM or GS reset ?

EDIT: I understand now that changing the EEPROM would not change anything except possibly allow use of original drivers .

The datasheet I followed doesn't mention that they are reset with the GM and GS resets. It certainly survived all sort of MIDI files I tried so I think the setting will stick.
EDIT: Thänk you very much ~