I don’t know anything about your specific issue, but have you traced out the path that the 5v supply is coming from and measured the resistance along that path? If there’s any active components in there like a voltage regulator or some other driver, perhaps that’s where the problem is.

Upon further investigation, I believe the problem is not with the SB16, but instead with the "passive" game port to MIDI adapter I have. I originally thought the SB16 contained the MIDI current loop circuitry, and that my game port MIDI adapter was completely passive, but that appears not to be the case.

Instead, it appears that the SB16 TTL MIDI signal is routed directly to game port pin 12, and that pin is not directly connected to any of the MIDI output pins. I am thinking there must be some components within the cable, and perhaps those are faulty.

To riplin's point, +5V on the game port, pin 1, seems to be routed to MIDI pin 4 with a 2.2K resistance through the cable, likely explaining the large voltage drop I see when the cable is in use.

in the following page there is a schematic for a working gameport midi cable: http://midi.teragonaudio.com/hardware/pc_intfc.htm
simpler "out only" version here: Simplest way to get Gameport -> MIDI out
and if you do not really want to make your own then something like the 10eur DB15MIDI might be what you are looking for: https://www.serdashop.com/DB15MIDI

Evidence is pointing to the adapter as the issue, so now have questions it. I am a bit nostalgic for it because I have had it since I was a kid, and it's a pretty nice adapter (other than not working with the SC88) that brings out the joystick ports. I am curious if the adapter has become defective somehow over time, or if the problem is simply with its original design.

Unfortunately, it is potted, so I cannot examine its contents without destroying it. But I am really curious about the 2.2K resistance I measure between the +5V on the joystick port (pin 1) and MIDI pin 4. I wonder if that high resistance is intentional to limit the amount of current coming from the SB16, either being overly cautious, or perhaps because the SB16 was not designed to supply that much current. Either way, the MIDI spec is 5 mA, so the adapter cannot possibly meet that spec with the 2.2K resistance in place.

If the high resistance in the original adapter is intentional, then should I be concerned about buying or building an adapter which pulls the full 5 mA from the SB16?

In summary, I built an externally-powered MIDI thru using inverters and a 4.7K resistor as a load instead of an opto-coupler, and the SB16 with the weak MIDI out adapter was able to drive it just fine. But in the end, I went with a cheap commercial MIDI thru that the SB16+weak MIDI out adapter was able to drive, because it looked better than my homebrew thru.

I still don't know if my MIDI out adapter is defective, or just designed with such a small drive current, but in the end, I have a working system, so I'm happy. Thanks for all the advice!