Do you have a reference for that?
SCSI-1 drivers are meant to be able to sink 48mA per line. Here's an old IBM document that states that explicitly:
http://ps-2.kev009.com/eprmhtml/eprmb/h1293.htm, RaSCSI folks have this to say:
https://github.com/akuker/RASCSI/wiki/Transceiver-Comparison, and pulling the datasheet for a totally random SCSI controller out of my downloads folder, here's what it has to say:
The STM32's GPIO lines, per the datasheet, have an absolute maximum rating of 25 mA (this is from the
STM32F103C8 datasheet which is I believe what my bluepill boards have on them:
While this datasheet doesn't give an absolute maximum rating for the sum of all GPIOs, the maximum total current that's allowed to flow through the ground pin is 150mA:
And this would be easily overloaded by even only 3 of the lines sinking the maximum 48 mA.
Even if we look at a SCSI bus which only contains the BlueSCSI and some termination, SCSI terminators are supposed to aim to supply 24 mA ish per pin of source current when necessary (see
"Understanding the single-ended SCSI Bus", chapter 4). So even just with a normal standard-compliant terminator, you're potentially overloading the power handling capabilities of the MCU if six lines are sinking at once.
So, it's out of spec, the question is how much does this matter in practice? And that's a more complicated question I think, because it's really a question about risk. This will shorten the life of the microcontroller, and eventually the thing will die. But how long "eventually" is hasn't really been quantified yet, nor is what the upshot of it dying is. I think it's reasonably unlikely it would take the SD card out with it, so if you're in a position to consider bluepill boards as basically consumable (although with a long replacement window), the risk might be OK. (Though it feels rather wasteful)
For hobby use—it's probably OKish. I've got bluescsis of a slightly modified design in some of my compacts that I don't use very often, and I'm OK with the above for that. If it was something I used daily I would probably reconsider; and if, like the OP, I was doing
actual work on the things, I wouldn't even consider BlueSCSI.
The real problem is that as far as I know the BlueSCSI folks get weirdly defensive when you point out, however politely, that their hardware is running outside of spec (or say anything even remotely critical about their hardware), and so this seems unlikely to get fixed unless they can pretend it was their own idea somehow. Big sigh.