Question Potential addition to the ATX spec - serial line digital readouts / control?

[VirtualLarry]

Here's a though. Maybe it's time to update the ATX PSU spec again, and this time, take that one wire, that's no longer used, in the ATX 24-pin spec, since like forever, and turn it into a serial digital data line (along with ground as the other signal line), and use it like an I2C bus, sort of, to communicate between the ATX PSU and the 24-pin ATX mobo.

I think that this idea has a lot of merit, and it would be an easy thing to do, going forward, with newer mobos and chipsets, IMHO.

They already updated the SATA ATX power spec, so that a signal on the +3.3V line, going to a 3.5" higher-capacity (but not reason it couldn't apply across the board) HDD, SHUTS DOWN the HDD.

If they can re-purpose lesser-used power lines for signalling like that, on the SATA connector, why can't we re-purpose that (-5V, or -12V, I forget, it was only use for onboard sound card amplifiers like 10 years ago) line that is currently basically an NC on current modern ATX PSU 24-pin connectors. Why can't we turn it into an I2C-like serial data line, where the PC can poll the PSU and read out voltages, current, temps, all kinds of good data out of the PSU's digital control logic chip(s).

They could call the updated spec "ATX digital I/O".

 

[mindless1]

It would not surprise me if this eventually happens, once such features are integrated into both ends and are practically free to put in the silicon.

For the time being it would require additional circuitry in the PSU, and on the motherboard, so it seems like something that won't appeal to the OEMs which are most of the target market, due to raising both their product costs and support costs from people worried their PSU isn't at 12.000V or the temperature isn't 24C.

AFAIK, "most" PSU do not contain any "digital control logic chips" that sample and provide this kind of data. Controlling it doesn't need to be more than an analog setup even if it's arguably digital switching.

I suppose I am against the idea (for the time being, until the controllers have this integral) because once it's in the spec, any PSU that doesn't have it would be non-compliant to spec, which raises my cost too if everyone starts doing it yet I don't really need this feature.

It would be handy to know if PSU temps shot up due to a faulty fan but as far as current, what do I care as long as the system is stable? Just about any enthusiast level mobo can give you the voltage at the mobo which is the relevant measurement point for stability, though it can't tell you the latency in fast switching circuits, which used to be a problem back in the Athlon era which moved them to use of the 12V rail for CPU, after Intel had already recognized this and moved to 12V for CPU.

 

[serpretetsky]

I think it's pretty common in server power supplies (pmbus). But leave my ground alone! Just use both the negative signals. Also the i2c circuitry on the mobo will have to be -12V tolerant.

 

[VirtualLarry]

I think it's pretty common in server power supplies (pmbus).

I did not know that.