HATs and other add-ons • Re: Rpi 5 built-in power switch ("PSW"), and the ATX Watchdog hat

Is that an advertising post or a request for support?

A request for support with a lot of details on what I'm trying to accomplish, and what components I'm using to do it...

Think it's advertisement as the power button on Pi5 is well described in the documentation https://www.raspberrypi.com/documentati ... wer-button

The J2 jumper is located between the RTC battery connector and the board edge. This breakout allows you to add your own power button to Raspberry Pi 5 by adding a Normally Open (NO) momentary switch bridging the two pads. Briefly closing this switch will perform the same actions as the onboard power button.

That doesn't distinguish between (for example) a pair of pads in parallel with the PSW switch, versus a pair of pads that send a signal to a completely different I/O pin that also causes the PMIC to turn on and/or the Pi to shut down. In both cases, "briefly closing this switch will perform the same actions as the onboard power button", as the doc page says, but the two setups are different from the perspective of what I'm considering.

So what additional info would you get from a schematic?

I think I said this in the original post, but I am trying to figure out:

• What power rail is the pullup resistor connected to? 5V? 3.3V? 1.8V or 1.1V for the CPU or DRAM? (...DRAM is super unlikely...)
• Related, what happens on the pi side if this rail gets power from the pullup on the ATX hat, when only 5VSB is provided, but 5V to the rest of the pi is off? I might need to figure out a diode of some kind, for example. Or this might sink the whole idea, especially if the voltages are different.
• I only thought of this question just now, but: Which of the two pads is connected to GND? If I get that backwards I'll short the ATX hat's connector to GND all the time (the two PCBs' GNDs are connected through the GPIO header), and the button will act pushed all the time.
  • That one can be determined with some resistance measurements actually; the GND pin on the GPIO header is documented.
• What other components are also connected to that circuit? How is the PMIC involved -- is it the simple circuit that I expect, with a pullup resistor and the switch to GND, with the line going to one of the PMIC's inputs? Or is it something more complicated, to handle the 5-second delay for the hard shutdown, with for example an RC circuit? I'll need to make sure none of those components can be affected by the ATX hat in some unexpected way. Or vice versa.

Measure them and see if pushing the actual button impacts on their levels (any logic analyzer/scope/multimeter should do).

Sure. Once I got access to my multimeter again earlier today, I measured a short between the two pads when the button is pushed. That tells me that the button and those pads are connected, so there's a small amount of hope. But when the button isn't pushed, there's still a 10k-12k resistance (which varies over time) between the two pads, so *something* else is going on there. Possibly as simple as a 10k pullup to +5V, and an extra variable resistance from +5V to ground through one of the unpowered components.

As I mentioned above, I did not think about the which-side-is-GND question when I had access to the meter. I'll have to see about that one tomorrow.

..
It controls an ATX supply, and needs a switch connected to two of its pins for powering everything on. With a jumper (that isn't part of the kit, but has a mounting location present on the PCB), it will provide regulated 5V power to the pi via the GPIO header. (It will provide whatever max current the connected ATX power supply can do.)
..

Think about that after you've read https://www.raspberrypi.com/documentati ... wer-supply; seems like you've misse something.

The only thing I see there that you might be thinking I missed is the 5.1V requirement. But that's a requirement on VBUS (since all the supplies there are USB), not on the rest of the 5V rail, AFAICT from both the raspberry pi 4B and 3B+ partial schematics. In fact there's a 5V zener (...from the part number, looks like a TVS) between VBUS and GND, which will suppress any voltage above 5V in any case. So plugging 5V into the right side of that TVS will have the same effect as 5.1V at USB VBUS, on the left side.

Unless you mean the recommendation for a 27W supply for the Pi 5 maybe? The particular ATX supply I'm using is specced at 15A on that rail (75W), with a limit of 100W on the sum of the usage on 5V and 3.3V. (And I'm doing ~nothing with 3.3V.) So it can provide way more than the recommended 27W supply can.

Or is there something else that my brain is skipping right over?

Thanks!

Statistics: Posted by bkadzban — Thu Apr 25, 2024 6:07 am

---