I have a custom hat that (amongst other things) has a GPS module (ZOE M8).
Power for the GPS module is taken from Vdd on the 40-way connector via a transistor switch (so I can shut it down if I so choose). And to improve acquisition time over reset, the GPS module's backup power is fed from a supercapacitor via a diode.
This has all worked fine on Pi 4B. But when I tried this on a Pi 5, the Pi 5 crashed horribly, in that I could not access the filing system.
I could, for example, pwd. But not ls, nor sudo shutdown. And attempts to reconnect by ssh were rebuffed:
ssh -Y pi@192.168.1.71
kex_exchange_identification: read: Connection reset by peer
Connection reset by 192.168.1.71 port 22
The crash seems to happen when I turn on power to GPS with my transistor switch. When this happens, Vdd momentarily drops from 3.3V to 2.9V while the supercapacitor charges. It is all back to normal within a few microseconds. I get the same amount of glitch on Pi5 and Pi4. But on Pi5 this seems to take down the filing-system while the Pi4 keeps running as if nothing had happened.
Is this expected? Does this match other people's experience?
I attach an oscilloscope trace IMG_6434 GPS Turn On.jpg showing the glitch in the Vdd line (yellow) and GPS_Vdd=VRef (magenta).
As well as the schematic of the transistor switch GPS Power Switch.png
And the schematic of power to the GPS module GPS Power with SuperCap.png (VRef is fed directly from GPS_VDD).
Thanks,
Rover.
Power for the GPS module is taken from Vdd on the 40-way connector via a transistor switch (so I can shut it down if I so choose). And to improve acquisition time over reset, the GPS module's backup power is fed from a supercapacitor via a diode.
This has all worked fine on Pi 4B. But when I tried this on a Pi 5, the Pi 5 crashed horribly, in that I could not access the filing system.
I could, for example, pwd. But not ls, nor sudo shutdown. And attempts to reconnect by ssh were rebuffed:
ssh -Y pi@192.168.1.71
kex_exchange_identification: read: Connection reset by peer
Connection reset by 192.168.1.71 port 22
The crash seems to happen when I turn on power to GPS with my transistor switch. When this happens, Vdd momentarily drops from 3.3V to 2.9V while the supercapacitor charges. It is all back to normal within a few microseconds. I get the same amount of glitch on Pi5 and Pi4. But on Pi5 this seems to take down the filing-system while the Pi4 keeps running as if nothing had happened.
Is this expected? Does this match other people's experience?
I attach an oscilloscope trace IMG_6434 GPS Turn On.jpg showing the glitch in the Vdd line (yellow) and GPS_Vdd=VRef (magenta).
As well as the schematic of the transistor switch GPS Power Switch.png
And the schematic of power to the GPS module GPS Power with SuperCap.png (VRef is fed directly from GPS_VDD).
Thanks,
Rover.
Statistics: Posted by ClassicRover — Mon Aug 05, 2024 2:18 pm — Replies 6 — Views 79