In some situations power delivery to the Pi is not the issue. It is power delivery from USB to the SSD. I ran into this myself and it was weird. * When the Pi

In some situations power delivery to the Pi is not the issue. It is power delivery from USB to the SSD. I ran into this myself and it was weird. * When the Pi 4B first came out, I connected a cheap Inland (Microcenter house brand) SSD, booting from that, and it ran fine for a while. (Months?) "Better" SSDs like Crucial MX100 had problems under heavy load and I suspected that the instantaneous power demands of the SSD were not met by the USB port on the Pi. (The Pi 4B supports UAS and older Pis did not and the older ones never had problems with USB connected SSDs.) * At some point it started to run into difficulties with the SSD disconnecting under heavy load, requiring a power cycle to restore operation. At that point I began using a powered USB hub to work around this. * Recently a user on some forum (which I have forgotten) suggested that one particular SSD model from Kingston worked fine w/out the need for a powered hub. I acquired one and tested [1] and found this to be true. Just for grins, I also tested some other SSDs that had previously only worked with a powered hub and they worked fine when directly connected. This leads me to believe that the RpiOS folk have been tweaking some settings in the "firmware" [2] that have modified the power delivery from the USB ports. [1] Testing involved running several disk benchmarks as well as 'stress-ng' to load the storage as much as possible. [2] I'm unclear what the Rpi folk mean by "firmware". As an embedded systems developer "firmware" is something written to non-volatile solid state storage like flash or eeprom but when I updated firmware on a Pi, it installed a new kernel too.