2
\$\begingroup\$

I have a Raspberry Pi Zero 1.3 and a standard USB OTG unpowered hub. When I plug in certain USB devices, the RPi reboots instantly. I believe this is due to "inrush":

https://www.raspberrypi.org/forums/viewtopic.php?t=25803

What is the correct way to modify the system to prevent this inrush problem (eg. can I just solder a large electrolytic capacitor across the supply rails)?

\$\endgroup\$
2
  • 2
    \$\begingroup\$ try using a powered USB hub \$\endgroup\$ – jsotola Aug 11 '20 at 4:29
  • \$\begingroup\$ @jsotola Some (but not all) powered USB hubs solve the problem -- it seems there need to be separate power rails for the USB connection and the hub power supply. However a powered hub is not a feasible solution for the situation I need to deploy this device in, so I need to find some way to modify the Raspberry Pi Zero to handle inrush, if possible. \$\endgroup\$ – Luke Hutchison Sep 6 '20 at 18:49
2
\$\begingroup\$

You need a needle-tip solder iron and study hands to solder in electrolytic capacitors. The bottom side might have space if it is not mounted behind a touchscreen display, like I do often. Use a fine 31 mil (0.031") diameter rosin core solder, 60/40 mix. No one cares if it is not RoHS compatible as it has lead in it.

It is only 5 volts so any small diameter can capacitor rated for 10 volts will do. Shop at DigiKey, Newark, or Mouser for electrolytic capacitors that are long of body but narrow in diameter. The blog about inrush is correct, so to fix it try 100 uF capacitors, 1 by the USB-C power connector (bottom of the board), and 1 by the extra USB ports where you find both 5 volts and ground pins. Two of them should absorb the surge caused by plugging in certain USB devices.

Be sure the main power supply is 5 volts 3 amps or more, with no extensions or spliters added to the power cable. The Pi can be power hungry when running programs. Also consider adding a battery backup as a hat for the Pi.

I am adding a photo as to where to place a single 100 uF 10 volt narrow body electrolytic capacitor. It can be hot-glued on top of the Pi logo without affecting the slow signals to the header. NO glue or touch U3, as it processes any imported video you plug in.

enter image description here

\$\endgroup\$
5
  • \$\begingroup\$ I got a decent 5V 3A power supply to ensure that the brownout was not caused by insufficient power supply. This did not solve the inrush problem. Given that the power supply must have a fairly large internal capacitance, as part of its voltage regulation circuitry, I don't have high confidence that just soldering two 100uF capacitors across the rails is going to work (although I still need to try that). Is there anything else that can be done? I guess I can try to find a powered USB hub? \$\endgroup\$ – Luke Hutchison Aug 13 '20 at 7:46
  • 1
    \$\begingroup\$ I would try everything sensible and low cost. The power traces on the Pi are narrow and cause brief voltage drops with sudden loads. Not making this up. \$\endgroup\$ – user105652 Aug 13 '20 at 10:02
  • \$\begingroup\$ I tried connecting a 100uF capacitor, and even a 1000uF capacitor, at the points you indicated. Neither solves the problem -- the Raspberry Pi Zero still reboots when I plug in a USB device that draws reasonable power (e.g. a USB Ethernet dongle). This is the case both with a Raspberry Pi Zero, and a Raspberry Pi Zero W. \$\endgroup\$ – Luke Hutchison Sep 6 '20 at 18:13
  • \$\begingroup\$ Then you are 1) Drawing too much current with your add-on devices or 2) Not supplying enough current to PWR IN. Power supply is too weak or power cable is too skinny and possibly too long, so a sudden load will always cause a brief voltage drop. Consider using a power spliter before the Pi so it does not have to source the current. A 5 volt Pi will reset at only 4.6 volts to 4.8 volts, not much room to work with. \$\endgroup\$ – user105652 Sep 6 '20 at 22:46
  • \$\begingroup\$ As far as I can tell, none of that is true per se (I mean, the current draw is too high, sure, but it's not a deficiency of the power supply or cable -- I even bought a 3A power supply to prove it -- but also the devices that cause the reboot work fine on any other computer, so the power draw is not a problem for other USB systems). I have been researching the design of circuitry that solves the inrush problem, and it typically caps initial current draw at some lowish value, and delays connecting the data lines until the voltage has settled. \$\endgroup\$ – Luke Hutchison Sep 7 '20 at 23:11

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.