powering Raspberry Pi from bench supply with USB

[metebalci]

I know I can power RPi from GPIO and using it like that, but I wonder if I can with USB. So I get a USB cable and connected +5V and GND to bench supply, and to comply with USB power spec. I shorted the D+ and D-. However, RPi is still not properly booting, I dont have a screen attached so not sure if it writes anything but I see some activity in LEDs then it seems like it is rebooting. What am I missing ?

[Ian.M]

A screen?

Before the Pi 4, the Pi's power in port was very dumb, IIRC simply Vbus and Ground with no data pins connected.  The Pi 4 has some resistors to signal the charger its expecting 5V power (which was incorrectly implemented on early  board revisions).

Many cheap USB cables have too much resistance to reliably power even a Pi Zero!  Measure the 5V at the GPIO header, to check you are actually delivering the correct voltage to the board, and scope it to check it isn't dipping.  *DON'T* be tempted to crank up the voltage past 5.25V in an attempt to compensate for a crappy cable.  Instead, cut it down to 3" long and splice it onto much thicker wires back to your PSU.

[metebalci]

A screen?

Before the Pi 4, the Pi's power in port was very dumb, IIRC simply Vbus and Ground with no data pins connected.  The Pi 4 has some resistors to signal the charger its expecting 5V power (which was incorrectly implemented on early  board revisions).

Many cheap USB cables have too much resistance to reliably power even a Pi Zero!  Measure the 5V at the GPIO header, to check you are actually delivering the correct voltage to the board, and scope it to check it isn't dipping.  *DON'T* be tempted to crank up the voltage past 5.25V in an attempt to compensate for a crappy cable.  Instead, cut it down to 3" long and splice it onto much thicker wires back to your PSU.

I checked with the scope and sometimes it seems it actually falls down, not sure to what extent it is OK. But it seems to be working now. I dont know what was the problem. I guess it might be because of the cables.

[Ian.M]

I believe anything below approx 4.63V will trip the undervoltage detector.  USB connector contact resistance can be significant - I have one Pi with a SATA HDD (in a USB<=>SATA 'pod') that needs its power in USB jack and plug cleaned and reseated at regular intervals to prevent the Pi detecting undervoltage, and the HDD restarting.

[rpiloverbd]

Personally I always recommend using an official RPI power supply to power up the RPI. You know, RPI needs 5v,2A. Is your power supply really giving this output?

[Ian.M]

5V 2A is grossly inadequate even for a Pi 2B if you have four fully loaded USB ports.   OTOH if all that's on there is a single unit load wireless HID mouse/keyboard dongle, the same Pi 2B could be happy on a 500mA PSU (that's 500 Western mA as Chinese mA are usually more wimpy, like 'Sears horsepower').  See https://www.pidramble.com/wiki/benchmarks/power-consumption

[metebalci]

Personally I always recommend using an official RPI power supply to power up the RPI. You know, RPI needs 5v,2A. Is your power supply really giving this output?

Just in terms of basic specs that should not be a problem I think for my power supply (RS NGE) (or is there sth special about USB power?). I also normally use USB adapters.

[metebalci]

5V 2A is grossly inadequate even for a Pi 2B if you have four fully loaded USB ports.   OTOH if all that's on there is a single unit load wireless HID mouse/keyboard dongle, the same Pi 2B could be happy on a 500mA PSU (that's 500 Western mA as Chinese mA are usually more wimpy, like 'Sears horsepower').  See https://www.pidramble.com/wiki/benchmarks/power-consumption

My use case for the purpose of this question is only RPi with SPI/GPIO connections, no USB devices, even no keyboard etc. but wifi. It is consuming around 0.5A like this.

[mazurov]

I shorted the D+ and D-.

This looks wrong to me. I has been powering USB devices from a bench supply for many years and don't remember having any issues. The only connections are plus to vbus and minus to gnd, this worked for all devices so far including pies up to v.3.

[metebalci]

I shorted the D+ and D-.

This looks wrong to me. I has been powering USB devices from a bench supply for many years and don't remember having any issues. The only connections are plus to vbus and minus to gnd, this worked for all devices so far including pies up to v.3.

It is from USB charging specs. Less than 200ohms between D+ and D- means it is a dedicated charging port, not capable of enumeration, so just like a power supply. I guess if a device gets power by USB (but no data comm use) or maybe for old devices it may not matter, but it matters for some devices at least.

[Renate]

... not capable of enumeration ---

What do you mean? Everything is capable of enumeration! "That's one USB power supply!"

I used to run a shorting wire from the four USB sockets to the regular 5V.
Then I could feed the power in on a USB A plug instead of the wimpy micro USB.

[metebalci]

Not an expert, but I take it to mean no other/some more/normal USB signalling. No handshake, negotiation etc.

[Renate]

Oh, yeah, I get USB enumeration and low-level USB.
I guess that you're not familiar with Count von Count from Sesame Street.
He likes to count things.