1

I have reviewed a lot about powering the PI using the 5V pins on the expansion connector but I have not seen the solution to this problem. Here are some of the tests I ran:

(Setup: Nothing plugged in the micro USB connector, PI is headless, no USB devices, ethernet connected, PI 3B+, Raspbian buster)

  1. 30A 5.01V switching PS plugged into ping 2+4 and grounded: PI starts, the red LED goes on, the activity LED comes on, goes off, device reboots.
  2. 5A 5.05 SPS: same problem.
  3. Benchtop PS set for 5.00, 5A capable: same problem.
  4. Wall wart that comes with the PI(generates 5.1V), using a breakout board from the micro USB connector (this wall wart always works) and connecting VBus to 5V: the same problem.
  5. 300W ATX power supply with ATX connector breakout board: the same problem happens.

(This could be what Pi crashing when powered through 5V pin is about).

Warning: I know that raising the voltage beyond the +5 (5% tolerance) may cause damage.

So I started raising the voltage on the SPS (you can do some minor adjustment) and the power bench. When you get to about 5.5V, you can get the PI to boot. However, the red LED will not stay on reliably. If it's on, and you SSH to the PI, it will go off, then on. I cannot get it to stay on.

So I decided to see what does that voltage needs to be to keep the red LED on (just as if powered by the wall wart plugged in the micro-USB connector). I had to raise the voltage to 6.6V. No magic blue smoke, no chips overheating. (I was OK in sacrificing a PI). I would not be comfortable running a PI with this voltage unless someone can tell me that there is an expected voltage drop in the circuit (I have not seen one).

So I decided to see what happens during boot with the current. When powering with 5.5, the red LED starts on but goes off as soon as the current demand of the pi crosses 500mA. And goes back on when it drops under again. Doing SSH creates more activity, current goes above 500mA, the red LED goes off, and then on when nothing happens on the PI and current draw goes back down to 450mA or so.

Additional information: I am not powering the rPIs directly to the 5V pins, I am powering through a breadboard and a PI cobbler (I have the Adafruit and the SparkFun models).

So my question is: who has powered a PI with a 5V supply on the 5V pins and gotten the red LED to behave and the PI to stay on?

PS Sorry for the long post but I would rather supply all the info I have upfront.

Stephane Bourque
  • 107
  • 8
  • You are doing something wrong. Without a clear photo it's impossible to say what. Let's have a photo of the wall wart connected up. – joan Mar 10 '20 at 20:54
  • I'm powering the RPi 4 via the pins, and it works fine. What you describe sounds like a mix of wiring problems and measurement problems (you absolutely positively can't measure 5.4V anywhere on the Pi if you feed 5.1V in!) which I believe cannot be reasonably reproduced. If you can't get it to work with your power supplies, get the official one. – Dmitry Grigoryev Mar 11 '20 at 10:36
  • @DmitryGrigoryev you are right. My mistake. 5.45 came from when I had another PSU. – Stephane Bourque Mar 12 '20 at 23:46

2 Answers2

0

I learned this the hard way with a 3B. Powered it with a "5V" power supply, and the 3B's CPU literally desoldered itself from the board. Why did that happen?
I later discovered that the power supply was actually 5.2V with no load, but that was enough to create a short circuit inside the silicon.

I bought a 10A 5v PSU for a couple 4Bs to power them through the GPIO port, and initially everything worked fine. For months. I made sure to adjust the voltage to exactly 5.00V, using a multimeter. But over time, the cheap PSU's voltage regulation failed. In 6 months, it was 7.4V!
Why those Pi4s didn't fail, I'll never know.

Now I use a setup like this:

12v power supply --> 5v buck converter with voltage display --> Pi4's USB-C

The buck converter steps down the voltage and displays the voltage being fed into the Pi. Mine looks like this: enter image description here

Botspot
  • 1,759
  • 6
  • 27
0

The answer is ohm's law...

It appears that the pi cobblers provide enough resistance to create a sizable voltage drop when powering through the cobbler. Resistance is anywhere between .7 - 1 ohm. When the pi is booting, that's not a problem (100mA), but as Linux is loading, draw gets to 500-700 mA, which gets you a drop anywhere from .4-.7 volt. This takes a 5V supply below the required voltage and the pi reboots.

Moral of the story, do not power through a cobbler...

Thanks everyone.

Stephane Bourque
  • 107
  • 8
  • I have no idea what a "cobbler" is, but if you refer to the GPIO connector pins, such connectors are rated for a contact resistance of 0.02 Ohm. I power my Pi using those, and it works fine even if I plug 3 HDDs in (I have USB current limiter disabled) which gets the current close to 3A. – Dmitry Grigoryev Mar 13 '20 at 09:06
  • The cobbler is a combo PCB-Flat cable that allows you to bring all the expansion port on a breadboard. So you have pins->pcb->pinds->flat cable->expansion port. It's sold by folks at SparkFun and Adafruit (possibly others)to let develop hardware on a breadboard. Here is a link https://www.adafruit.com/product/2028. – Stephane Bourque Mar 14 '20 at 01:55