@whitneydesignlabs I am not sure I understand. Are you trying to run a Pi5 inside a robot? Most everyone I know that does that sort of thing has the Pi4 (the 5 has issues)  as a server with ESP's, NANOs, and the odd Pi Zero(w) as the on robot hardware.

However, if you are trying to get a Pi5 working as a mobile, make sure you have the new 5A power supply or better still drop down to a Pi4 that only needs 3A. Sure there are people on the net claiming it can work with less, but at what stability. If the Raspberry Org folks say 5A I would use 5A.

BTW, good idea to study PD, it is a little complicated but USB-C without PD only delivers 3A at 5V. I don't know enough about it to be sure, but maybe you need a certain level of device 'smarts' at both ends of the charging/power circuit to go past 3A.

Insufficient power is probably the most common problem I see with Pi's in general.

@zander Yes, I am actually running two Pi 5s in the robot and a handful of ESP32s, Teensy, Pico etc. (all battery operated starting with a native 36vdc). Dropping back to Pi4s is not really a design change I want to make. Need to process images for machine learning, and also run an onboard LLM. These tasks need the Pi 5's compute and speed.

I have the OEM Pi power supplies for my 110VAC mains. But that doesn't work on the robot unless I add an inverter, which crossed my mind as a way to solve this issue. But not a very efficient approach going from dc to AC mains voltage and then back down to 5vdc..  That is an option, I suppose, but kind of bulky and not very sexy.

Might learn more about USB C protocols to make my own. I was hoping not to have to reinvent the wheel on this. 

@whitneydesignlabs Why not go the 18650 route. That is what almost all mobile Pi's use. Superficially I would think the hover board battery would do, but maybe there is something about it that is causing a problem. We know that 18650's work, the trick is finding non-fake 18650's. I just invested in a good battery charger/discharger and discovered almost all my 18650's were useless. At least now I can order from Amazon then test them and get a free return if needed.

If that isn't something you want to do, I understand so then you need to build a monitor for the Pi5 power to see what is the problem. I suspect you will find you are experiencing brownouts.

@zander I certainly can use 18650s in a custom pack. I have them on hand, and a spot welder for tabs.  I was trying to stick to the 36v packs from the Hoverboards, since I have a lot of them in my collection already, they have built in BMS and I have lots of chargers for them.  If I go the 18650 direction, what is wiring plan?  We only get 3.7v or multiples thereof from the 18650s, so I am unclear what method gets that to 5v and how to wire the USB C connector. Also curious how that would be any different than my test #3 above, using the 5v, 10A dc-dc converter with output wired to USB C breakout?

@whitneydesignlabs I only mentioned the 18650 in case the batteries you are using now have some sort of flaw. The main problem you are likely having is the Pi5 needs too much power. This is just one of the flaws in the new Pi that many have already spoken of. I can't readily find a complete board description, but if there is another 5V input other than the USB-C try that, maybe POE?

Without PD you can only get 3A via USB-C.

@zander I will look into powering the board via the pins on the 40pin header.  It also has a 4 pin PoE connector. I will also explore that idea as well. Thanks!  BTW, this is the car charger I also tried. It claims to do USB C PD. But I still get power issues with it:
https://www.amazon.com/dp/B0C3B51KDJ?ref=ppx_yo2ov_dt_b_product_details&th=1

@whitneydesignlabs There is a bug on the forum, you have to post amazon links via some shortener. Unless the part that plugs into the car has the smarts to communicate with the raspberypi USB chips, then it will only deliver 3A. Keep in mind I know little about this other than what I have read in a google search. 

I know that all SBCs can act erratically if they don't receive enough power. The Pi5 docs emphasize this.

EDIT: The screen grab shows what I am using. I was using the Amazon supplied URL shortener bugt just last week it auto-updated to this one.

Link didn't work. But here is the description of it on Amazon:

120W Car Charger USB C for iPhone 15 Pro, JOYROOM 3 Port Super Fast Car Charger Adapter PD 100W&35W QC 4.0 Car Phone Charger Fast Charging for iPhone 15/14/13 Pro Max/Samsung S23/22/Laptop

Here is a discussion about powering the Pi 5 with the gpio pins, and a configuration change:
https://forums.raspberrypi.com/viewtopic.php?t=358008

Might be an approach for me to try with the 5v output dc-dc converter.

@whitneydesignlabs Get yourself a USB tester so you can see how many amps you are getting. I doubt you are getting more than 15W but the tester will confirm. Here are a couple

https://amz.run/7aFa

and 

@whitneydesignlabs There is some good info there but also some cautions. 5A is a lot of power for small wires, and what are the GPIO pins rated for? Power Delivery as far as I understand is a two way protocol, the consumer (Pi5) has to send a message requesting a certain amount of power and the producer (the cigarette adapter) has to understand that message then respond with an appropriate message. Then and only then can you get more than 3A to flow.

Try a little test, first do a test using the Pi 27W official wall wart. Do whatever you do with the robot and observe the power usage. Now do the same with the other power source.

Test complete. I built a harness to use two 5v gpio pins and two gnd gpio pins. I supplied 5vdc to these pairs from a 10 amp, 36v to 5v, dc-dc converter. (double the pi's rated power requirements)  I also enabled max usb current in the config file. It boots and runs fine with no issue, until... I launch a resource intense LLM. As soon as it hits the CPU hard, I get power warnings.

Going back the the oem wall wart, and it works as expected.

There really must be a solution to power a Rpi 5 from a battery. Any battery, any size, I will make a custom pack if needed.

My robot will feel a bit restricted if it has a mains power cord tethered to it all the time... (JK, my robot does not have feelings, but it makes a pretty convincing attempt!) 🙂

Seems like there must be some off the shelf item, or plans to build something, that can provide full 5amp USB C PD from a native DC voltage. Please tell me my robot is not going to have to lug around an inverter. (eye roll)

@zander I have a similar device incorporated into a USB C cable. Even when Pi5 is on the car charger with USB C PD, the cable's watt meter never shows more than about 6-7watts. I was expecting to see more like 15-25w under heavy use.  So it seems that the PD negotiation is just not negotiating or something.

Sorry for multiple post. I am trying to solve this one issue with about 3 different approaches simultaneously. (my brain works in parallel sometimes) 🙂

@whitneydesignlabs I suspect the majority of devices claiming PD are in fact only claiming PD compatibility, NOT power. In other words they will work ok with PD but will not honour or supply anything beyond 2.4A.