@zander Could be the case. I paid about $30 for this phone charger.  Which is a fair chunk of change if it is faking it.  I avoided the cheapos, and got this one due to being all aluminum and claiming 120W. Seems to charge my phone pretty fast, but I am obviously blind to the details when connected to my phone. But as noted above, when connected to the Pi5, it doesn't seem to do the job as expected.  My robot is built (being built) largely from repurposed parts and scrap yard finds, so I can't afford to throw $30 bucks at every phone charger on Amazon testing them out one by one.

I am sure a solution will present itself sooner or later. I can't be the only person trying to run a Pi5 at full throttle from a battery.

If I can't get a pi5 running on battery, and have to include 120vac mains power, I might as well move to a small form factor PC, which would out perform the Pi5 in AI and machine learning anyway.  But that solution would put a serious hit on battery life.  The robot would be very electron thirsty.

@whitneydesignlabs Why are you blind to what is happening when charging the phone? Just put a USB meter in series. It's no surprise that it works with the phone, they invented it I think.

You should really attack this problem from the other way. How does the Pi5 work. Using the official Pi5 wall wart, insert the USB power meter and then run some task that will push the Pi5 into full 5A power use. If it works as it should, then you know for sure the Pi5 is able to request more power (i.e. 50% of the PD protocol)

I would NOT believe that any other device is really doing the PD protocol and not delivering 5A.

You will need to find a battery board that understands PD in order for the battery pack to supply the higher current.

I think it is very likely that it is not possible to do this at least yet. Remember, the Pi5 can request via the PD protocol that the battery pack deliver 5A but if the battery pack does not have the ability to communicate (does not have a PD chip) then you will not get 5A.

Your efforts may be better placed in getting a Pi4CM running optimized code plus using a profiler to discover the hot spots and cool them off.

Good luck.

Sorry, I am not completely blind when connecting a phone to the USB C car charger. I can still see the watt meter built into the cable. Just not any details on whether there is some power negotiation going on.  My phone is currently 100% charged and it only reads a couple watts. I will let my phone run down low today, then re-connect and gather some more wattage data.  But might be a moot point, since I have already determined this device can't seem to power the pi5 properly.  But I guess the extra data point would be good to have.

I appreciate the suggestion, but Pi 4 does just not have the compute to run a large language model locally. The Pi5 w 8BG ram can just barely do it, with small, optimized models.  And it definitely needs the CPU running at full tilt to do so with any usable speed.  (yes, I am definitely pushing the latest budget technology to the max, lol...)

I have started exploring Amazon's  "power banks" offerings for phone charging. Here is the description of one candidate:

INIU Power Bank, 25000mAh 65W USB C Laptop Portable Charger, PD QC Fast Charging 3-Output External Battery...

It is a $66 gamble on if it will work, but I am running out of options. I suppose if I buy it, and it still can't juice up the Pi to 100% power, the power bank can be used for other projects, or phone charging while travelling etc.

I might find it is the exact same issue as the cigarette lighter version, except this has it's own battery.  If I take the plunge and buy it, or a similar one, I will report my findings.

Cheers

@whitneydesignlabs Ok, but when the phone (I am assuming a recent iPhone otherwise, this will not work) is fast charging (below 80% if the relevant setting is set) you don;t need to see details about the protocol, just the fact that the current is more than 2.4A will confirm PD is in operation.

If the Powebank is a good one like ANKER that may work. If it doesn't, you can just return it. I am no longer able to do that test as my ANKER swelled up and I had to throw it out. I do have a non-ANKER powerbank but do not know if it is PD capable. I will let you know.

If even the Pi5 does not have enough compute power, there are a few alternatives but they may all require higher power so maybe the first problem to solve is how to get more than 15W out of a battery pack. I am sure it can be done, but finding the right parts may be a bit of a challenge. Remember, at least in the USA it is legally allowed to 'puff' about your products. You and I may call it lying, but the advertisers have a deal with the US lawmakers that allow it to be called 'puffery'.

Thanks for bouncing all these ideas around with me. Of all the challenges on this project, providing adequate power is not one I had foreseen. I am driving pairs of motors big enough to move a human being, a robot arm that can curl a 5lb dumbbell, but can't run an SBC... 🙂

https://youtube.com/shorts/H-bJ82tZCBs?si=RonTqCdi-q2EadmU

My phone is a cheap Chinese Android, so it probably can't do PD stuff properly.  I'll run it down low, then slap it on the car charger anyway, just for giggles.

You have a point on Amazon returns. I tend not to take advantage of their return policy. But in this case, with an end-consumer-ready device like this, I might make use of it if the power bank doesn't cut the mustard. I will go ahead and order it for testing and evaluation.

I should have it tomorrow... got to give Amazon some kudos for fast delivery, despite all the puffery. 🙂

@whitneydesignlabs I found some of your on-line info.

I bit the bullet and ordered a $129 unit of the brand you mentioned, Anker, model A1336. Seems like it is one of the most up-to-date, large Ah capacity, and on-board display, which will be useful:  It doesn't seem to mention the words PD in the Amazon ad, the ports are labeled IQ. Hope I didn't make a buying mistake and get the wrong protocol.. lol 

Anker Prime Power Bank, 20,000mAh Portable Charger with 200W Output, Smart Digital Display, 2 USB-C and 1 USB-A Port Compatible with iPhone 15/14/13 Series, Samsung, MacBook, Dell, and ...

Here are the details:

• 200W Total Output: Equipped with two high-powered USB-C ports and one USB-A port totaling 200W output, quickly charge two laptops simultaneously at 100W each for maximum efficiency.
• Lightning-Fast Recharge: The 100W rapid recharge via the USB-C port enables the power bank to be fully recharged in 1 hour and 15 minutes.
• Power On the Move: With a compact size of 4.9 × 2.1 × 1.9 inches, the 20,000mAh power bank is designed to fit seamlessly into your bag, making it convenient for travel and ensuring you always have reliable power on the go.
• Get Real-Time Information: Stay informed with the smart digital display that provides real-time information on remaining battery capacity, power input, and power output, giving you complete control and visibility over the power bank.
• What You Get: Anker Prime 20,000mAh Power Bank (200W), 2 ft / 0.6 m USB-C to USB-C charging cable, travel pouch, welcome guide, our worry-free 24-month warranty, and friendly customer service.

I'll report back on test results and/or if I ordered the wrong one and have to exchange it.

@whitneydesignlabs WOW, last time I ordered one that size I paid $26 for 10,000mAh but it was on sale, now that is the regular price. The fact it doesn't say PD but does say QC at the USB port makes me think it is NOT PD. I did a search for PD on Amazon and found the following.

https://amz.run/7b9c

That says it is PD 30W and what you need is 25W for the Pi5. Since it is ANKER and it is iPhone 15 compatible that just reinforces the decision.

Since that is CDN$ your price will be much less. Order it now and send the other back as soon as you get it.

I've done some Googling since my last post. It seems (?) that IQ protocol is claimed to be compatible with PD.  On the link you posted, if you hover over the pic showing the USB C ports, it also has the IQ symbol and not PD. I am not opposed to ordering this one, too. But any thoughts on whether both units (the one I ordered, and the link you sent, which I might still order also) be IQ which can also do PD?

Shutting down the "labs" and experiments for the night. Phone was at 71% running hotspot all day long (few to no clients). The USB C car charger with watt meter cable, shows 8-9watts, when connected to cheapo phone. Highest wattage I have ever seen from the car charger. No where near 27 watts.  I'll report back when the Anker arrives manana, via Amazonia Post.

@whitneydesignlabs What I consider the important part is where ANKER says

with PD 30W Max Output

IQ is a different protocol that I don't care about, it is proprietary, PD is a USB standard

Thanks @zander for all the ideas! Might order the PD specific Anker unit also.  However results are in from the IQ Anker now:
In a nutshell, it works better than anything I have tried so far. So I am logging this issue as SOLVED for now, meanwhile continuing to look for better options.

Boot-up fine, but still with a the usual warning about the power supply not able to supply 5A. (ugh) Warning clears, and no more warnings from here on. Even under heaving CPU load.  Fluctuating draw from 6-12 watts, then back to 2 or 3 watts at rest/idle.

Ran a multi-modal LLM model called Llava, locally, on the Pi5 from Anker powerbank battery for the stress test. It took about 9 minutes to process the image, and about 1 min to display the analysis results of the prompt. (The robot will use online APIs when available, but will be able to fall-back to a local AI system like this, when Internet is unavailable)  Active cooling running the entire time.

I'll post a screen capture for those interested in those results.  I only post this off-topic info in this thread because it was the driving reason to figure out this mobile battery problem.  If there is interest to discuss local LLMs on an SBC, we could start a new thread, or find a relevant topic.  If I get the other USB C PD specific Anker powerbank also, I will test and post results. It would be nice to compare the two and use the best.

@whitneydesignlabs If the max watts you saw was only 12, then that is not even Pi4 level of 3A. I have no interest in LLM, others may.

That is kind of my thought, too. Still not getting full 5A, but at least it is not shutting down and throwing errors non-stop. I ordered the exact model you posted above. Should have it tomorrow. I will test and post results.

So even though Anker says IQ mode is fully compatible with all PD devices, it still seems like that is not entirely accurate... I'll know more tomorrow, but now I have to decide if I keep this $150 Anker for other projects, or return it. I think I have some time with Amazon, so more testing probably in order before I make a decision.

It is a nice unit, but I want my five amps, wah. 🙂 (Said like a spoiled kid in a candy store...)

Side note. I used the Raspberry Pi 5 wall wart to charge the Anker. Yep, kicks out near 27W for charging, but Anker won't spit 27W back out for device load.

@whitneydesignlabs How are you measuring the current flowing from the battery to the Pi5 USB-C connector?