Framework 16 Charging Efficiency-60W/100W/140W/240W

Here are some of my findings from testing the FW16 with various USB C chargers I have at my disposal. All tests are on Windows 11 on the balanced power mode using a FW16 with the 7840HS and no 7700S dGPU. Power measurements were made with a WITRN C4 power meter and HWINFO while running an in-game benchmark.

In general, the laptop appears to be louder and run at a slightly higher power with USB C Extended Power Range (EPR) chargers, but benchmark performance sees little improvement. It seems that Framework only uses the maximum voltage available, so even if your battery is fully charged and idling it will still charge at 48V on a 240W charger, 28V on a 140W charger, etc. It appears that power conversion inefficiencies may be adding ~28W of heat into the system while using a 240W charger. That number could increase on a system with the 7700S where over 200W could be requested from the USB C charger.

Battery Power 60W USB C 100W USB C 140W USB C 240W USB C
Peak Package Power 41W 30W 42W 47W 54W
Average Package Power 23W 20W 26W 29W 34W
Peak Charger Power N/A 58W 86W 129W 198W
Average Charger Power N/A 58W 86W 128W 163W
Peak Charge Rate -60W 37W 53W 82W 84W
Average Charge Rate -40W 20W 37W 65W 83W
Laptop Power Conversion Efficiency (Estimate) N/A 93-96% 93-95% 86-89% 83-85%

Laptop power conversion efficiency is (Average Package Power + Baseline Power+Fan adjustment+Average Charging Rate)/Average Charger Power.

Estimated efficiency is not accurate due to small inconsistencies in test runs and a lack of insight into power used outside of charging or powering the CPU package. I estimated the baseline power draw of the laptop at 17W while on battery and used that number +/- a few watts to accommodate fan speed changes to isolate the power conversion loss.

12 Likes

Where did you find a 240W charger to test with? My understanding was that none yet existed (see the discussion here: FW16 Plugged-in Battery Drain - 240W Charger Option - #4 by Snipe)

I have the eval_xdps2222_240w1 evaluation board from Infineon. https://www.infineon.com/cms/en/product/evaluation-boards/eval_xdps2222_240w1/.
Infineon also has a REF_XDPS2222_240W1 reference design that is much smaller and more likely to resemble consumer 240W chargers. It seems like Arrow was involved in the development of the reference design. https://www.arrow.com/en/research-and-events/articles/arrow-and-infineon-240w-usb-pd-reference-design-board

12 Likes

Brilliant find and as @Christian_Denney noted: somewhat unexpected

I’ve applied for the arrow & infineon 240w board. Will see if they’ll reply.

1 Like

Great work.
Have you adjusted for the power supplies efficiency rating, or does the board you’re working with allow you to configure the output wattage? I’m concerned the power supplies engineering may have a large influence over the results rather than the Framework’s.

That’s a USB power meter, so the efficiency of the power supply doesn’t matter for this overview. Would be interesting to see the value for the 180W adapter, though.

I measured the power going into the laptop with a USB c power meter so the power supply efficiency can be disregarded.
Here is an example of my measurement setup:

2 Likes

Oh I see, thanks!

I’m always intrigued that these meters, it doesn’t seem to matter what brand, display the current to so many decimal places. I doubt that they can actually measure to more than two decimal places accurately, knowing the sort of technology that is behind the actual current measuring process.

Can I ask about the idle efficiency of different power inputs please. I we losing much at idle with an overly powerful adaptor?

I ran 2 power tests while idling on the desktop with a fully charged battery. I could only test the 20V (60W) charger and 48V (240W) charger as my meter bricked itself somehow. The 60W charger provided 0.4661WH and 240W charger provided 0.5273WH over the 2 minute test. The 48V charger provided ~13% more energy which matches the overall efficiency change going from a 20V charger to 48V (83%*1.13=93.8%).

The 60W charger showed a little over 11W for most of the time with some occasional jumps that brought the average to around 14W.

The 240W charger showed 14ish watts for most of the time and some jumps that brought the average to almost 16W.

I also ran two 4 minute tests while the laptop was charged and sleeping.
20V (60W) charger provided an average of 5.2W.


48V (240W) charger provided an average of 10.4W or double the 20V charger.

I am not sure why the sleep power is so different between the 2 chargers but I can’t do any more external measurements.

3 Likes

Was this fully loaded? i.e heavy gaming with a low battery. AFAIK Framework Laptops don’t utilize 100% of the charger capacity, my FW13 only draw 85W on a 100W charger, from your picture 198W, 198÷240≈85/100 this value is expected

It would be nice, if you (or someone else with a similar test setup) could add a column with the values from the official 180W charger from Framework for reference…

P.S.: for some unapparent reason the link to the initial posting didn’t show up, even though I hit its reply button.

Yes, the peak power draw was at full load while gaming.

The FW16 also took ~85W with my 100W charger, although I wouldn’t expect every charger to provide a max of 85% of listed power. My 140W charger provided 128W or 91.4% of listed output. Also, the FW16 without DGPU shouldn’t even have a big enough load to exceed 200W so I can’t check the max charging output for my 240W charger.

1 Like

I ran some more tests after finally getting the fw-fanctrl-dev-windows tool working on windows. These new test results are slightly more accurate because the fan speed is constant, however, there are still plenty small errors from timing and measurement.

I found it interesting that the changes in this test caused the battery life to decrease using the 100W charger. Additionally, the 240W charger dropped from 194W to ~140W after about 3 minutes and the benchmark run ended up having slightly lower performance than with the 100W charger. I measured the baseline power draw from everything but the CPU package at 33W this time which is a lot more than I expected compared with the previous test at 17W.

All results are from a 10minute CineBenchR23.2 Multi Core benchmark on Win11 with Best Performance mode, 80%Fan speed and Max Brightness [FW16 7840HS no DGPU]

Battery 100W USB C 240W USB C
R23 Multi Core Score 15005 15629 15556
Average Core Frequency 4.01GHz 4.26GHz 4.18GHz
Peak Charger Power N/A 88W 194W
Average Charger Power N/A 86W 156W
Peak Charge Rate -85W 7W 85W
Average Charge Rate -76W -3W 57W
Peak Package Power 58W 58W 65W
Average Package Power 43W 53W 51W
Power Conversion Efficiency N/A 94.5% 90%
Average Core Temp 77.5°C 93.8°C 91.7°C
5 Likes

How did you get the tool working? I have not been able to successfully install it. I am pretty sure the CrosEC driver is what is not working but don’t really know what else to do from here.

I think the issue that prevented CrosEC from installing on my first few attempts was caused by not restarting my laptop after enabling unsigned drivers. Try enabling unsigned drivers and rebooting before running the install.bat file.

Darn I tried that. I keep getting a “The system cannot find the file specified”. Well thanks for the response.

I hope to get a 240W eval kit as well but sadly I don’t work with USB power much.

I had no issues running the install.bat file by double clicking it. It seems like the error message you listed is somewhat common for batch files. I would suggest searching the web for help.

1 Like