I’ve been using the new 240W charger from Framework and it has been nice in some areas but I’ve noticed that when playing game the power still drains. I was playing Arknights Endfield and checked HWiNFO what the power looked like and I saw it draining about 5-15W. Once my power drained to about 85-90 the performance would drop and I’d get consistent lag spikes. The TDP of the system with the 240W charger is 145W. While the battery is full and being drained from the system uses 100W for the 7700S and 30-45W for the CPU. Once the system stops draining from the battery the dGPU draws an unstable 80-98W and I get those consistent slowdowns. I’m not sure why the system draws from the battery when the 240W charger has plenty to run everything at full and then some. Does anyone else experience this? I also know it’s not a thermal issue because the highest hotspot temp is 93C the CPU and the GPU doesn’t get any higher than about 85C.
I’m running Windows 11 25H2 with the newest driver bundle.
Here on the forums, the performance of changing power targets based on battery state is known as the “battery flipping” problem.
There are a few things you can do:
Use balanced mode instead of high-performance. This seems to only pull approximately what the PSU is capable of (unmodified). The battery can still drain under a high enough load down to 95% and charge back up to 100%, but the performance drop IME has not been noticeable, unlike the very drastic drop when in high-performance mode.
Frame limit your games to something lower than the max you get. This can help draw less power.
Due to measurement error and safety concerns, the 240W power supply is coded to only provide 205W (90%). I’ve found that using ectool to force it to 240W via setting the input current value higher has allowed me to use high-performance without draining the battery (any more than balanced does), and this only pulls about 215W from the PSU instead of the 205W. This method requires running a command every time you plug the device in though. Do note that by doing this, you’re essentially ignoring Framework’s safety preferences. You can read more about battery flipping here (read up from there): [TRACKING] Battery flipping between charging and discharging / Draws from battery even on AC - #625 by TechPriestNhyk
Use ectool to set a larger battery drain threshold, which would effectively allow it to drain down to 20% (or whatever you choose) before charging again, but this will only delay your performance drop, not prevent it.
I see a total system draw of about 205-215W depending on whether I’m using ectool or not. Idk where the extra ~65W goes compared to a simple 100W + 45W TDP calculation, but I’ve measured with a power meter that it’s definitely going somewhere. RAM, SSD, display and gpu memory are certainly some of the contributors.
I got that 145W TDP limit from the EC driver power limits Here as well as the framework control software where it says the TDP limit of 145W and there’s another thing where it says a number that changes while changing out of 183W. When I do have the system maxed put the CPU can only ever pull 45W and the GPU 100W. I really do think know why the laptop can’t pull more even if it has it available.
You would need to measure at the AC outlet to know for sure. It seems internal software measurement can be off, maybe by a lot. And they don’t even seem to account for all uses of power or conversion losses at all.
There is a youtube video with a thorough test of the power adapter itself, giving the conversion losses there. You’d subtract that from the wattage you get at the AC plug.
The conversation loss would only be a tiny amount like around 5%. The 145W TDP is only for the CPU and GPU so it doesn’t account for the rest of the power used by the system.
The main problem is that the 7840hs + 7700S can draw peak powers of 400W+. The peaks can be very short, but something needs to supply that or one gets a voltage droop that causes stability problems.
It takes time to switch on extra power from the battery so it becomes a race to add power from the battery before the voltage droops too much.
Add to that, that the current sensing in the battery charger chip might not be very accurate, there is no voltage sensor for detecting Vdroop or Vripple and you then have a very difficult problem to solve, particularly during high load, such as games.
In summary, its a very difficult problem to solve.
This problem is kind of swept under the carpet on desktop pcs, because even if a desktop gpu is rated at 200W, and cpu at 100W, users quite happily install 750W PSUs into them, thus handling the peaks nicely.
The power drain I see is a constant amount. Also I can see how many watts the CPU and GPU burst to in hwinfo and they don’t exceed the power if the 240W charger
Hwinfo uses measurements that are averaged over time, so will not show very short higher peaks.
I had to write some special ec firmware to make instantaneous measurements that showed higher peaks.
I still don’t get why I’d see a constant power draw while gaming. Gaming on the balance power option brings the TDP down to 120W and I don’t see any power draw while gaming. Idk, I’m not very knowledgeable on this stuff. Just doesn’t make sense that the laptop can’t use the full 240W power from the charger and not drain the battery. Plenty of other laptops with higher power definitely are capable of achieving this.
The laptop pulls more than 200W under load. If you want to see for yourself, you can pickup a power meter like the one called “Pluggable”. The cpu/gpu cores aren’t the only things that need power.
I ordered one of those tiny USB C to C 240W power meters so I’ll be able to see volts, amp, and watts being fed into the port. It was like $9 and it’ll be a fee days before it gets. I’ll keep y’all updated on what I read while under load. I just don’t believe the laptop is consuming more than 240W on the high performance power mode in Windows.
Just as a heads up, it’s probably pulling about 215W, the Laptop throttles the PSU to only provide 205W by default. At least, that’s what mine pulls when I measure it.
Welp as promised I have got some data from the tester although with a scare and a sacrifice. The tester I got was a Hagibis 240W power meter from amazon and it did initially work. I played Arknights Endfield for a bit and looked at the reading and this is what I got.
High performance power option
220w-232W mostly around 232W and never above with 6-8W battery drain.
Balanced power option
190-220W mostly around 205W with sometimes less than 1w of battery drain.
The tester eventually stopped displaying stuff which was strange. I tested it with other devices and chargers and it still didn’t work. I just left it plugged into my phone and forgot about it when I literally saw it blowing smoke while I was holding my phone. It’s safe to say to never buy these types of testers . The data was good but I’m gonna try to get my money back lol. Also I did try to test the power draw using a FurMark2 and CPU burner at the same time and got the same power readings.
With all that said I do find it strange how it never went above 232W. The voltage reading was also always at or around 48V regardless of power draw. I personally wouldn’t be too upset of the battery drained a little on high performance mode, it’s just how it will suddenly drop performance and it never did on the 180W charger.
I may buy a proper PD3.1 240W USB tester in the future but the only one I could find that goes up to 240W at 48V was $100. I do have a USB PD tester already but it only allows for 156W. It’s the it’s the FNIRSI FNB48P if anyone is curious.
This is the one I use. It works up to 240W as advertised, though mine stopped working for a while though. It’s working again now though, so idk what that was about.
@zachary_stout I actually find it really strange that you were able to get 232W out of the PSU, considering framework throttles it to 205W with the electronic controller. I’ve personally never seen a spike higher than 220W, and only by modifying the EC safety limits can I get it to average something higher than 205W (e.g. 215W with BF6). If that reading was good and repeatable though, perhaps your device is an example of the reason for the throttling they’ve put in place.
It wouldn’t make sense to put a power draw limit lower than 240w for the charger if both the laptop and charger are rated for 240w. At least that’s what I think. I do notice a strange behavior with my laptop when I stress my laptop. Sometimes when the dGPU is fully loaded the CPU can’t pull more than 45w which makes sense. Other times the CPU can pull up to the 54w TDP, then the GPU power draw becomes unstable sometimes even getting limited to ~57w. With the cooling modifications I’ve done to my laptop, it really has no thermal throttle limit. In fact hwinfo said the throttle reason was power limits instead of thermal limits.
. The tester I got was a Hagibis 240W power meter from amazon and it did initially work. I played Arknights Endfield for a bit and looked at the reading and this is what I got.