High Battery Drain During Suspend (Windows edition)

[Matt_Falcon]

@2disbetter

If you are talking about default Windows installation that is a Microsoft setting

90% certain that’s not true - unless MS has some serious crackhead defaults if it detects Newspeak Standby. I see these initial values (on a new install) change from PC to PC, and I KNOW that 20 minutes isn’t anywhere near a typical value. MS does have defaults if the system doesn’t specify, but that’s up there in the hours territory, not 20 minutes. ha.

(I changed it to 120. I dare say that’s the most-sparse power profile page I’ve seen in my life… I really appreciate that the “processor” settings are finally gone, because most people abuse that by setting min=100/max=100 and burn through so much power unnecessarily)

I’m not “seeing” 20 minutes of sleep before hibernate. I’m seeing “20 minutes” as the “Hibernate after / On battery (minutes)” setting in the Power settings. By default. Only on this PC.

As to the audio device, it appears to be an IDT codec, not a Realtek which is the bundled driver in the Framework package. Thus no, no driver neither by WU nor the bundle.

[2disbetter]

@Matt_Falcon

90% certain that’s not true - unless MS has some serious crackhead defaults if it detects Newspeak Standby. I see these initial values (on a new install) change from PC to PC, and I KNOW that 20 minutes isn’t anywhere near a typical value. MS does have defaults if the system doesn’t specify, but that’s up there in the hours territory, not 20 minutes. ha.

I have 2 Framework laptops and develop software for Windows. By default Windows on the Framework doesn’t even have hibernation enabled. (As in, you can’t select it from the power menu) As being part of the s0 suspend scheme hibernation is set to a ridiculous number that makes no sense.

@Ryan_Porter

The default hibernation setting on my Framework laptop is to wait 596,523 1/3 hours before hibernating. I changed mine to 30 minutes.

How yours was set to 20 minutes is a mystery to me, but I’ll stand by saying that it was not Framework’s doing. One possibility for it is If you are syncing your settings via OneCloud and your MS account, this could be a setting that you setup on a different computer, that was just synced to your Framework laptop.

@Matt_Falcon

As to the audio device, it appears to be an IDT codec, not a Realtek which is the bundled driver in the Framework package. Thus no, no driver neither by WU nor the bundle.

As for this, it should look something like this:

Note: this is from our family computer which is a Dell All-in-One. I’m just using it to indicate what you should be seeing.

Depending on your batch number you either have a Realtek audio chip or a Tempo audio chip detailed here:

Solving for Silicon Shortages

If your device is hibernating because of a setting which tells it to do that, then we don’t have anything here to talk about.

However if you’ve changed that setting, and you are still only seeing 20 minutes of standby time before hibernating, then I would point to your audio drivers being the culprit. There is a good chance the new Tempo chip works fine with window’s generic drivers, and your audio is not the problem at all.

[Matt_Falcon]

@2disbetter

By default Windows on the Framework doesn’t even have hibernation enabled.

Hibernation is a hidden component of sleep, and has been since … what, Windows 7? It’s been 10 years since Hibernate has been a visibly exposed user option. We can assume that, if anyone is referring to Hibernate, they know how to view under the covers and tinker with things. Yes, hibernate is enabled by default on every Windows system. It’s just not available for you to click, by default.

(just because the option to explicitly hibernate is hidden by default, doesn’t mean hibernate defaults to being disabled)

Windows 11 had the same 20-minute default behavior. I was a bit upset.

@2disbetter

As for this, it should look something like this:

You’re showing “jacks” there, not audio devices/codecs.

A “batch number” is something that I keep seeing referred to, but nowhere in the ordering or fulfillment process (about a week and change from ordering to receiving) did anyone tell me what batch I had. In mine, it seems I was affected by this chip change. Great, it’s fine. It works. I’ve got no problems with it so far, but I just thought it’s weird that the driver pack didn’t include an OEM customized driver for audio. Obviously I don’t have a Realtek codec in my model. (according to that blog post, I have a Tempo chip - formerly IDT, which … indeed, lines up with what I said earlier, ha)

Anyway, this discussion is getting very frustrating, as I feel like I’m talking to a brick wall. These concepts – that Windows takes some power settings from configuration in OEM-specific firmware – are well established by now. Windows has hard-coded defaults, but it will yield to settings planted in system firmware that it pulls during setup. The settings of the power plan Framework gives it are not the standard defaults Windows typically has. One of them is to hibernate after a 20 minute timer.

If you don’t believe me for some reason, go set up a new Windows 10 install in VMware Player, see what its power plan defaults are. Then go clean-install a Framework laptop. You will then find that, gasp, the clean Framework install sets a 20-minute hibernate timer, and the clean VMware install didn’t. (default Windows behavior is also to say “low battery” at 20%, reserve at 7% I think, and critical at 3%… but Framework is reserve at 4% and critical at 2%, so there’s that too)

[2disbetter]

@Matt_Falcon

If you don’t believe me for some reason, go set up a new Windows 10 install in VMware Player, see what its power plan defaults are. Then go clean-install a Framework laptop. You will then find that, gasp, the clean Framework install sets a 20-minute hibernate timer, and the clean VMware install didn’t. (default Windows behavior is also to say “low battery” at 20%, reserve at 7% I think, and critical at 3%… but Framework is reserve at 4% and critical at 2%, so there’s that too)

I’m not trying to frustrate you. I am merely saying that I have installed Windows 10 4 times on my own Framework laptop and one time on my wife’s Framework laptop. At no point or time was ‘hibernation after’ set to 20 minutes. It was set to the time I mentioned in the previous post. So how yours is being set to 20 is beyond me.

Regarding the audio inputs, yep! Doh! I just went and checked on my wife’s framework, because I got hers later. I think her’s was batch 5. Her device manager reports that she is using window’s generic drivers. On mine, which is a batch 1, I see that my microphone array and speakers are realtek based, which indicates that the driver is installed. (You can also check under sound, video, and game controllers.)

All of that to say, and to not keep derailing this thread (sorry), audio doesn’t seem to be the issue here at all. Just timers under power settings. Glad that is figured out.

[Matt_Falcon]

Yup, and I sorta suspect that this default timing I’m seeing on mine may be some kind of artifact of build differences, somehow… maybe I need to do more testing to isolate what caused it to get set so ridiculous each time I install Windows. It’s definitely not a good setting. haha.

I try and give defaults a chance, a lot of the time*… but my first out-of-box hours with the Framework quickly found it chewing through battery like crazy. Now on my 4th day with the thing, and I hardly even need to bring a battery pack along for all-day battery life! I forgot my USB-C cable at home today, and I’m hardly even sweating it. (35%? nah i’ll be fiiiiine)

* where I know they’re not toxic - which is the opposite of the default “hide file extensions for known file types” which MS still keeps enabled by default 20 years after “clicking an .exe” is known to be a fatal mistake yet MS still hides extensions by default so a malicious .exe with a .doc icon can still exist

[feesh]

Note to anyone who gets a mainboard replacement from an early batch (and therefore gets switched from Realtek to Tempo DAC): Uninstall Intel SST as it becomes a problem device during standby. After doing so, I’m not seeing any audio device on the top offenders list.

Also, got a chance to test the expansion cards’ contribution to standby drain, just for a few more datapoints:

[michel]

@feesh

Also, got a chance to test the expansion cards’ contribution to standby drain, just for a few more datapoints:

It’s probably the coffee not kicking in yet, but could you elaborate on the 4 different readings?

[feesh]

I let it sleep for 1h with different configurations of expansion cards plugged in. First row is USB-C only, subsequent rows add on non-USB-C cards. You can see the resulting battery drain in the CHANGE RATE column.

[dkter]

My laptop just started doing this the other day (as in, it doesn’t actually sleep when I close the lid or press the sleep button, the fan continues to run and power LED stays solid)

what’s weird is that I have a Tempo codec, and most of the issues in this thread seem to be from the Realtek

[Shiroudan]

I’m also getting a pretty bad case of this it seems. 2.5wH in 35 Minutes def isn’t great.

This seems to be the main offender.

[Jason_Dagless]

I switch off sleep, fast start and hibernation. No power drain at all.

I’m either using the laptop…or I’m not. Fast NVMe means it’s back up in seconds.

I’m not called on that often to hack into some bomb system with 12 seconds to go…

[feesh]

Update for BIOS 3.09 – no apparent differences in standby drain (as expected).

[Avendor]

This is what hangs my laptop up. Haven’t seen any solutions out there for it and it prevents modern standby more often than not.

[Second_Coming]

@feesh

Update for BIOS 3.09 – no apparent differences in standby drain (as expected).

I don’t recall Framework officially responded to the ‘expansion card power drain while in sleep / suspend’ issue here.

Looks like they can’t fix it / improve the situation.

I’m waiting to see if adopters of the 12th gen mainboard also see similar drain behaviour.

[feesh]

Also curious to see how sleep drain is on the 12th mainboards. Here’s one datapoint that shows 5% drain across ~5 hours with 2C, 1A, 1HDMI which looks like a decent improvement.

[Second_Coming]

@feesh

Also curious to see how sleep drain is on the 12th mainboards

I don’t have a 12th gen Framework mainboard…but for a field comparison with other 12th gen laptops, I have the Asus Zenbook Pro 15 Flip OLED…it’s drained 3511mWh in 10h30m in standby (Network connected standby). (It has 2 TB4 ports, 1 USB-A, 1 HDMI, 1 microSD card slot and a headphone jack) So around 0.335W…

[Anthony_Bucci]

High Battery Drain During Suspend (Windows edition)

I don’t have a 12th gen Framework mainboard…but for a field comparison with other 12th gen laptops, I have the Asus Zenbook Pro 15 Flip OLED…it’s drained 3511mWh in 10h30m in standby (Network connected standby). (It has 2 TB4 ports, 1 USB-A, 1 HDMI, 1 microSD card slot and a headphone jack) So around 0.335W…

For those that don’t feel like doing math .33W would drain the framework roughly 3% in 5 hrs

[Tomato]

A LTT video about framework came out today that actually mentioned that they apparently figured it out an a fix is incoming. Take that with a grain of salt but it is something.

Link with timestamp: Should I take back my $225,000? - YouTube

[Second_Coming]

@Tomato

they apparently figured it out an a fix is incoming

Not sure if that means fixed in the 12th gen, only. Or also possible to be fixed for the 11th gen.

[Ray519]

With a 12th gen Framework, 32GB of RAM and a WD SN850, but all USB-C, Windows reports a sleep consumption of ~450mW so far…