Dogwood, how about some updates on thermals?

Hardware Librem 5
22

I don’t know, I think it has been talked about ad nauseum around here. But Purism is a small fish in a VERY BIG and VERY CROWDED pond. Not a lot of things when it comes to the supply chain, etc are really in their favor.

Also don’t forget that their very business model goes completely against the established industry.

Just based on the sheer complication of such a task, I wish people would stop focusing on how Purism doesn’t behave like Apple or Google in the business world, and just accept that enough of their products have gone out, and the way they go out, and the way they are pre-ordered, is an established thing. I’m not saying to give bad business practices a pass. I’m saying don’t focus on them now. What good is having your bone, if you lost the yard to bury it in?

I hope that one day they can get to the revenue streams that allow them solidify their business ends, but until then, just remember, there isn’t another company out there doing what they are.

7 Likes
23

I regret the wording there, and have edited my post to better reflect the situation. TL;DR: Phone will not be on its way until August.

1 Like
24

This seems relevant, considering the title of this thread:

16 Likes
25

It is :slight_smile: thanks!

26

One question that I couldn’t quite see answered is … time between charges with modem on, WiFi off, screen off?

In my situation, if WiFi has relevance, it is probably because I am at home, in which case charging is always a possibility … whereas if I am out and about, that’s where charging may be difficult or impossible, I probably don’t need WiFi, but the modem is probably on (unless I have HKd it because I am sick of the government tracking me :wink: ).

I would also be interested in seeing how the battery goes with heavy GPS usage. I know that on my current spiPhone, this is horrible. If doing in car navigation, you pretty much have to put the phone on charge in the car. Otherwise the battery percentage is moving about as quickly as the blue spot on the map. LOL.

2 Likes
27

What causes the higher power draw in Dogwood?

5 Likes
28

The 12V fast charging is the big news in this update. Getting a 64% charge in 60 minutes isn’t great, but it certainly beats slow charging.

You can get fast charging by varying the voltage (eg, Snapdragon’s Quick Charge) or you can vary the amperage (eg, Oppo/Vivo/OnePlus/Realme’s VOOC) or do both (eg, Quick Charge 5.0 and SuperVOOC)

It looks like Purism chose the method of varying the voltage, whereas PINE64 chose the method of varying the amperage for its fast charging. From what I understand varying the amperage has advantages in terms of limiting the heat in the wires because it has lower resistance, but varying amperage is harder to do than varying voltage.

I think that the Librem 5 is using 1.5 amps, so its max charging is 12V x 1.5A = 18W, whereas the PinePhone is 5V x 3A = 15W. Interestingly, the bq25890 chip on the Librem5 allows for varying either the voltage or the amperage, so Purism could do either with the Librem 5.

2 Likes
29

I’m far from an expert on this, but isn’t it the opposite? Higher voltage has lower resistance and the lower the resistance the lower the temperature of the cable will be because the electrical transfer in the cable loses less energy. That’s the reason why the electrical grid runs at higher voltages than your household, so it’s more efficient and gets less warm.

4 Likes
30

I am especially excited to hear that they are working on suspend to RAM next. I believe this will be the most important thing. There are large parts of the day when I don’t have access to my phone, and being able to just suspend the phone will be excellent. If they are able to allow the radio (modem) to wake for phone calls, that is extra gravy, but it isn’t crucial to me. I’d consider that a feature. My phone takes calls when I want it to, and not the other way around.

5 Likes
31

Everyone is different though. For me, if Purism is going to have some kind of standby / sleep / suspend-to-RAM then being able to be woken for and by an incoming call is essential.

2 Likes
32

I am convinced and confident that the librem5 system will improve over time.
regards

1 Like
33

Only 10h in air-plane mode and screen off! :sweat_smile:
My LG G4 use only 3% battery in 8h in this mode, that leaves a lot of room for optimization…

2 Likes
34

Your LG g4 isn’t running desktop class software either. I think for many it is hard to truly understand the significance of the L5. We aren’t talking about a watered down mobile OS, we are talking about full up desktop class Linux. This means many things, which will become more apparent as we go.

10 hours with a tiny battery (by comparison to a laptop) is impressive and there is still a lot of room for improvement.

5 Likes
35

The reason why the battery is so bad even when everything is turned off is because suspend is not implemented yet so the CPU will still be running 100% of the time even though it’s locked in your pocket (in a low-power mode however).

Purism still has to implement suspend on the Librem 5, putting the CPU into sleep for shorter periods of time while in your pocket and batch together requests from apps who wants to execute periodically. This will significantly improve standby time and it’s what android and iOS does.

I wouldn’t have my hopes up for any more significant improvements of the on-screen battery life though, I could be wrong though.

2 Likes
36

From Ohm’s Law (V=I*R), you can get higher voltage by either increasing the current (I) or increasing the resistance ( R). So yes, you generally get less heat when you put more current through a high voltage line than a low voltage line.

The issue is that batteries degrade faster when they use high voltages. Therefore, high voltage chargers have to convert down to a voltage that the batteries can take, and you always have an efficiency loss in that conversion which generates heat. For example, in a perfect world, you can take 10V 1A and convert it to 5V 2A, but if we use the bq25890’s best efficiency of 93%, that means you end with 5V 1.86A and 0.14A were lost as heat.

Oppo claimed that its high-amperage VOOC charging generated less heat than Snapdragon’s high-voltage Quick Charge. From Wikipedia:

VOOC uses a higher current than normal USB 2.0 charging; VOOC operates at 5 volts and 4 amperes to match the internal battery voltage, which removes the need of regulating the voltage inside the mobile phone, which would produce additional unwanted heat.[2] The VOOC Flash Charge circuit technology reportedly lowers the temperature of the charging adapter, and creates an interface from the adapter to the phone, which, OPPO claims, improves the speed and safety of the charge.[3]

At this point, the best fast charging varies both the amperage and the voltage, so they are generating heat in multiple ways.

3 Likes
37

To see what’s going on, let’s look at a simplified model. The power supply (the USB wallwart) will “offer” a voltage to whatever load is connected. This is normally 5V. The load, in this case the charger (inside the phone) and the battery, will draw some current I from the power supply. The supply will regulate its output to 5V regardless of the current.

Before you apply Ohm’s Law, you need to consider cable resistance, Rcable:

USBcharging
USBcharging921×437 28.3 KB

The same current flows through the power supply, the cable, and the load. So there’s a voltage drop over the cable and an associated power loss as heat: I * I * Rcable = I²*Rcable.

As you can see, the output voltage of the supply doesn’t matter for the cable loss. But output voltage plays a role for output power (and thus power delivered to the battery). Here, you trade higher voltage for lower current.

So, charging at 5V and high current means more heat in the cables. For cheap cables, only intended to carry data signals, enough current will eventually turn the cable into a (blown) fuse…

Also consider Kirchhoff’s Law: Vsupply - Vcable = Vload. High current in thin wires means lower voltage at the load, quite possibly violating the 5V spec. Not so much a problem for the charge circuit, but processors, memory, etc may suffer. (There’s normally a down-converter for the lower supply voltage of these chips.)

Raising the voltage deals with the thin wires, but there won’t be a happy end if an unsuspecting device is subjected to 18V. So there has to be some signalling to make this safe, and the device has to be built such that it can cope with a wide range of input voltages. It’s likely that charge-circuit efficiency will suffer to some degree.

Conclusion
If the cable can sustain 3A without damage, then you can transfer 15W @ 5V, or 54W @ 18V. The cable loss will be the same in both cases, but @ 18V there will probably be additional heat loss in the charge circuit. However, this loss is nowhere near the 39W power transfer gained.

[This attempt at an explanation disregards supply and device negotiating use of additional wires in the cable, or other some clever trick. I’m guessing the quick-charge standards employ every means they can, a bit like Power-Over-Ethernet.]

5 Likes
38

Hi lipu, please post some videos, pictures, news, even some guessing about Dogwood would be great :slight_smile:

1 Like
39

Thank you for the post. I think suspend to ram also sounds really interesting.

What I read here some times is the possibility to wake up upon phone call from the modem. Will this also cover matrix calls via wifi?

Via purism I got to know matrix when they launched the librem5 campaign. Since then my phone is mainly already where we want to go with the phone, a matrix driven phone, using only wifi or data to chat and call via matrix and no regular phone calls anymore. I guess as matrix is not using the standard calls from the modem but data this would not wake up the device, so I would have the same situation as on android (without gapps) which has no push for matrix. Any other speculations?

40

I don’t know what you mean by “desktop class software”, it doesn’t run pure GNU/Linux.
If you use your L5 as desktop mode, you don’t care about battery.
But it’s first a smartphone, and without a good autonomy, a smartphone has much less interest.

That has to be the priority, especially with so much delay, one would have hoped that his kind of basic thing to be fixed since the devkit was released.

What does NXP do anyway? Shouldn’t they be implementing these kinds of functions?

1 Like
41

How is it not pure GNU Linux?

5 Likes