EU Regulation 2025/2052: All power adapters 25-100W must be USB PD by Dec 2028

[tszaboo]

I suspect it will be even worse. As there is also the requirement to buy stuff without charger (eg: laptops) that many things simply will not include a power adapter. Instead, adding an extra 29.99 line item for some extra greed.

Or just buy a third party charger from Amazon which costs £10 instead.  This is the key benefit: you do not have to take the adapter that came with the product.

IKEA Sjöss
4 EUR
20W charger:
5.0V 3.0A 15.0W
9.0V 2.22A 20.0W
12.0V 1.67A 20.0W
15.0V 1.33A 20.0W
GAN FET, 88% efficiency

[tom66]

I suspect it will be even worse. As there is also the requirement to buy stuff without charger (eg: laptops) that many things simply will not include a power adapter. Instead, adding an extra 29.99 line item for some extra greed.

Or just buy a third party charger from Amazon which costs £10 instead.  This is the key benefit: you do not have to take the adapter that came with the product.

IKEA Sjöss
4 EUR
20W charger:
5.0V 3.0A 15.0W
9.0V 2.22A 20.0W
12.0V 1.67A 20.0W
15.0V 1.33A 20.0W
GAN FET, 88% efficiency

Even better.  You can be certain that product will be safe and likely reliable. It will meet EMI requirements. Such is the nature of being a large business with the regulations that come with that.

[spostma]

I hope that the Type-C PD/PPS 1-wire signaling standard one day will become the technical basis
for non-Type-C power supplies wih 3-pin barrels (the kind many brand laptops use).

This would open an elegant migration path to higher voltages (>48V) and currents (>5A) than power supplies Type-C cables and connectors could handle.

[Randy222]

12V 20A, 24V 10A are much more popular than 48V 5A. There are several reasons. 5A may reduce energy loss in the cable, but 48V increases energy loss in the DC-DC converter of the device. The cost could also be higher for the components to withstand 48V, including safety compliance (48V is higher than 36V safe voltage)

Fortunately the mandate is only 25-100W so high power devices could still maintain 12V or 24V

I suspect it will be even worse. As there is also the requirement to buy stuff without charger (eg: laptops) that many things simply will not include a power adapter. Instead, adding an extra 29.99 line item for some extra greed.

The requirement? This is literally Apple from decades ago removing headphone and chargers to cut cost and increase net profit, but now legally enforced. bribelobbying is probably involved

"25-100w" seems a bit foolish if PD is already in the 240w area. Why does the EU mandate say "25-100w" ?
Less amps is really the goal (less I²R, etc. So to do that you raise volts.
But what battery device has a batt that is actually 48v or 60v?
Non-batt devices also don't typically run at 48v or 60v. This means there has to be power wasting conversion going in inside that device, IF (if) the device PD side still allows for the higher volts to obtain lower amps on PSU.

There's no big win here, none.

[default0.0player]

"25-100w" seems a bit foolish if PD is already in the 240w area. Why does the EU mandate say "25-100w" ?
Less amps is really the goal (less I²R, etc. So to do that you raise volts.
But what battery device has a batt that is actually 48v or 60v?
Non-batt devices also don't typically run at 48v or 60v. This means there has to be power wasting conversion going in inside that device, IF (if) the device PD side still allows for the higher volts to obtain lower amps on PSU.

There's no big win here, none.

One example is gaming laptop, they are typically 100W~450W. If the mandate includes 240W, half of the power management of gaming laptops would be redesigned, and that's going to cause problems.

For non-batt devices, you can reduce one DC-DC converter by making the power supply the ideal voltage for the "main" circuit of the device.

For batt devices, take laptops for example, there are three "power paths". Power adapter(VADP) -> battery(VBAT), power adapter -> main(VSYS) and battery -> main. Typically the VSYS is 12V~20V. There are typically two MOSFETs and one DC-DC converter. For AC power, the VADP is directly connected to the VSYS, this ensures max efficiency, the battery's voltage is not constant, DC-DC has to be used there so no room for efficiency improvement. For battery power, the VBAT is directly connected to VSYS, DC-DC converter is only used for charging, switching between VADP and VBAT is done by switching on or off for the respective MOSFET(AFET and BFET). There are many buck converters in a laptop, low powered ones to get +5V and +3.3V for some components and dedicated, high powered for CPU and GPU called voltage regulator module (VRM), these converters have to maintain their own voltages so no room for optimization.

In conclusion, to improve efficiency we should optimize the three main power pathes, VADP -> VSYS, VADP -> VBAT and  VBAT -> VSYS. The correct answer is direct connect(with MOSFET as on/off switch), DC-DC and direct connect. This ensures high efficiency for both AC and BAT operations.

As a result most laptop have a single voltage input, if you force them to support PD things get complicated. To support a wide range of input, you have to use a DC-DC in VADP -> VSYS, which reduces the efficiency significantly. I have a laptop like this that support both DC5525 and PD, it's significantly hotter when using PD due to the energy loss of the DC-DC converter.

However there's a new technology called narrow voltage direct current(charging) (NVDC), which removes the DC port and use USB-C PD exclusively. The VBAT - VSYS is connected in parallel and there's a buck-boost instead of buck converter in VADP -> VBAT. The advantage is that it accept a wide range of the input voltages, ideal for PD standard, the disadvantage is lower efficiency(buck-boost instead of direct connect) and lowered battery lifespan(peak load is supplemented by battery even with adapter connected). This is good for small laptops for portability, no need to carry a dedicated adapter all the time since most modern phone chargers that have PD can charge it.

If you force PD on gaming laptops, to get higher power it would require 48V as the current tops out at 5A, you either make half of the components to withstand 48V or use an additional DC-DC converter on top of the NVDC converter to reduce the voltage from 48V to 19V, this causes 10 - 15% additional energy loss within the computer, which requires more cooling.

"Less amps goal" is for transmission lines that are often thousand kilometers long

[tooki]

As a result most laptop have a single voltage input, if you force them to support PD things get complicated. To support a wide range of input, you have to use a DC-DC in VADP -> VSYS, which reduces the efficiency significantly.

I have a laptop like this that support both DC5525 and PD, it's significantly hotter when using PD due to the energy loss of the DC-DC converter.

Using PD input doesn’t mean you have to support every voltage configuration.

If you force PD on gaming laptops, to get higher power it would require 48V as the current tops out at 5A, you either make half of the components to withstand 48V or use an additional DC-DC converter on top of the NVDC converter to reduce the voltage from 48V to 19V, this causes 10 - 15% additional energy loss within the computer, which requires more cooling.

Half the components of what? The local DC-DC converters? Sounds like a worthwhile modification to me.

"Less amps goal" is for transmission lines that are often thousand kilometers long

Sure, but it’s equally important in low-voltage systems. There was a long discussion on here not too long ago about doing away with 100-240V mains in households and just using 12V DC, since so many of our gadgets use that. But the downsides would be enormous. Copper is expensive enough as it is; the last thing we need is to cause devices to need even thicker wires.

Like… 240W at 12V requires thick wires and beefy connectors. Have you not been following the ongoing problems with 12VHPWR connector on GPUs? They keep melting because current imbalances end up running huge currents through individual pins. It’d be FAR better to redesign around 48V supply and then convert that directly down to the final load voltages.

[Randy222]

One example is gaming laptop, they are typically 100W~450W. If the mandate includes 240W, half of the power management of gaming laptops would be redesigned, and that's going to cause problems.

So if my device consumes 80w, but I wanted to escape the 25-100w mandate, my device can sink 100.5w ?

[NiHaoMike]

Like… 240W at 12V requires thick wires and beefy connectors. Have you not been following the ongoing problems with 12VHPWR connector on GPUs? They keep melting because current imbalances end up running huge currents through individual pins.

The huge design flaw is using a lot more pins than necessary. As such, a lot of the connector area went into insulation between pins. The more sensible way would be to base it on just 2 big pins similar to Anderson Powerpole or XT90.

[Siwastaja]

Like… 240W at 12V requires thick wires and beefy connectors. Have you not been following the ongoing problems with 12VHPWR connector on GPUs? They keep melting because current imbalances end up running huge currents through individual pins.

The huge design flaw is using a lot more pins than necessary. As such, a lot of the connector area went into insulation between pins. The more sensible way would be to base it on just 2 big pins similar to Anderson Powerpole or XT90.

This is tooki's pet story - and yet the problem is only with some GPUs, and not with CPU supplies in the same proportion at all, which share exactly the same design - paralleling 8 Molex pins at 12V for supplying more current. Current sharing is not a failure; it's well-known thing that can be taken into account in proper engineering ways just fine, including derating, test, measurement, and quality control. As has been shown, Anderson Powerpoles and XT90 fail, too, and paralleled connector pins are used all over the different fields of electronic and electric engineering just fine.

In the end, the Vcore supplies around 1V and 100-200A end up being pushed into the chip via hundreds of paralleled tiny connections, too! And it works, and tooki can't do anything about it; it's fundamentally the only possible real choice.

But for this context, it doesn't matter - the end result is the same - default0.0player is wrong, and even for modestly short distances, upping voltage to reduce current makes total sense. CPU power supply voltage was upped from 5V to 12V (local buck regulation either way!) ~20 years ago exactly for this reason. Whether it's a clunky, specialized single high-power connector pin, or a bunch of paralleled mundane lower-cost pins (either have good track record when done right, bad track record when done wrong) - the end results is same: big and expensive, exactly what's not desirable. And the cable copper, and the energy loss in the cable, both cost very real money even with just 2 meters - no need for thousands of kilometers.

Remember that voltage regulation % improves quadratic with increased voltage. 24V is 4x better than 12V.

Very high current, very low voltage works well only within a hard-wired PCB, for the shortest possible distance. That's when you just need it, because the chip needs that voltage.

10A is already a lot of current for a connector, and hoping for a small, cheap, performant connector that can do that reliably, not to talk about 15-20A, is chasing an unicorn. Limiting to around 5A and upping the voltage instead totally changes the game, and the target is using local DC/DC anyway, so all that is needed is a differently tuned design, of roughly the same cost and complexity. Upping the voltage is an almost free lunch.

[Monkeh]

which share exactly the same design

Exactly the same design apart from using a different connector, for a different current density, in a different location with regards to mechanical strain. Almost as if they're not at all the same thing, and one is long proven while the other has been a serious problem since introduction.

[tooki]

Like… 240W at 12V requires thick wires and beefy connectors. Have you not been following the ongoing problems with 12VHPWR connector on GPUs? They keep melting because current imbalances end up running huge currents through individual pins.

The huge design flaw is using a lot more pins than necessary. As such, a lot of the connector area went into insulation between pins. The more sensible way would be to base it on just 2 big pins similar to Anderson Powerpole or XT90.

Oh, I 100% agree.

[Siwastaja]

which share exactly the same design

Exactly the same design apart from using a different connector, for a different current density, in a different location with regards to mechanical strain. Almost as if they're not at all the same thing, and one is long proven while the other has been a serious problem since introduction.

What I meant: share the exact same concept - what I wrote after that, was important: design specifics differ. Both are low-cost connectors, with current capability upped by increasing number of paralleled pins. That's the common part. What destines it to failure vs. success is what differs: connector type used, pin count used, possibly quality control in parts used (I don't know). Seems to work for CPU - it nailed the design, the GPU didn't. The CPU success proves the concept is fine; tooki is repeating ad nauseam the concept of paralleling does not work because he can recite one failure. That's what we are dealing with.

On the other hand, MC4 PV connectors proved us that even with a well capable connector with low current density, relatively low current for its size, and no parallelization / current sharing to deal with at all, manufacturing tolerances due to poor standardization can kill the concept. Actually paralleling those would save the day, since the healthier connection would cover for the worse.

And the fundamental truth in paralleled connections is that the derating formula the engineer will be using is the deciding factor whether the end result is reliability-increasing or -decreasing: whether it's "another connection takes up the slack and saves the day" or "another connection takes up the slack, fails itself, leaves even more slack to others, and the whole thing fails in thermal runaway". So the same mechanism leads to the two totally opposite outcomes!

This twist ends here from my side. You guys know I'm right, and so does everyone else; the issue is one unlucky fixation which needs repeated rectification, but I won't pursue it further. We can go back to the EU regulation part.

[Monkeh]

What I meant: share the exact same concept - what I wrote after that, was important

But ultimately the point is that user-mated connectors are a high risk for current sharing failures.

tooki is repeating ad nauseam the concept of paralleling does not work because he can recite one failure. That's what we are dealing with.

Or he's pointing out that the concept has inherent risks, and you've quite some bee in your bonnet about it for some reason.

I've seen current sharing on connectors fail in many other places. 12VHPWR is just an example of how badly it can fail when it meets consumer applications.

This twist ends here from my side. You guys know I'm right, and so does everyone else.

Ah, there's that planet sized ego. I suspect a similar condition in those who put 12VHPWR out there.

[tooki]

Like… 240W at 12V requires thick wires and beefy connectors. Have you not been following the ongoing problems with 12VHPWR connector on GPUs? They keep melting because current imbalances end up running huge currents through individual pins.

The huge design flaw is using a lot more pins than necessary. As such, a lot of the connector area went into insulation between pins. The more sensible way would be to base it on just 2 big pins similar to Anderson Powerpole or XT90.

This is tooki's pet story - and yet the problem is only with some GPUs, and not with CPU supplies in the same proportion at all, which share exactly the same design - paralleling 8 Molex pins at 12V for supplying more current. Current sharing is not a failure; it's well-known thing that can be taken into account in proper engineering ways just fine, including derating, test, measurement, and quality control. As has been shown, Anderson Powerpoles and XT90 fail, too, and paralleled connector pins are used all over the different fields of electronic and electric engineering just fine.

Huh?!? It’s certainly not a “pet story” of mine, it’s just a well-known, still unresolved example of current-sharing gone wrong, which could easily have been avoided by using a higher voltage rather than sticking with 12V. I have discussed it in a whopping total of three threads before this one — none of which I started — and the last time was over a year ago. You’re just attacking me because you and I don’t agree on the practice of paralleling high-current pins. But you argue me on that point with far more investment than I have. 12VHPWR is simply an example I’m aware of. I don’t actually care about it anywhere close to as much as you apparently think I do, so I resent your insinuation of it being some kind of obsession.

Current sharing works fine if everything is well controlled. But that’s a very big “if” that seems to break down in the real world, and I suppose that’s why connector manufacturers generally recommend against the practice.

As for why the CPUs don’t appear to suffer the same issue: that’s a very good question. Clearly there is a reason, perhaps that they’re different connector types (motherboard connectors use larger contacts), perhaps vibration (more on a GPU? I dunno.), perhaps due to cable management (GPU cables bent at tighter radii?).

Indeed:

which share exactly the same design

Exactly the same design apart from using a different connector, for a different current density, in a different location with regards to mechanical strain. Almost as if they're not at all the same thing, and one is long proven while the other has been a serious problem since introduction.

Yup.

In the end, the Vcore supplies around 1V and 100-200A end up being pushed into the chip via hundreds of paralleled tiny connections, too! And it works, and tooki can't do anything about it; it's fundamentally the only possible real choice.

Hundreds, and they won’t individually be anywhere near their limits. Very different from a small handful of pins pushed to a substantial % of their maximum rating when everything is working properly.

But for this context, it doesn't matter - the end result is the same - default0.0player is wrong, and even for modestly short distances, upping voltage to reduce current makes total sense. CPU power supply voltage was upped from 5V to 12V (local buck regulation either way!) ~20 years ago exactly for this reason. Whether it's a clunky, specialized single high-power connector pin, or a bunch of paralleled mundane lower-cost pins (either have good track record when done right, bad track record when done wrong) - the end results is same: big and expensive, exactly what's not desirable. And the cable copper, and the energy loss in the cable, both cost very real money even with just 2 meters - no need for thousands of kilometers.

Remember that voltage regulation % improves quadratic with increased voltage. 24V is 4x better than 12V.

Very high current, very low voltage works well only within a hard-wired PCB, for the shortest possible distance. That's when you just need it, because the chip needs that voltage.

10A is already a lot of current for a connector, and hoping for a small, cheap, performant connector that can do that reliably, not to talk about 15-20A, is chasing an unicorn. Limiting to around 5A and upping the voltage instead totally changes the game, and the target is using local DC/DC anyway, so all that is needed is a differently tuned design, of roughly the same cost and complexity. Upping the voltage is an almost free lunch.

Exactly.

[tooki]

tooki is repeating ad nauseam the concept of paralleling does not work because he can recite one failure. That's what we are dealing with.

Only in your imagination, because I had the gall to disagree with you in public.

Mentioning something twice, a full year apart, is hardly “ad nauseum”.

This twist ends here from my side. You guys know I'm right, and so does everyone else; the issue is one unlucky fixation which needs repeated rectification, but I won't pursue it further. We can go back to the EU regulation part.

You’re the one with the fixation, bro. And that condescending, unyielding, arrogant attitude isn’t going to convince anyone of anything other than your nasty temperament.



And I never said it can’t work. I have said that it can be problematic, and the fact that a) there are examples of it being problematic, and b) that connector manufacturers advise against the practice, lends credence to the theory that it can be a problematic practice. And therefore, what I’ve said is that it’s probably better to avoid it if you can. And the abundance of designs that do avoid it lend credence to that being a prudent course of action.

If paralleling were inherently advantageous, as you contend, then we’d see a lot more designs that do that. But instead we see the opposite, with single fat connector pins being the more widespread solution.

On the other hand, MC4 PV connectors proved us that even with a well capable connector with low current density, relatively low current for its size, and no parallelization / current sharing to deal with at all, manufacturing tolerances due to poor standardization can kill the concept. Actually paralleling those would save the day, since the healthier connection would cover for the worse.

Or they’d fail sequentially if the tolerances on all of them happened to lean in the same direction.

And the fundamental truth in paralleled connections is that the derating formula the engineer will be using is the deciding factor whether the end result is reliability-increasing or -decreasing: whether it's "another connection takes up the slack and saves the day" or "another connection takes up the slack, fails itself, leaves even more slack to others, and the whole thing fails in thermal runaway". So the same mechanism leads to the two totally opposite outcomes!

Sure. But the context here was USB-C connectors, i.e. small, low-cost, mass-market connectors. Not hundreds of bond wires. Not a handful of bolted-down high-tension power transmission lines. The issues with the 12VHPWR plug are an example of a pitfall with current sharing in small connectors, and thus, it’s wiser to go with a higher voltage if possible.

[default0.0player]

Like… 240W at 12V requires thick wires and beefy connectors. Have you not been following the ongoing problems with 12VHPWR connector on GPUs? They keep melting because current imbalances end up running huge currents through individual pins.

The huge design flaw is using a lot more pins than necessary. As such, a lot of the connector area went into insulation between pins. The more sensible way would be to base it on just 2 big pins similar to Anderson Powerpole or XT90.

This is tooki's pet story - and yet the problem is only with some GPUs, and not with CPU supplies in the same proportion at all, which share exactly the same design - paralleling 8 Molex pins at 12V for supplying more current. Current sharing is not a failure; it's well-known thing that can be taken into account in proper engineering ways just fine, including derating, test, measurement, and quality control. As has been shown, Anderson Powerpoles and XT90 fail, too, and paralleled connector pins are used all over the different fields of electronic and electric engineering just fine.

In the end, the Vcore supplies around 1V and 100-200A end up being pushed into the chip via hundreds of paralleled tiny connections, too! And it works, and tooki can't do anything about it; it's fundamentally the only possible real choice.

But for this context, it doesn't matter - the end result is the same - default0.0player is wrong, and even for modestly short distances, upping voltage to reduce current makes total sense. CPU power supply voltage was upped from 5V to 12V (local buck regulation either way!) ~20 years ago exactly for this reason. Whether it's a clunky, specialized single high-power connector pin, or a bunch of paralleled mundane lower-cost pins (either have good track record when done right, bad track record when done wrong) - the end results is same: big and expensive, exactly what's not desirable. And the cable copper, and the energy loss in the cable, both cost very real money even with just 2 meters - no need for thousands of kilometers.

I don't think 12V should be used for laptop's power input. The PSU of a desktop is internal so a desktop takes 100-240V AC as input, not low voltage DC. The sweet spot for high-performance laptop is around 19-24V. Lower voltage means higher cable losses while higher voltage leads to more DC-DC conversion losses.

Let me put the typical diagram for a quick explanation

The VSYS is the input voltage of the CPU and GPU's power supply where it's converted to Vcore around 1V. Which voltage would you prefer on VSYS? If it's 48V, then go HPB and "make half of the components to withstand 48V" is true, including power supply for low powered ones like the keyboard and screen backlight, 5V usb power, etc. However you can maintain high efficiency by only using the DC-DC converter in the diagram for battery charging.

If the VSYS is not 48V, whether it's 19V or 24V, go NVDC and use the buck converter for both battery charging and powering the laptop, this decreases the efficiency by routing 100% of the input power (not just for battery charging) to the DC-DC converter. The DC-DC's voltage regulation requirement is also going to be much higher, if the voltage drops a little it'll lead to battery discharge even when the load is lower than the adapter's limit.

If Anderson connector is used you can simply go 19V 13.5A or whatever, use HPB and get efficiency and better peak response.

It's also way simpler, just 2 "pins", + and GND, unlike the USB-C that has 24 tiny pins that's more complex and can't handle high current.

There are three voltages you can choose, VADP, VBAT and VSYS. VSYS is the voltage of the input of all power converter downstream, CPU/GPU, 5V usb, 3.3V for SSD, etc. Choosing a high VSYS will reduce the I^2*R losses of the main power rail but the subsequent converters for these components needs to be beefed up with higher cost to withstand the input.
The battery voltage can be chosen by varying the series connected cell count, 2S for low-powered portable laptops and 4S for high powered gaming laptops.
Then the VADP and then the power circuit itself. NVDC is good for PD as you only need to replace the buck shown in the diagram to buck-boost and it'll happily work with 5-48V, all subsequent components don't need to be changed. However, the efficiency is lower because the buck-boost is used 100% of the time with adapter.
IPB on the other hand, is less compatible with PD because the VSYS must be exact the same as VADP if adapter is used, in other words, if the VADP is lower than the battery's minimum voltage, the computer won't draw any power from it. However it's more energy efficient as the buck (or buck-boost) is not used to power the laptop itself with the adapter or the battery. If the adapter is underpowered, the buck converter can work in reverse, boosting VBAT to VSYS. This only happens in less than 10% of the time, in comparison to the NVDC circuit that the buck-boost is used 100% with the adapter

If 48V 5A is so good why hardly anyone actually use it. I think it's better to wait for a DC-DC converter that accept a large voltage range like 5-48V and very high efficiency like >95%

[Randy222]

If 48V 5A is so good why hardly anyone actually use it. I think it's better to wait for a DC-DC converter that accept a large voltage range like 5-48V and very high efficiency like >95%

48vdc is old telco-era spec, good in many ways (like less power wasting PSU's inside every count of device). 240vac PSU to all the equip that had high efficiency dc-dc with 120/240 50/60Hz auto switching PSU, was just easier for the worldwide supply chain, the equip worked anywhere, whereas 48vdc powered racks is not everywhere.

48/5 is just some silly spec for a tiny C port that is new in PD world. The trap is, building a PSU spec ahead of actual device engineering. "spec it and they will use it" is a fail from the start.

[tom66]

Most laptops taking a 48V input will just run off the battery bus instead of the adapter directly.  This is already common practice in designs, as the battery current is independently monitored it's possible to charge the laptop with constant current even if the AC adapter power exceeds the charge rate momentarily.  It also has the advantage of not requiring any failover to the battery in case the AC adapter is momentarily overloaded, which can be one reason AC adapters are oversized.

I run my Lenovo Legion work laptop on a 100W USB-C PD adapter just fine, even though it normally runs on a 240W AC adapter. The laptop works fine like this, with battery charging at 40W and the remainder for the system.  As system load exceeds 60W, the battery charging rate reduces; as system load exceeds 100W, the battery gently discharges. In most cases battery does not discharge for a long period of time since it takes a lot of CPU+GPU load to exceed 100W and thermal throttling is a thing too.  The same will no doubt be commonplace with 48V adapter inputs.

[tszaboo]

I run my Lenovo Legion work laptop on a 100W USB-C PD adapter just fine, even though it normally runs on a 240W AC adapter. The laptop works fine like this, with battery charging at 40W and the remainder for the system.  As system load exceeds 60W, the battery charging rate reduces; as system load exceeds 100W, the battery gently discharges. In most cases battery does not discharge for a long period of time since it takes a lot of CPU+GPU load to exceed 100W and thermal throttling is a thing too.  The same will no doubt be commonplace with 48V adapter inputs.

I was charging mine from a limited 24W, 12V charger because that was at hand. The laptop was charging, but I had a videoconf, that run down the battery while it was plugged in.
It takes almost no effort and no cost to make the charger run from different voltage rails. Especially, battery charger ICs, they cost like 1 USD at small 1K quantities at TI. And the laptops sell for 1000 times more. They shouldn't cheap out on compatibility, but because they do, we have to legislate it.

[default0.0player]

Most laptops taking a 48V input will just run off the battery bus instead of the adapter directly.  This is already common practice in designs, as the battery current is independently monitored it's possible to charge the laptop with constant current even if the AC adapter power exceeds the charge rate momentarily.  It also has the advantage of not requiring any failover to the battery in case the AC adapter is momentarily overloaded, which can be one reason AC adapters are oversized.

I run my Lenovo Legion work laptop on a 100W USB-C PD adapter just fine, even though it normally runs on a 240W AC adapter. The laptop works fine like this, with battery charging at 40W and the remainder for the system.  As system load exceeds 60W, the battery charging rate reduces; as system load exceeds 100W, the battery gently discharges. In most cases battery does not discharge for a long period of time since it takes a lot of CPU+GPU load to exceed 100W and thermal throttling is a thing too.  The same will no doubt be commonplace with 48V adapter inputs.

Because in both cases, the VADP is 20V, you can still connect VADP to VSYS. With the OEM 240W charger, the battery is charging at constant current-constant voltage just like any other li-ion charger. With the 100W PD, the battery circuit works in "reverse boost", the buck converter can momentarily work in reverse, boosting VBAT to VADP.

If you use 48V charger, which VSYS voltage are you going to design? Since the VADP is now 48V, using the same VBAT (12-18V) means the VSYS must be designed to handle a wide range 12-48V for both adapter and battery operations, and all downstream buck converters components for +5V, +3.3V, Vcore, etc, must be able to handle 12V-48V as well. If VSYS is still 20V, you'll need an additional buck converter to reduce the voltage to 48V to 20V, causing additional energy loss

[tom66]

Because in both cases, the VADP is 20V, you can still connect VADP to VSYS. With the OEM 240W charger, the battery is charging at constant current-constant voltage just like any other li-ion charger. With the 100W PD, the battery circuit works in "reverse boost", the buck converter can momentarily work in reverse, boosting VBAT to VADP.

If you use 48V charger, which VSYS voltage are you going to design? Since the VADP is now 48V, using the same VBAT (12-18V) means the VSYS must be designed to handle a wide range 12-48V for both adapter and battery operations, and all downstream buck converters components for +5V, +3.3V, Vcore, etc, must be able to handle 12V-48V as well. If VSYS is still 20V, you'll need an additional buck converter to reduce the voltage to 48V to 20V, causing additional energy loss

Why so complicated?  The input USB-C supply is just a buck or buck-boost converter that can put current on the battery bus.  A current sensor on the path to the battery limits charging to, say, 4 amps at 11.1V.  The USB-C converter can do up to 100W, or even 240W, with maximum current modulated by the battery current sensor (simple control loop - limit battery current to 4A and limit total current to lower of supply and IC rating).   All hardware in the laptop runs off this 11.1V rail.  This means that you can support anything from a 15V adapter to a 48V adapter; if you support a boost converter too, then you can even charge the laptop from 5V USB-C at 15W, it'll be slow, but many times that doesn't matter.

This method means the adapter can be, say, only rated for 25W and still charge the laptop or reduce the rate of discharge whilst in use.  A lot of times that is more than enough e.g. I want to use my laptop on the train, if I plug it into a portable 25W adapter then it might discharge slowly but it will do so slower than if it had no adapter, so it effectively extends the runtime I have even if it doesn't actually charge the laptop.  So I don't need to carry a more bulky 100W adapter around, and when I go for my lunch break the laptop will make up the delta whilst in sleep.

[Randy222]

"PD" is also a consumer PITA.
Take PD 3.1, nice spec right? But consumer needs to be aware that PD 3.1 PSU is a tiered spec, it does not mean 240w is available !

I just bought a new PD 3.1 PSU, $88(can) for just the low end of PD 3.1 (140w). It comes with a "PD 3.1" cable but does not say what level of PD 3.1 it will work with. I also got some extra 3.1 C cables that are 240w capable, those actually say "240w" on cable heads.

For cables, if it's labeled "PD 3.1" does that mean it must support 240w?

[default0.0player]

Why so complicated?  The input USB-C supply is just a buck or buck-boost converter that can put current on the battery bus.  A current sensor on the path to the battery limits charging to, say, 4 amps at 11.1V.  The USB-C converter can do up to 100W, or even 240W, with maximum current modulated by the battery current sensor (simple control loop - limit battery current to 4A and limit total current to lower of supply and IC rating).   All hardware in the laptop runs off this 11.1V rail.  This means that you can support anything from a 15V adapter to a 48V adapter; if you support a boost converter too, then you can even charge the laptop from 5V USB-C at 15W, it'll be slow, but many times that doesn't matter.

This method means the adapter can be, say, only rated for 25W and still charge the laptop or reduce the rate of discharge whilst in use.  A lot of times that is more than enough e.g. I want to use my laptop on the train, if I plug it into a portable 25W adapter then it might discharge slowly but it will do so slower than if it had no adapter, so it effectively extends the runtime I have even if it doesn't actually charge the laptop.  So I don't need to carry a more bulky 100W adapter around, and when I go for my lunch break the laptop will make up the delta whilst in sleep.

In this way, VBAT is the same as VSYS and connected to VSYS in parallel, VADP is connected to a buck-boost to convert to 12V(11.1V). However I'm worried about the battery life, if the control loop is not fast enough, the battery could micro-discharge when the laptop is suddenly loaded, and micro-charge when the laptop is suddenly unloaded, it may happen even with 240W charger

"PD" is also a consumer PITA.
Take PD 3.1, nice spec right? But consumer needs to be aware that PD 3.1 PSU is a tiered spec, it does not mean 240w is available !

I just bought a new PD 3.1 PSU, $88(can) for just the low end of PD 3.1 (140w). It comes with a "PD 3.1" cable but does not say what level of PD 3.1 it will work with. I also got some extra 3.1 C cables that are 240w capable, those actually say "240w" on cable heads.

For cables, if it's labeled "PD 3.1" does that mean it must support 240w?

Search for "EPR" cables instead of "PD3.1", most of them are capable of handling 240W, or just search for "240W"

[tom66]

In this way, VBAT is the same as VSYS and connected to VSYS in parallel, VADP is connected to a buck-boost to convert to 12V(11.1V). However I'm worried about the battery life, if the control loop is not fast enough, the battery could micro-discharge when the laptop is suddenly loaded, and micro-charge when the laptop is suddenly unloaded, it may happen even with 240W charger

It's understood from studies on electric vehicles that regen/accelerate frequently, that micro charge-discharge seems to be beneficial to battery life.  In any case, it's still possible to limit maximum charge percentage, which is the single biggest improvement to battery life you can make in a stationary power application.

https://www.sciencedirect.com/science/article/pii/S2352152X2201338X

[Randy222]

"PD" is also a consumer PITA.
Take PD 3.1, nice spec right? But consumer needs to be aware that PD 3.1 PSU is a tiered spec, it does not mean 240w is available !

I just bought a new PD 3.1 PSU, $88(can) for just the low end of PD 3.1 (140w). It comes with a "PD 3.1" cable but does not say what level of PD 3.1 it will work with. I also got some extra 3.1 C cables that are 240w capable, those actually say "240w" on cable heads.

For cables, if it's labeled "PD 3.1" does that mean it must support 240w?

Search for "EPR" cables instead of "PD3.1", most of them are capable of handling 240W, or just search for "240W"

Ahh, right, the other marketing term, "EPR". Why even look for EPR, why not "TB5" or "Thunderbolt5" ?

But correct me if I am reading it wrong, all PD 2.x is SPR, all PD 3.x is EPR. "EPR" does not mean it must support all three levels, 140/180/240w of PD 3.1. Like what I got, a PD 3.1 140w PSU, it's a PD 3.1 ESR device, but does not support 180/240w.


The issue is like the start of "HD" when it came to marketing. HD vs HD(FHD). My old panny plasma came in a box that said "HD" on it, yet the actual screen canvas was below HD specs, the panel has a scaling engine that downscales an HD input to fit the actual canvas. That marketing was technically a violation of the "HD controlling committee" or whatever it calls itself.

Now we have marketing of USB-C cables, a lot of product just says "PD" of some version, now we have "EPR" attached, much of which most consumers have not a clue what that means.

Would just be better to sunset the older PD specs when a new PD specs comes out, forcing makers of PD crud to not make older spec.


I guess the best cable to get is, TB5/USB-C.