Power over USB-C - it's a mystery

[Peabody]

I have a new Thinkpad laptop, with power from the AC adapter delivered via a USB-C port.  I bought what I hoped would be a short "port saver" USB-C extension cable:

https://www.ebay.com/itm/193067529050

Everyday I move the laptop from one room to another, then back again, and I have AC adapters in both rooms so I can stay plugged in without having to move the power supply.  So I wanted to find a short extension that would stay permanently connected to the computer, thus saving wear and tear on the port.

The cable arrived, but it doesn't work with the laptop.  I've tried all possible permutations of both power supplies, both laptop USB-C ports, and even connector orientation, and the thing just doesn't work.  No power is getting through.

However, it works fine with my cell phone and either a standard 5V USB wall wart or the fancy Motorola TurboPower adapter.

So basically, there's nothing wrong with the Thinkpad USB-C ports, there's nothing wrong with the Thinkpad AC adapters, and there's nothing wrong with the extension cable.  Right.

I was hoping the extension would just have straight passthroughs of all pins, but apparently not.  Can anyone explain what the problem might be, or where I could get a USB-C port saver that would actually work with my laptop and its AC adapters?

[Mr. Scram]

USB C has "smart" and "dumb" charging modes and it sounds like your laptop is trying to use the former and failing due to the port saver. The Raspberry Pi 4 had similar issues.

[ve7xen]

USB-PD is pretty complicated, and without PD successfully negotiating the power requirements of the your laptop, it will degrade to the non-PD power spec, which is 5V @ 1A IIRC. Your laptop can't do much with that, but your phone can (and might use non-PD mechanisms to pull up to 2A @ 5V). So my guess here is that your 'port saver' is breaking the PD negotiation somehow. There are quite a few things that could be going wrong here:

- Your laptop requires > 15W charging (or I believe any of the alternate voltages), and the cable assembly is not an 'EMCA' cable with active marking to indicate it is safe at that power, so is refusing to charge. High current charging requires an appropriate cable.
- I don't think extension cables have ever been spec-compliant, so either by design (PD cable negotiation detects this or the cable is somehow marked as an extension) or accident (CC lines aren't connected) it may be causing PD negotiation to fail
- Your cable may be defective; if the CC lines are broken, the phone would still charge in 5V@1A mode, but PD would fail

My Pixel indicates that PD is successful by saying 'Charging Rapidly' instead of 'Charging'. I'm not sure if your phone does the same, but if you see this it probably means that PD is working, and the cable is just not marked for high current use.

There do seem to be some USB-C female to male cables that claim to support 20V@3A charging, so it does seem like they're out there, but USB-PD has greatly complicated the purchase of charging cables, and vendors are not marking them very well (the apparent lack of standardized markings doesn't help).

In my experience USB-C has been quite robust, and that was one of its primary design goals, so personally I don't feel that this sort of hack is necessary.

Edit: Here is one example that at least claims to support 20V@3A charging: https://www.amazon.ca/Extension-UseBean-Aluminum-Extender-Nintendo/dp/B083DG8YYR/ref=sr_1_32?keywords=usb-c+female+male+cable&qid=1583449045&sr=8-32

[thm_w]

Edit: Here is one example that at least claims to support 20V@3A charging: https://www.amazon.ca/Extension-UseBean-Aluminum-Extender-Nintendo/dp/B083DG8YYR/ref=sr_1_32?keywords=usb-c+female+male+cable&qid=1583449045&sr=8-32

Good advice, basically OP bought a non-PD rated cable, so it will not be able to supply PD power levels. The wires inside that cable are likely hair thin as well, so if you could even force it that would not be a good idea.

That link states 10gbps, which means all data pairs are present. Although, it makes no mention of PD compliance, and apparently USB spec for cables does not specify type C female. So it may or may not work?

[Peabody]

Thanks for the responses.  I think I understand now.  I just bought the wrong cable.

[coppice]

USB-C is a mess, but the biggest part of that mess is the labelling of cables. You don't get a clear indication of what the cable does and does not do in its product description, and there is no marking on most cables as to what they offer. If you have multiple USB-C cables, for different applications, you'd better tag them.

[angrybird]

You don't need an emarked cable unless you're >60W which is >3A in 20V mode.  Your laptop will be charging in 20V/3A mode if PD2.0 fixed voltage mode, or at a variable voltage (15-21V if memory serves me correctly) at up to 3A if the charger and laptop are PPS capable.

For PD you need the USB 5V, GND, D+, D-, CC1, and CC2 wires populated on both ends of the cable, this should be most if not all of the Type C cables you encounter though I believe some cheap ones only populate one of the CC wires.

There is one more problem - Common mode range of the Type C/PD circuitry.  If I recall correctly, you cannot have more than 250mV of ground offset between the charger and the laptop.  More than this and you will have common mode issues with the CC pin communication on one end or the other, or both.  So the longer you want to go, the better cable you will need as your ground side impedance of the cable needs to stay pretty low.

[amyk]

USB-C is a mess, but the biggest part of that mess is the labelling of cables. You don't get a clear indication of what the cable does and does not do in its product description, and there is no marking on most cables as to what they offer. If you have multiple USB-C cables, for different applications, you'd better tag them.

I haven't looked at USB-C closely if at all, but aren't the cables just pin-to-pin connections? Or is that too "old fashioned" now?

[angrybird]

There are many failure modes not considered by the Type C connector workgroup and I think his concern about connector failure is quite valid.  It would be nice to have a USB port that has a replaceable "connector saver" like you see on some RF equipment...  I love these things.  It removes fear of costly damage

[angrybird]

My VNA's are up on a shelf, do you have a portable?  I'd like to get something a little lighter, but due to VNA cost I can't fully justify that until what I have is either insufficient for a job, or the wife pre-approves it 

[ve7xen]

USB-C is a mess, but the biggest part of that mess is the labelling of cables. You don't get a clear indication of what the cable does and does not do in its product description, and there is no marking on most cables as to what they offer. If you have multiple USB-C cables, for different applications, you'd better tag them.

I haven't looked at USB-C closely if at all, but aren't the cables just pin-to-pin connections? Or is that too "old fashioned" now?

Mostly yes, but e-marked cables are required for high current charging, which do involve some negotiation with the cable. I was lazy and found the 15W number on Google, but the actual spec calls out 3A rather than the wattage, which makes more sense.

Problem is that the CC lines and the high speed data lines aren't 'required' for USB2 to function, or charging at low current (cell phones), so plenty of cheap cables don't bother with them. USB3 data rates require the high speed data lines, but not CC either. I assume it says somewhere in the USB-C spec that the CC lines should always be connected, but when people expect a cable to cost $2 and the connector it plugs in to can also handle 10Gbps and simultaneously 100W power transfer, and something's going to give. USB-IF should have come up with a marking strategy to help here.

[GromBeestje]

Marking strategy? In my opinion, the whole "one connector for everything" is the problem. People expect, when the connector fits, it works. Telling them to look at the markings, it makes things too complicated.

[angrybird]

Table 3-1 shows the compliant cable assemblies.  For laptop charging, you need #CC2-3, emarker is not required.  I am not sure if there are laptops charging at 100W yet but if there aren't, it is likely there will be some soon.  For those you will need a cable with the emarker that supports 100W.  Also note in the table, any cable for USB 3.2/USB4 requires an emarker regardless of power requirement!

[Peabody]

My recommendation: just plug right into the port. USB-C ports are rated 5000 cycles, and assuming you plug it twice a day, that's 2500 days, or 7 years.

I bet you will have a new laptop in less than 5 years, so why bother?

Nah.  In 5 years it will be barely broken in.  The Windows 7 computer this one replaced is 10 years old, and I would still be using it if Microsoft still provided security updates. Of course the old computer used a barrel jack for power.  To bad this one doesn't.

[Mr. Scram]

Nah.  In 5 years it will be barely broken in.  The Windows 7 computer this one replaced is 10 years old, and I would still be using it if Microsoft still provided security updates. Of course the old computer used a barrel jack for power.  To bad this one doesn't.

What's the predicted lifespan of that barrel jack?

[tooki]

USB-C is a mess, but the biggest part of that mess is the labelling of cables. You don't get a clear indication of what the cable does and does not do in its product description, and there is no marking on most cables as to what they offer. If you have multiple USB-C cables, for different applications, you'd better tag them.

I haven't looked at USB-C closely if at all, but aren't the cables just pin-to-pin connections? Or is that too "old fashioned" now?

Not even close. Not only do USB-C cables need to identify themselves via a chip in order to allow high power charging, but Thunderbolt 3 piggybacks on the USB-C ports, and thus use USB-C plugs, but a Thunderbolt cable is not interchangeable with a USB-C cable (and vice versa). Oh, and a USB-C cable can be USB 3, or just USB 2.  It’s a hot mess.

[angrybird]

Nah.  In 5 years it will be barely broken in.  The Windows 7 computer this one replaced is 10 years old, and I would still be using it if Microsoft still provided security updates. Of course the old computer used a barrel jack for power.  To bad this one doesn't.

What's the predicted lifespan of that barrel jack?

I have a thinkpad I purchased new in 2008 and the barrel jack still works like new.  I've taken this laptop all the way around the world 30-40 times, it used to be my only travel PC and I still use it daily.

USB-C is a mess, but the biggest part of that mess is the labelling of cables. You don't get a clear indication of what the cable does and does not do in its product description, and there is no marking on most cables as to what they offer. If you have multiple USB-C cables, for different applications, you'd better tag them.

I haven't looked at USB-C closely if at all, but aren't the cables just pin-to-pin connections? Or is that too "old fashioned" now?

Not even close. Not only do USB-C cables need to identify themselves via a chip in order to allow high power charging, but Thunderbolt 3 piggybacks on the USB-C ports, and thus use USB-C plugs, but a Thunderbolt cable is not interchangeable with a USB-C cable (and vice versa). Oh, and a USB-C cable can be USB 3, or just USB 2.  It’s a hot mess.

Not all cables need to identify themselves.  It's a lot more straightforward than it was a couple years ago.  Emarkers are required for anything USB3 or USB4 but not required for USB2 + PD cables that are for <= 60W

[steenerson]

For PD you need the USB 5V, GND, D+, D-, CC1, and CC2 wires populated on both ends of the cable, this should be most if not all of the Type C cables you encounter though I believe some cheap ones only populate one of the CC wires.

I thought there was only supposed to be one CC wire, connected to one of the CC pins on each end. This way each end sees one CC pin get pulled up/down and only has to decide if it's own plug is upside down or not. If both wires were connected, the devices would have to coordinate to determine if there's a twist.

[angrybird]

For PD you need the USB 5V, GND, D+, D-, CC1, and CC2 wires populated on both ends of the cable, this should be most if not all of the Type C cables you encounter though I believe some cheap ones only populate one of the CC wires.

I thought there was only supposed to be one CC wire, connected to one of the CC pins on each end. This way each end sees one CC pin get pulled up/down and only has to decide if it's own plug is upside down or not. If both wires were connected, the devices would have to coordinate to determine if there's a twist.

You are correct!  It is required for orientation detection!  My statement was backwards, it applies to the connector, not the cable!  Thank you 

[tooki]

USB-C is a mess, but the biggest part of that mess is the labelling of cables. You don't get a clear indication of what the cable does and does not do in its product description, and there is no marking on most cables as to what they offer. If you have multiple USB-C cables, for different applications, you'd better tag them.

I haven't looked at USB-C closely if at all, but aren't the cables just pin-to-pin connections? Or is that too "old fashioned" now?

Not even close. Not only do USB-C cables need to identify themselves via a chip in order to allow high power charging, but Thunderbolt 3 piggybacks on the USB-C ports, and thus use USB-C plugs, but a Thunderbolt cable is not interchangeable with a USB-C cable (and vice versa). Oh, and a USB-C cable can be USB 3, or just USB 2.  It’s a hot mess.

Not all cables need to identify themselves.  It's a lot more straightforward than it was a couple years ago.  Emarkers are required for anything USB3 or USB4 but not required for USB2 + PD cables that are for <= 60W

So… some cables require chips, some don't, some cables will support above 60W, some won't, some are USB 2, some are USB 3, some are Thunderbolt (which looks like a USB-C cable, but isn't)… how's this not a hot mess?

[tooki]

So… some cables require chips, some don't, some cables will support above 60W, some won't, some are USB 2, some are USB 3, some are Thunderbolt (which looks like a USB-C cable, but isn't)… how's this not a hot mess?

Simple, with increased quantity and reduced per-product IP cost on EMCA technology, they will be dirt cheap. Soon we will see $15 100W 10Gbps cables sold in BestBuy next to a same $15 dumb USB 2.0 cable.

Huh? Where did I say anything about cost? I said it's a hot mess (i.e. it's a confusing clusterfuck), which it is.

[tooki]

Huh? Where did I say anything about cost? I said it's a hot mess (i.e. it's a confusing clusterfuck), which it is.

I meant if you get the holy cable easily accessible, who cares if a cable does what?

I assume you're saying "if a one-cable-to-rule-them-all can be made cheaply, then the issue becomes moot"?

Well, it can't be made at any price: Thunderbolt cables look just like USB-C cables, but will not work for USB, and USB cables won't work for Thunderbolt devices. You cannot make a cable that is both USB and Thunderbolt. (This may change with USB 4, which is Thunderbolt-based, but then those cables won't work on USB-C devices that aren't USB 4 capable.)


Not to mention that if there's a way to make a cheaper cable, even for some applications, somebody will, as we see today with gadgets that include USB cables containing only power wires, standards be damned.

[RiffRaff]

If you look at the progression of the Type C and PD specs over the past few years you will see that there has been a lot of work put in to reducing the complexity of the cable specs.  You have to remember, you are talking about a Universal Serial Bus here.  If it's universal, it's probably not going to be simple.  Ever seen a Swiss Army Knife? 

[tooki]

Active tbt cables are not usb.

Literally my point: they aren't USB, but they look like it, because they have the same connector on both ends. The only visual difference is the TB logo on a TB cable, which is meaningless to someone who doesn't use TB.

Passive ones (up to 80cm for 40G and 2m for 20G) are usb compatible. Passive cables also can be emca for 100w.

So, again, a bunch of cables that look the same aren't the same.



You are trying to explain to me things that only prove my point: USB-C is a confusing mess. A consumer cannot grab a cable with USB-C connectors on it and be in any way certain what it will and will not be capable of supporting. With older USB, while there is the occasional exception, it's quite predictable overall.

If you look at the progression of the Type C and PD specs over the past few years you will see that there has been a lot of work put in to reducing the complexity of the cable specs.  You have to remember, you are talking about a Universal Serial Bus here.  If it's universal, it's probably not going to be simple.  Ever seen a Swiss Army Knife? 

Well, USB was fairly simple for a long time. Its complexity really ramped up with USB-C, though.