Shared publicly  - 
 
57th #USB #TypeC analysis: +AUKEY Official Amp PD Duo 29W Wall Charger [Model PA-Y7].
tl;dr: AVOID. Interesting idea, poor execution. Will overcurrent itself (and other devices) due to design flaws. Charger has PD2.0 spec violations. USB-A splitter has USB-C 1.2 spec violations. AUKEY could have easily done what they were attempting to in a safe and compliant manner if only they had done things differently.

This sample was provided by​ Adam Baxter​ from my queue of "USB-C Stuff to Analyze". So thank him for the test!

This is a neat idea, but AUKEY made some significant errors in execution that make it unsafe. The splitter bit is particularly unsafe. "Half-clever, but not clever enough". As a result it will overcurrent the AUKEY charger itself if used normally. And the charger has behavioral problems that make me very wary in general.

I'll try my best to describe in this format: (1) the component/feature, (2) how AUKEY messed it up, (3) why it causes problems, and (4) how they could/should have done it.

---------------------------------------------
Item #1: 29W USB-PD charger.
(1) The AUKEY has 5v/3a 9v/3a 14.5v/2a profiles. The D+/D- pins have BC1.2 DCP signaling. It has "sawtooth" IR drop compensation that kicks on after a time-delay (at 5v only).
(2) This violates the Power Rules. A 29W PD2.0 charger needs 5v/3a 9v/3a 15v/1.93a levels -- any others optional/extra. Sawtooth IR drop compensation is sub-par. The time delay is just wacky.
(3) 14.5v is an "Apple/hack" proprietary level. 15v is the "PD2.0 spec" mandatory level. Without the required level, many devices (like 15v laptops) simply won't work.

Technically a 27w (9v/3a) charger wouldn't need 15v, but this is a 29w charger so it must. Plus, I really don't like when a manufacturer uses a hack to support proprietary Apple levels that even Apple is in the process of moving away from to rejoin the USB-PD standard... which they co-wrote.

Some devices detect how much to draw by increasing current until voltage drops past a certain level. ("Ramping"... the Pixel does it.) Sawtooth IR drop compensation is bad enough in screwing with the algorithm. But by adding a time-delay that kicks on at random times it freaks out devices even more. As a result some may charge slowly/not at all.

BC1.2 only on the D+/D- pins is actually "compliant" per the new spec. But I argue using an Autocoding chip (that supports BC1.2 AND other methods) on the Type-C D+/D- lines is more elegant, since legacy adapters will then work with all devices. This isn't ideal, but it is far more practical. Translation: if an Autocoding chip is used, C-to-Lightning adapters will charge the iPad/iPhone at 2.4a. (But that's technically Apple's own fault for breaking spec in the first place, and not making their C-to-Lightning cable properly.)
(4) Leave 14.5v as an "optional" extra level. Use 5v/3a 9v/3a 14.5v/2a 15v/1.93a. Use linear IR drop compensation. (My opinion only:) Put an autocoding chip on the D+/D- lines.

Item #2: USB-PD traffic
(1) AUKEY uses a chipset that spams SRC_CAP messages with HARD RESET commands in-between. It also issues an unstructured VDM at the end when no response is received.
(2) The chipset violates the USB-PD spec by not dropping VBUS voltage to vSafe0v during HARD RESET as specified in [Section 7.3.12 - Source Initiated Hard Reset]. Since no contract was negotiated, no Unstructured VDM should have been issued per [Section 6.4.4.1 - Unstructured VDM].
(3)The hard resets are bad enough on their own (and are signatures of yet another Apple-directed hack). Likewise, the Unstructured VDM is some kind of weird proprietary AUKEY thing, possibly a backdoor command. [I don't know -- Unstructured VDMs by definition are vendor-specific and proprietary.]

By ignoring the spec's required safeties and procedures, it hints there are other unsafe USB-PD behaviors I simply haven't discovered.
(4) AUKEY needs to fix their firmware, plain and simple. Or use a different USB-PD controller chip from someone who knows what they're doing. [I don't know whose chip this is.]

Item #3: Overcurrent behavior
(1) AUKEY uses constant-current limit at 3.5a at all voltage levels. (It does not use constant-current shutdown.)
(2) While 3.5a overload is "acceptable" for 5v/3a and 9v/3a, it is wholly unacceptable for 14.5v/2a. Not having shutdown/reset to vSafe5v on overcurrent condition (at 14.5v) is also very bad.
(3) Not having shutdown means if a transient/"high resistance" short circuit occurs, the charger will force 3.5a of current through until "whatever" is shorting burns up/carbonizes and the voltage goes back up to the last setting.

Keep in mind this happens at 14.5v too, as shown in the graph. It never cuts power. (Would you want to charge your $2500 Macbook Pro with something that behaved like this?)
(4) Overcurrent protection needs to be dynamic per-voltage level. Constant-current shutdown is the gold standard.

(My opinion only:) If an overcurrent situation is detected, the charger should SHUT OFF and require the cable plug to be cycled manually before delivering power again. That manual action forces the user to investigate out what is causing the short, rather than zapping their device with power until it burns.

Item #4: the C-to-A dual splitter
(1) The AUKEY splitter actually has an Apple 2.4a/12w autodetect chip inside it, powered off 5v input from Type-C. The Type-A ports have D+/D- of 2.7v. This means each port advertises Apple 12w at 5v/2.4a.
(2) 5V/4.8A WILL OVERLOAD THE TYPE-C PORT. The Type-C input also does not check for the Rp advertisement voltage on the CC line and limit itself as necessary.
(3) This is bad, plain and simple. 2.7v on D+/D- means "I have 2.4a, come and get it". The charger can only output 5v/3a before overloading. 4.8a > 3a. Do not pass go, do not collect $200.

By not checking the CC line voltage, it will overdraw any compliant Type-C device it is plugged into. This includes Nexus/Pixel phones (which only offer 5v/500-900mA) and some chargers like the Apple USB-C chargers (which only advertise 5v/1.5a with CC resistor signaling due to being noncompliant).

Using this splitter would allow Type-A devices to overdraw the Type-C device with 4.8a!! Those are port-destroying current levels.

(4) This "splitter" idea would have been simple to achieve in a safe/compliant manner if AUKEY only used a USB-PD chip inside the adapter, then used some logic to restrict the current advertisement on the D+/D- lines of the USB Type-A ports (or shut one off entirely) as needed.

To power both USB-A ports at 24w, it would need USB-PD SNK chip negotiating 24w/9v=2.66a  @ 9v (or 1.6a @ 15v, etc) and convert the voltage internally. If the chip couldn't negotiate that much, the firmware could alter the D+/D- signal on the Type-A port to keep it within whatever budget it had. ("Smart" USB-PD or "dumb" resistor signaling, the ARM Cortex M0 chip could handle it.)

If connected to a phone/low-power charger with only 500-900mA (or less, as instructed via USB-PD), since USB "by default" offers 100-500mA, AUKEY would need to physically disable/de-power one (or both) Type-A ports using a FET.

Or to simplify it by a lot -- only use 1.5A BC1.2 DCP signaling on both Type-A ports -- and ONLY turn on the FET if the Rp advertisement from the Type-C input is 3.0A (or 1.69v on CC). (1.5a + 1.5a = 3.0a/5v.)
---------------------------------------------

Again, all of the above is ENTIRELY possible -- but would require a bit more components, logic, and engineering elbow grease. I've attached photos documenting all of the above statements.

Pay close attention to the animation of the splitter overloading the Apple 87W power supply. Note I didn't run the same experiment on my Pixel -- although I did attach a photo showing it shows up as a device. I would have overcurrented it by trying to draw 4.8a out of a 500-900mA port.

All in all, this is a highly flawed USB-PD charger -- and adapter -- and I cannot recommend it, since it doesn't take appropriate precautions to be safe. It may work -- but "you'll shoot your eye out, kid."

#USBC  
21
1
Johnny Cheung's profile photoSean Graham (Telek)'s profile photoLinh Nguyen's profile photoNathan K.'s profile photo
12 comments
 
+Sean Graham Thanks I caught it during my third edit pass. I typically post (so Plus doesn't eat the post :/) then go through over the next couple hours fixing things.
 
+Michael Anderson This is actually a "neat" idea and shows initiative. But +AUKEY Official  has shown they simply don't have the engineering to follow through on safety edge-cases.

If you make something, you have to think of all the "what ifs". If AUKEY had just asked / consulted / done peer-review prior to producing this, +Benson Leung or I could have easily told them how to do it safely.

It literally could have been as simple as moving a couple wires around -- or changing a few lines of code. At worst, adding one or two MOSFETs. The circuits are already there, and the unique packaging and design shows they are putting in some effort. But as is, it is noncompliant and POTENTIALLY hazardous.
 
Hmm. I wonder whether their 12V car chargers have the same flaw
 
+Sean Graham That list is fully up to date. That is where I publish my data before writing it up here.
 
+Linh Nguyen​​ Can you define what you mean by "fatty" charger? There are plenty decent ones out there.
 
+Nathan K. You can ignore my comment, I for some reason missed this was a wall charger. I was trying to say battery 🙂
 
+Linh Nguyen I'm testing the Imuto battery now. Will see how that turns out.
Add a comment...