Shared publicly  - 
 
WHY is it bad for USB-C to use a single resistor on both CC lines?
tl;dr: Because it will make your active/e-marked cables fail to charge. And your device think it is a headphone jack. You could also potentially blow your DAC or charging circuit.

(Note: corrections, comments, or clarifications are welcomed.)

Recently I posted a critical review of a charger that bridged CC pins and used only one Rp resistor. The manufacturer stated it was for "compatibility" with non-compliant devices. Benson explained this is bad, but people may not understand *HOW* serious this matter is. Cue this post.

https://www.chromium.org/chromium-os/cable-and-adapter-tips-and-tricks
https://plus.google.com/+BensonLeung/posts/4xq4EDjXMw8

(You can follow along with the pictures below using a basic understanding of circuits. Or memes.)

In the first image, I show how things "should" work. Two separate Rp resistors. By varying Rp you change the voltage on the CC line, vRd-USB. The charger and device sense this voltage to find each other. The device also learns how much current it is allowed to pull. (This is where Benson's crusade began!)

In the second image, I show what happens when you start cutting corners and "things go bad". Some manufacturers -- saving $0.001 by omitting a resistor ( http://goo.gl/t8C8zq ) -- use a single Rp and bridge the CC pins instead.

This is a violation of Section 4.5.1.2.1: "Initially, a Source exposes independent Rp terminations on its CC1 and CC2 pins, and a Sink exposes independent Rd terminations on its CC1 and CC2 pins"

If using an e-marked/active cable (as will be mandatory soon),  there is an Ra pulldown on the second CC pin inside the cable. With an unsafe charger there are now two paths to ground. You now have something called a "voltage divider" merging both CC pins. The formulas get a bit messy, but you can simplify the circuit. The resulting vRd-USB makes no sense to the device. It may charge slowly, not at all, or behave erratically.

First-rate [e-marked] cables like +StarTech.com's Thunderbolt 3, or Belkin's USB3.1Gen2 line won't even work with this kind of spec-violation!

(Edit: I direct this to all reputable manufacturers. If you are reading this, *please* consider joining Benson Leung in calling out non-compliant USB-C products. Manufacturers violating spec will make *you* and your well-engineered cables look bad. I can guarantee in the future, someone will post "Why isn't my ___ charging at full speed with my $40 name-brand cable? It works fine with my cheap [non e-marked] cable.")

In the third image, I show how a non-compliant charger may even cause an active/e-marked cable to appear as a "Audio Adapter Accessory"...  in other words, a headphone jack! If the partner device is "DRP" capable, it may misinterpret the CC bridging as the presence of Ra on both pins. This is a unique alternate mode that repurposes the USB D+/D- pins to output analog audio signals from the phone's DAC. (There is already one phone on the market that lacks a 3.5mm audio jack, with more coming.)

You may have read about how Benson is loudly warning Qualcomm QC is not OK on USB-C. This is one reason why. QC 3.0 modulates the voltages on... you guessed it... the USB D+/D- lines to change the voltage delivered to the phone. QC had to do this on USB-A since there were only 4 wires, [4 pins, which limits max current]¹. But USB-C has [2 pins/1 wire]¹ dedicated to negotiating voltage, and [at minimum 5 wires/11 pins, which allows carrying extra current]¹. (Apple chargers have D+/D- signaling too, but is merely the charger telling the phone it MAY take extra current. Not instructing the charger to jack up the voltage.)

The possibility exists -- however remote -- that an [bad] QC-over-USB-C charger, with an [bad] bridged CC pin, with a [GOOD] active cable, and [GOOD] phone, will result in the phone being fried.

(Disclaimer: this is a worst-case-scenario, based on my best understanding. I may have erred in a calculation -- darnit Jim I'm a MechE not an EE -- but it is my humble attempt at explaining how this one flaw [bridged CC pins] can blow up bigtime. I'll explore other "bad" scenarios in other posts.)

¹: Thank you to Benson for pointing out the bare minimums! Please see "Table 3-11 USB 2.0 Type-C Standard Cable Assembly Wiring" for details.
https://plus.google.com/u/0/+BensonLeung/posts/HBWyVBmdVzw

²:  I've since learned the "LeTV Le Max 2" is the exact phone that is vulnerable to this "worst case scenario". Now that I have this in mind, I may do a post on how to trigger this behavior.
http://www.androidcentral.com/leeco-le-max-2-review

#USB   #TypeC   #USBC
19
21
Frank Tedesco's profile photoBenson Leung's profile photoNathan K.'s profile photo
13 comments
 
Where would we be without you, Nathan? I don't presume to understand, since Undergrad physics and Electronics left me decades ago. Our mantra is the same regardless of my comprehension. THank you for this.
 
Even if it's left you, not too late to catch it! Please let me know if you can find any errors, omissions, or oversights!😊
 
+Nathan K. Great write up. One small correction, USB-C has one wire to negotiate voltage as well as current (cc) but each receptacle must have two CC pins for symmetry and orientation detection.

Also, most USB 2.0 Type-C cables will have effectively 5 wires. The 15+ wire case is mainly for faster data and alt mode support.
 
Roger, I've gone ahead and added a link to Table 3-11 where that's described!

I inappropriately carried over what I learned from the Google Huawei Nexus 6P OEM C-C cable. For some reason, it has Vconn pins? (Are they NC internally? Or is the 6P OEM cable active? Only explanations that make sense.)

Edit: They are internally NC. Please see [3.5.2 USB Type-C to USB 2.0 Standard-A Cable Assembly]
 
+Nathan K. Huh. That's the first I've read about that (6P cable).

I don't have one of those cables with me, but if I did I'd query it with Twinkie.

We have some experimental firmware for twinkie that talks to the emarker now.
 
+Nathan K. I'll put that on my to do list or ask someone back home to look closer at that cable.
 
+Benson Leung huh... weird... I just used a multimeter and a breakout to measure (GND->CC1/2) on A12->A5 and B12-B5. Only CC (not Vconn) is wired in the OEM cable it seems.

But I got 7.20V on the floating CC pin using the Huawei charger in one orientation?? That can't be right. Might explain the "not charging" reports some users were giving on the Nexus forum.
 
+Benson Leung The Huawei 6P OEM cable does only connect CC, not Vconn. So that's good. However, during testing, I noticed the OEM charger is sending floating voltages of up to 7.5V on CC. (Ref: http://goo.gl/KE5M7x )

This seems to imply CC v_supply is unregulated, or a constant-current source. (Section 4.5.1.2.1 Ip, 4.11.1)

I put a 5.1KR±1% resistor on it and voltage settled to 1.51V and Vbus triggered 5.01V as expected. (Ref: goo.gl/85IX30 )

Edit: Links fixed. But it isn't constant current, see post below.
 
+Nathan K. Ah of course! That clears it up. Constant-current source makes sense, and is allowed, as you can see in Figure 4-6. It's just easier to explain using Rps.

BTW, could you check your links? I get 404s for both of the links in your last comment.
 
Sorry about that, links fixed. It refers to the Nexus 6P OEM charger post album. Unfortunately, additional testing indicates it may not be a constant-current source.

(Ref 22k Rd: http://goo.gl/DSnCwj ) (Ref 3.1 vRp: http://goo.gl/rGn9ac ) This suggests the N6P charger CC voltage source is unregulated. I added a footnote to the Nexus 6P Charger Analysis.

Is unregulated power to CC Rp a bad thing? Or is it permitted in spec?

https://plus.google.com/102612254593917101378/posts/H3UKcWe59jP

"Edit 2: I constructed a simple 22kR network to test the constant-current supply theory. (330uA * 22kR = 7.26V) However, the voltage only reads 3.15V, proving it false. Therefore, the CC supply in the Huawei Nexus 6P OEM charger is likely unregulated."
 
+Benson Leung​​​​​, after speaking with an EE expert friend, and doing some additional tests, I've learned (a few?) USB-C chargers are vulnerable to "driven right leg" on the floating CC pin. When I measured AC voltage (on the CC Rp DC output), if I touch the CC line while ungrounded, I detect a 5V AC transient, with a 8-9V DC signal. (Due to floating ground I understand this is unavoidable.) I will share these images momentarily.

My friend suggested putting a 1MR pulldown on CC to mitigate this type of interference, at the expense of "vampiric power". Is this wise, or accounted for in the USB-C spec? (Some chargers have CC lines succeptible to common mode "EM noise"?)
 
+Nathan K. hmm. I actually don't know. I have consulted with one of my EE colleagues, and he asks #1) which charger exactly is this? The Huawei 6P charger? #2) which multimeter (make/model#) you are using to make the measurement.

We will try to repro here as well.
 
+Benson Leung I compiled all the helpful information I could into this G+ post: https://plus.google.com/102612254593917101378/posts/1K7b8HTN5zh

To answer briefly: (#1) BOTH the supposedly-good USB-C chargers I have: the OEM Huawei Nexus 6P OEM charger, and a iClever 30W dual-port A/C charger. (#2) a cheap multimeter, but highly recommended and the personal favorite of Larry Sears, EECS professor at Case Western Reserve University and founder of Hexagram/ACLARA RF Systems. 

NOT a Fluke by any means, but recommended (and provided to students) as "a decent cheap one". I've attached pictures of the manual.
Add a comment...