TPS2511 DCP Shorted + 1.2V not detected by a Google Nexus 7 tablet

Hi Larry,

How do you know the tablet is not being detected properly? If the battery in the tablet is nearly fully charged it may drawn under 1A current.

Hi Lee,

This was captured by triggering from Vbus (Vcc). The unit was plugged in and then the enable on the TPS2511 was switched on.

Below is another variation. the response is not identical from trial to trial. The end current was the same but I am not sure who is doing what in the signature.

Below is the Nexus 7 connected to its native charger. In this case the trigger was from Vcc but it was via connecting it to the AC line.

The note should say 800ma and not 80ma.

Simply forcing DM/DP to 1.2V works too. This is how it looked with a different charger. This did yield 800ma but I am not sure of the timing. If you look at Vcc

it seems to imply the current was dropping. I suppose that it could also be cable compensation increasing the voltage under load.

I didn't receive your attachment. You may need to switch to rich text format then use the paper clip icon to attach the file.

Hmm, that was using rich text. I see a place keeper in the auto email but not the picture.

I did not use "paste from Word" button. I just pasted it from the clip board. Maybe this does not work.

I used the Paste from Word function for one of the pictures. Let's see if that works.

Nope. Try saving images to a file and then attaching using paper clip icon.

Hi Lee,

One more try! There should be a Word document attached.

LarryTPS2511 Responses.docx

Not yet. I pinged the Engineer again.

Hi Lee,

Thanks for the update!

Larry

I have found sort of a solution. I ran across an adapter cable that has a USB A connector on one end and fans out to a number of different adapter connectors on the other end. This includes USB mini and micro. This cable has DM and DP shorted in the A connector. It may also contain some other circuitry but so far I have not completely figured it out.

The result is that the device sees a shorted DP/DM with some bias from the start. This is allowing the Nexus device to charge at the correct current. It also does not seem to bother the other devices looking for a DP/DM short. This does not work for a divider type device but presumably it also does not supply the correct connector for that.

So the bottom line is that a solution  might be to supply an adapter cable.

Hi James,

I am not really sure either. The reply was from TI. My take is that there is not any sort of a fix for the TPS2511. However I am sort of hesitant to accept switching to the TPS2513. We have too much field experience with the TPS2511. It works fine in most cases.

So far, the TPS2511 has worked better overall than the dozen or so "finished products" I have tested. Also, I don't really see any sort of 128ms delay that seems relevant to the issue. In addition, it strikes me that reducing a deglitch timer from 128ms to 1ms means there is basically no deglitch timer. Since the TPS2511 looks to be the newer part, I take it as the deglitch timer was increased to 128ms from 1ms. (Other parts state this but for a different reason)

Overall I am quite disappointed by the lack of information in the data sheet. At the same time, it seems quite problematic that a solution such as this would work at all. After-all there is really no standard to test against.

From what I have seen, this tablet does charge at about 440ma without "proper detection" so the fact that it does not charge at the normal rate of 800ma is not the end of the world. Some other solutions, non TI, do not charge at all on some devices. This is a much bigger problem.

I would like to test the TPS2513 but at this point I don't have the budget to go through the whole testing matrix to determine if the TPS2513 has some other issue. Again, being that there are no specifications, it has to be tested empirically.

We are now starting to think in terms of supplying an adapter cable as part of the solution. This stems mainly from the idea that it is likely impossibly to make every possible device work with a single solution.

FWIW

Larry

Hi Larry, Lee,

Thanks for the comments from you.

I have been using TPS2511 for some time. Here are my understandings about the part:

• TPS2511 default mode is divider mode.
• If D+ voltage drop across ~2.1V or D- voltage drop across ~1.6V with 128 deglitch timer, TPS2511 will switch from divider mode to 1.2-V mode and stay at this mode for several seconds.
• TPS2511 may go back to divider mode from 1.2-V mode if D+ voltage rises across ~2.1V or D- voltage rises across ~1.6V with 128 deglitch timer.
• If TPS2511 stay at 1.2-V mode for some time, the 1.2V reference will be disconnected and go to BC1.2 DCP mode.
• TPS2511 will reset to divider mode if D+(D+/D- are shorted) drop across 0.375V for a second deglitch timer.

You can see all this is aligned with BC1.2 specifications. The Data Contact Detect timeout is 500ms. So 128ms should be fine.

I also saw many reports about Nexus 7 charging too slow.

productforums.google.com/d/topic/nexus/wrTqHYhS-Vw

www.reddit.com/.../

www.technobezz.com/.../

However, I am not convinced from the TI designer about the 25ms pull-high. From Larry's waveforms, TPS2511 did change from divider mode to 1.2-V mode.

However, it might make sense to reduce the 128ms timer to a shorter time (e.g. 10ms). I saw in this E2E community about 128ms failed the test about UL certification (I will attach the link if I see it again)

Any comment?

Thanks for the additional input James.

I have been using the devices native charger as an example of what it should draw. In this case it is about 800ma. I am using a later version of the OS. This seems reasonable but the charger is rated at 2A. So basically I don't think there is an issue with the tablet in this regard.

It does look like the TPS2511 is "late" to respond to the initial detection try at about 30ms. It does not seem to ignore it but it does not change to shorted with 1.2V until over 150ms later. This looks to be too late and is simply missed by the Nexus 7. Then again, maybe the TPS2511 is not responding at all and the voltage changes seen are entirely due to the Nexus 7 continuing it's connection sequence. It does appear that it eventually thinks it's made a data connection given that D+/D- are driven to 3.3V

The native charger gets to shorted @ 1.2V in less than 100ms from application of Vcc.  There is a "glitch" to 3.5V that looks like some sort of a response. The data lines still wind up at 3.3V. Apparently the TPS2513 is faster and would act more like this. I should probably get one an try it. The very short deglitch timer does make me a little concerned about noise issues.

If I learn anything new I will add it to the thread.

Larry

I can't fully explain what is going on with this device. Clearly the device is not following BC1.2 as it seems to incorrectly go into SDP mode instead of DCP or CDP after performing primary detection. The TPS2513 tries to work around this by switching to 1.2V short mode after 1ms deglitch which causes the device to interpret primary detection in such a way that it does go into DCP or CDP mode.

Hi Lee,

Thanks Larry and Lee for the great comments.

Larry said TPS2511 was a newer part. But I read from datasheet that TPS2513/TPS2514 was a newer part. Can you please provide the comparison between 1ms Vs. 128ms timer? What is the reason to choose 128ms timer or 1ms timer? Thanks in advance.

Hi James,

Sorry about that. I was confused.

Larry

Actually the TPS2546 has a different algorithm. It can switch VBUS on and off, which is what it does when it switches modes. How and when is considered proprietary information.

Yes, switching Vbus was another interesting point. All the ins and outs of divider switching is yet another subject.