TUSB4041I: 1.1V power supplies keep breaking

I did a search before posting, but looking at related topics to my own post I find this https://e2e.ti.com/support/interface-group/interface/f/interface-forum/891653/tusb8041a-1-1v-core-supply---recommended-power-supply with a mention of up to 670mA @ 1.1V. I have checked if the TPS82676SIPR‎ runs hot using a Thermal camera and it does not. It's just below 30C/86F before glitching out.

Hi Jens:

     The thermal pad mentioned in above link is for the ground pad of hub device, did you have good soldering for the ground pad?

     also the high 1.1v current form TUSB8041 is due to USB3 signal, TUSB4041 should not that high, and 600ma should be enough.

     You should be send schematic by insert button below.

Best

Brian

Hi Brian,

Here is the schematic. The Thermal pad has a good solder connection. 

J

what input voltage from schematic below. to get 1.1v output, input   at least 1.7v to get 1A  current.

Best

Brian

I can see that has fallen out of this schematic? Sorry about that. IN voltage is 5.0V

Did you have any de-coupling caps close to 1.1V and 3.3v supply pins, like 0,1uf cap?

Best

Brian

Sorry. I was way too quick when I uploaded that schematic. As you found, it was not complete. Let me do this better? I've gone through my GIT history and created these from the last version of each of the Power Supply schematics.

This is the schematic for the TPS74611

This is the schematic for the TPS82676 & TPS7A4533

This is the schematic for the NCP139

As you can see from the final version of the TPS74611 files, there is lots of capacitance. Here is a screenshot of the layout:

As you can see, there are two 0.1uF caps (C6 & C13) close, a 22uF close to the input and a large 150uF bulk capacitor (C14). In addition, there are lots of 0.1uF around the TUSB4041 on the 1.1V rail as well.

ok, looks you have enough  decoupling caps around hub.

this may not directly impact, but seems have an issue.

Does hub connected to OTG device though Mux?  High voltage  for SEL pin  should be VCC which is  3.3v, but is also connected to Vbus, which is 5V.

Did you connected UBS devcie thoguh type-C port during the test?

Best

Brian

Hi,

Nothing was connected during the test. The MUX setup is per the official Raspberry Pi CM4 reference board schematic: 
https://datasheets.raspberrypi.com/cm4io/cm4io-datasheet.pdf (Figure 7. USB2 hub)

Hi Jens:

    When Nothing was connected to the type-C port, what TUSB4041 upstream port was connected to? Was 1.1v only used for TUSB4041? 

    Maybe a block diagram or full schematic could help for debug.

Best

Brian

This was the full schematic for the TPS74611 based attempt? https://flashgamer.com/a/USB2_hub_TPS74611.pdf 

Nothing else is using 1.1V either of the three full schematics. Just the TUSB4041.

J

Can you take waveform between  3.3v/1.1v/GRSTz for hub? 

Regards

Brian

As mentioned, the TPS74611 was two versions (and months) ago so I don't have the boards any more, but the 3V3 was solid and the timing for GRSTz looked right. The 1.1V looked correct when measured with a multimeter, but the scope showed that it was constantly booting. Many hundred times per second. I could not find out why, so I made the second version of the board, this time using NCP139AFCT120T2G that also showed similar problems briefly before dying.

The third board was made with two regulators (just to be sure). One is the TPS7A4533 from there reference design and that works great with it's 1.5A capacity. The second regulator on the third version is using the TPS82676 that I used for a USB 3.1 design with TUSB8041 some years ago. The TPS82676 is only 600 mA and worked fine in many thousand products with the more advanced TUSB8041 hub, but fails with the TUSB4041.

For all the failed regulators, I've been able to solder a few wires with 1.1V from my bench supply and the boards all work great. They just seem to require more than 1A on the 1.1V supply?

TPS82676 is DCDC converter which will switching on and off at 5Mhz, That's why you see it constantly booting on scope.

Maybe Ferret bead will help but  still not recommend to use it as supply for USB devices .

 LDO should not see re-booting issue.

for the board still has issue, can you take waveform on 3.3v/1.1v and GRSTz on scope during power on?

Regards

Brian

Hi Brian,

Here is the capture. Purple = 3V3, Yellow = 1V1, Cyan = GRSTz

Closer zoom on the trigger:

Hi Jens:

     How long is between 1.1v and GRSTz reach 3v?

Best

Brian

Hi Brian,

About 55ms?

are there a big cap on GRStz?

Best

Brian

No. It's a 1uF ceramic (CC0805KKX7R7BB105) as per the reference design. This was measured with the TPS82676SIPR‎ active. 

The below was measured with theTPS74801RGWT (as used in the Eval board) active. It looks very similar, except for the dip at the start. (Sorry about the multiple edits. I had to verify.)

1uf should not take 50ms, did you  control GRSTz by external signal? <3ms is prefered for reset signal.

Best

Brian

No. The connections is per the schematic https://flashgamer.com/a/USB2_hub_TPS7A4533.pdf Only the 1uF cap to GND.

But - is this even relevant for the 1.1V supply? I mean - the board works fine when using the TPS74801RGWT to generate 1.1V. It also works great when using an external 1.1V benchtop supply that I just flip on manually right after powering the board, so the timing seems to be trivial to my problem of power supplies that keep burning out.

Could it be that the TUSB4041i pulls more from the 1.1V rail than stated? I think I need to measure the current draw over time next.

We measured before and it never over 350ma on 1.1v, but you can try on your board t confirm.

Best

Brian