TUSB8043A: Operating concern, power supply

Hello VietCuong,

Can you confirm that the thermal pad of the TPS74601 is connected and there is adequate solder coverage?  Also have you tried increasing the bulk capacitance on the 1.1V rail to 10 uF?

Regards,

JMMN

Hi JMMN,

Yes, the thermal pad of the TPS74601 is connected. I tried increasing input and output capacitance on LDO but it still doesn't improve.

Regards,

VietCuong

Ok, are you sure there is enough solder to connect to thermal pad.  We use this device on our EVMs and we have not had thermal issues.  Does the hub work?  What state is it in and can you measure how much current is being drawn?

Regards,

JMMN

I have measured the supply current for usb hub is 0.35A at 1.1V  when the usb hub is started working.

I think LDO has no problem, I have tried 2 other 1.1V supplies and i have tried 0.5A for TPS74601, it is not as hot as provided for the usb hub.

Regards,

VietCuong

So you have tried two other 1.1V supplies and the board gets hot only around the 1.1V supply area?  Does the hub get hot?  The hub only draws power from the LDO it doesn't have any sort of internal regulator where it could back drive power.  Is the hub functional and are downstream devices active?

Regards,

JMMN

Downstream devices are working properly, the hub heats up but the LDO is much hotter than the hub.

The hub only gets 1.1V power from LDO TPS74601

Hi VietCuong,

I am moving your ticket over to the power forum to see if they can determine why the TPS74601 is getting hot in the system.  This is not caused by the TUSB8043A.

Regards,

JMMN

Hi VietCuong,

How much hotter is the LDO getting than the rest of the board? if you used the JEDEC layout then you would be looking at a ~60C rise on the LDO with the power being dissipated. 

Can you also share a more detailed layout around the LDO? it will help us determine what is going on here.

Thanks,

Mark

Hi Mark 

I used Flir to check the temperature around the LDO, about 62 degrees Celsius.
You can see more pictures below for more information. I wonder why when I use another 1.1V power supply, the situation is the same.

Thanks,

VietCuonghttps://e2e.ti.com/cfs-file/__key/communityserver-discussions-components-files/138/layoutimages.7z

Hi VietCuong,

Can you attach the files independently? My firewall will not let me download 7zip.

When you say you are using another 1.1V supply are you putting it on the same foot print or wiring it in externally?

Can you also provide a DC coupled scope plot? I wan to check how the device is regulating.

Thanks,

Mark

Hi Mark,

Yes, I have tried another 1.1V power supply to the hub.

You can see images below.

Thanks,

VietCuong

Hi VietCuong, 

Has your question been addressed?

Best,

Juliette

Hi Juliette

I have no answer nor the solution to this problem.

The temperature of the LDO has not exceeded the permitted level as in the datasheet, but I am concerned that the above operating conditions affect other supplies and the operation of the board.

Do you have any ideas ?

Thanks,

VietCuong 

Hi VietCuong,

I am a little bit confused by your response earlier. If you are supplying the 1.1V rail with an external supply and it is still at 60C then doesn't that suggest that it isn't the LDO that is causing the localized heating?

Regarding the LDO layout I don't see a lot of thermal vias around the LDO which helps spread the heat to the rest of the board, and the way that the input and output traces were routed really stops it from having enough topside copper to dissipate the heat correctly. 

Thanks,

Mark

Hi Mark,

You can review my questions above. You pay attention to AC/DC coupled scope plot, which is the output waveform of the source LD0 - 1.1V and also the input of the USB hub.

When I isolate the LDO from the hub, the waveform is a straight line, the ripple is very small.

I tried dummy load 0.35A which is equal to when the hub works, LDO doesn't heat up that much.

Thanks, 

VietCuong

Hi VietCoung,

Using an external supply likely improves the AC dips due to extra capacitance and/or higher current rating of the external supply, so that makes sense. However, it seems strange that the LDO would not reach the same temperature when you apply the dummy 0.35A load. This makes me think that the 0.35A load that you measured from the USB hub is the average current, however the USB hub may require transient spikes in current which are higher and correlate in time with the dips seen in the AC coupled scope plot. Most DMMs have default to averaging their measurements for ~16ms and the dips are a small portion of the total waveform so we may be missing a significant amount of current. 

We can reduce the AC dips by increasing the output capacitance as suggested earlier will likely improve the AC dip, however it will not reduce the temperature.

To reduce the temperature you can either reduce the power dissipation (likely difficult unless you have another, lower voltage rail since the load current is dictated by the USB hub) or the PCB layout can be optimized to improve thermal dissipation. We have an App Note which shows the difference different layouts can have on the thermal dissipation I have also included a power point on LDO thermal performance which provide guidelines on how to maximize thermal dissipation of a PCB. 

LDO Thermal Performance.pptx

Hi Kyle,

Thanks so much for your help.
- As you said, the 0.35A current that I measured when the hub is active is the average current.

- When the output capacitance increases, the operation cycle still occurs, it only decreases the spike.
- My concern now is the LDO output waveform when supplying to the active hub, if not for the active hub it is normal. When using other 1.1V sources, the cycle waveform remains the same and IC heats up.
- I don't know much about usb hubs, is it the way the usb hub (TUSB8043A) works that causes that waveform?

VietCuong

Hi Kyle,

Yes, that's right
when that happens, the 3.3V power rail (LDO input) also has a similar waveform.
Is that the normal operation of the TUSB8043A?
I am concerned that it will affect the power supply and other components.

Thanks

VietCuong

Hi VietCuong,

My expertise is in power and LDOs so I can't speak to the TUSB8043A specifically but many loads have periodic current needs so this is normal to see for many power rails. 

Perhaps JMMN can elaborate more on the USB hub's expected behavior.