| FORUM

FEDEVEL
Platform forum

USE DISCOUNT CODE
EXPERT30
TO SAVE $30 USD

Forum / Your Projects, Technical Discussion / TUSB8020B HUB getting really hot, any suggestions?

TUSB8020B HUB getting really hot, any suggestions?

elessargr , 09-05-2025, 07:38 PM

I remember seeing somewhere this
https://www.ti.com/tool/TIDA-00287
being sold and now I cant find it anywhere. I was thinking maybe to purchase this and check if indeed the temperature gets that high that quickly 🙁

elessargr , 09-05-2025, 07:39 PM

The weird thing is that yes it does have more vias for the IC but I see them only around the pad not in the center.

corlas , 09-05-2025, 07:46 PM

you can buy this if you want eval board but it's quite expensive https://www.ti.com/tool/TUSB8020BEVM#order-start-development

elessargr , 09-05-2025, 08:59 PM

Hi Corlas!
Yes I saw this also but indeed its expensive 🙂

Robert Feranec , 09-06-2025, 10:40 PM

yes, you may want to order it, it can be helpful

elessargr , 09-07-2025, 11:39 AM

https://www.krypton-solutions.com/products/ks2ph-2-port-usb-3-0-hub-reference-product/

Finally found it!
but looks like they dont sell it anymore :S
I reached out to them to see.

tomas.kaplan , 09-07-2025, 05:53 PM

Before you turn on your board, try to short the GRSTz signal with tweezers to ground. Hold it there, then turn on the board and release the signal. Thats gonna be quick test that could say a lot.

elessargr , 09-07-2025, 06:51 PM

Hi Tomas,
the GRSTZ (pin11) on the IC TUSB8020BI is actually connected to the ground with a capacitor before.
So you mean just to bypass the capacitor?

tomas.kaplan , 09-08-2025, 05:10 AM

Yes, exactly. Short it across the capacitor and turn on the board.. then release the short.

QDrives , 09-08-2025, 01:45 PM

Better yet, check the temperature before releasing the reset.

tomas.kaplan , 09-08-2025, 01:49 PM

True. It should be completely non-operational with reset shorted to ground.

elessargr , 09-08-2025, 04:55 PM

Hi team!
So tried this and the IC gets 35-38C. 🙁

elessargr , 09-08-2025, 04:55 PM

Room temprature here is 22C

QDrives , 09-08-2025, 07:42 PM

Rθja = ~31°C/W
You get over 10°C rise, so about 0.3W.
Most of the power is drawn on the 1.1V.
With no devices connected, it should draw about 40mA and 6mA from the 3.3V. This total to less than 70mW.

elessargr , 09-08-2025, 08:31 PM

Forgot to mention that once I normalize and remove the ground bridge from pin11 it goes to 50-55+C with USB3 connected to it and USB2 also. There isnt any traffic going through though.

QDrives , 09-09-2025, 02:10 PM

"*...with USB3 connected to it and USB2 also...*" -- on the input or output?
What current does it draw when you power it just with 1.1V?

elessargr , 09-09-2025, 04:24 PM

As a downstream devices.
I dont have a way of checking this unfortunately 🙁
I have only a multimeter and a microscope.

elessargr , 09-09-2025, 05:35 PM

At 0:06 I connect the PCB to the PC (USB3) while pin11 is short to the ground.
At 0:40 I remove the short.

Robert Feranec , 09-09-2025, 10:46 PM

hmm, it could be normal temperature for a sistuation with no heatsink or weak exposed pad thermal connection. 47C is not that much.

corlas , 09-09-2025, 10:49 PM

well yes, but while the IC is in reset state, it gets 10C+ increase? kinda weird.

Robert Feranec , 09-09-2025, 10:51 PM

then only the way to double check would be to try to measure the currents

QDrives , 09-10-2025, 02:30 AM

Do note that Junction-to-top is 16°C/W and Junction-to-board is 13°C/W. Your board cannot cool what is not there.

So in what "mode" Is the hub?
What if you remove R9? It should be in "Disconnected from host" mode.

elessargr , 09-10-2025, 04:14 PM

If I remove the R9 then Pin9 (USB_VBUS) would not get any power. This is what you mean?

elessargr , 09-10-2025, 04:44 PM

I found this video for USB testers/meters
https://www.youtube.com/watch?v=uzT8mbrr4_M
very helpful.

elessargr , 09-10-2025, 04:44 PM

I looked online and I see the POWER-Z C240 also is a good. Need to find out the accuracy for current and voltage before I decide to purchase it.

QDrives , 09-10-2025, 07:29 PM

Yes, it will get power, but should stay in the "not connected to host" mode.

elessargr , 09-11-2025, 06:15 PM

okay let me try that

elessargr , 09-11-2025, 07:55 PM

So this is with R9 removed.
The PCB I dont think its get any increase in temprature. Maybe few degrees since I can see the IC showing up but nothing as before.
Also when I connect it to the PC its not recognized. So I think the IC indeed is fully off.
Ignore the IC on the back.

QDrives , 09-11-2025, 08:32 PM

No, with R9 removed, the IC should still be powered. It is just in the "Disconnect from host" mode drawing little current.
The PC should not recognize it, as the IC thinks it is not connected.

If it does not get hot this way, we can mostly rule out any shorts below the IC between 1.1V and 3.3V to Gnd.
Then you have to check all other connections and modes.
One last 'failure' could be that the thermal pad has not been soldered properly to the board.
Then you need to check the temperature on the back of the board during power-up.

elessargr , 09-11-2025, 09:06 PM

hmm interesting.
Need to read and find out how to do the "check all other connections and modes."

Well I know that on the current prototype version of my PCB I dont have enough VIAs. I placed only 4. The back of the PCB is getting hot definitely. I have couple thermal videos that shows this. Front side on the IC I see 55-62C and on the back is 53-56C

QDrives , 09-12-2025, 01:59 PM

I would verify pin assignments first, both in schematic and layout.
Not only of the hub, but also of the things connected to it (like connector and mux).
Also check that there is no short to the connections encircled to Gnd.

elessargr , 09-12-2025, 05:29 PM

https://streamable.com/oyixnb
I think the last item about shorts is okay since I did a test and remove the IC and place another one from my previous prototype just to check if it was an IC problem or not. Yes I put extra solder before removing the IC.

Regardless I will go over those and check the pins

elessargr , 09-16-2025, 05:25 PM

So looks like ChargerLAB POWER-Z C240 doesnt pass USB3.0 🙁
And also I noticed that the IC doesnt get hot with it by touching it. I will do tests with the thermal camera and the C240 to see the results

elessargr , 09-19-2025, 06:30 PM

okay so bought another USB power meter (USBC-VAMETER)
which has the capabilities to pass USB3.0

Here what we have:
1 -PTv1.6 (no downstream port connected)
--V: 4.96-5
--A: 0.06/7
--W: 0.3/35

2 -PTv1.6 (one downstream port connected HS)
--V: 4.96-5
--A: 0.14
--W: 0.69

3 -PTv1.6 (two downstream port connected HS)
--V: 4.96-5
--A: 0.15
--W: 0.74

4 -PTv1.7 (R9 removed)
--V: 4.96-5
--A: 0.04
--W: 0.2

5 -PTv1.7 (pin11 GRSTZ grounded)
--V: 4.96-5
--A: 0.15
--W: 0.74

6 -PTv1.7 (two downstream port connected 1x SS & 1x HS)
--V: 4.96-5
--A: 0.25/26
--W: 1.24/28

I have two PCBs which are different prototypes.
Prototype 1.6 (PTv1.6)
does not have any downstream port connected but I can connect two HS. SS I cant since there was a mistake on the traces.
&
Prototype 1.7 (PTv1.7)
have both downstream ports connected with traces in the PCB.

QDrives , 09-19-2025, 09:50 PM

I assume the "A: xxx" is the current (5V) in Ampere?
Is the 40mA in measurement 4 realistic with everything connected?
Comparing the values of 6 with the datasheet, I would say that you device does not require (much) more than the datasheet mentions.

elessargr , 09-19-2025, 09:53 PM

Hello QDrives,
Yes you are correct its 5V. Sorry I just added this

elessargr , 09-19-2025, 09:56 PM

On measurement 4 the Resistor9 which provides power to pin9 (USB_VBUS) is removed. Yes the downstream ports are connected since there is no way for me to disconnect them on my PCB PTv1.7.
But with the R9 removed the IC we said is in idle state and thinks that no downstream ports are connected (disables them).

elessargr , 09-19-2025, 09:59 PM

> Comparing the values of 6 with the datasheet, I would say that you device does not require (much) more than the datasheet mentions.
So from your experience the IC uses the right current and 63C is fine. Correct?

elessargr , 09-19-2025, 10:02 PM

This is from measurement 4

QDrives , 09-20-2025, 02:03 AM

R9 does not provide power. It is part of a voltage divider to give a signal that the host is connected.

QDrives , 09-20-2025, 02:05 AM

Yes, I would say that the current you measure, match the current (power) mentioned in the datasheet.
The temperature, well, you need to improve your layout to reduce the temperature --> bigger coper area on the top and more vias.

elessargr , 09-20-2025, 07:47 AM

thank you soo much sir!
trying to see what other improvements I can make to my PCB before making the next prototype.

elessargr , 09-20-2025, 12:54 PM

I have put bigger vias, larger gnd areas and I will ask for those vias to be filled with solder so they can transfer more heat to the other layers. I think this should be good 🙂

QDrives , 09-20-2025, 06:56 PM

See if you can route that 3V3 outside of the IC. Optionally even let both pins have their own via.

elessargr , 09-20-2025, 07:31 PM

You mean on the first layer (red) right? the ones that are on the inside of the IC that I have.
Need to check if I can more traces around to make space for those.

elessargr , 09-20-2025, 10:08 PM

Managed to remove the other vias from under the IC but the two top I cant. Too many traces on the other layers 🙁 I will keep checking