FEDEVEL
Platform forum

elessargr , 02-14-2026, 05:45 PM

> Your hub does have a PwrCtlX pin....
You mean the pin5 on the USB HUB (PWRCTL1/BATEN1) ?

Plus those two?
- PWRCTL_POL/SS_DN1 (pin21)
- FULLPWRMGMTz/SMBA1/SS_UP (pin36)

I dont use battery for this PCB to where its connected to if you mean that.

QDrives , 02-15-2026, 01:10 AM

Where does your power come from? All from the PC through USB?
Anyhow, I think you better check the signals with an oscilloscope.

elessargr , 02-15-2026, 10:51 AM

> Where does your power come from? All from the PC through USB?
Yes, only from the PC (host device).

elessargr , 02-15-2026, 10:51 AM

I am waiting for the lad that has an oscilloscope to do the checks and tell me the results

QDrives , 02-15-2026, 04:56 PM

How many boards have you got?

elessargr , 02-15-2026, 05:27 PM

4x in total.
2x of the previous prototype that has same layout and 2x of the latest prototype.
Both of them have the same issue with the USB2.0 on the ETH IC.

QDrives , 02-16-2026, 11:52 AM

So you can look with the oscilloscope both at one with pull=up resistor and one without it.

elessargr , 03-08-2026, 09:39 AM

okay so, update here.
The lad that has the oscilloscope was able to do some checks and mentioned to me that there isnt anything abnormal in the D+/D- values from what he can see during the cold start in this configuration that I have.

elessargr , 03-08-2026, 09:39 AM

It happen also to have the USB-ETH adapter as an individual PCB with the same parts as we were doing some test prior, on an older prototype version, and when it was power up after a cold start it was detected without issues under a USB2.0.

elessargr , 03-08-2026, 09:39 AM

So this weird state its probably coming from the USB HUB (TUSB8020BI). We are thinking that it has to do with the power coming from the HUB IC and it may being able to get configure with the PWRCTL pins.

elessargr , 03-08-2026, 09:41 AM

Not sure if the PWRCTL can be configure in a way that once the HUB IC is waken up then it powers up the downstream port ETH IC.

QDrives , 03-08-2026, 07:10 PM

The power control pins do not power your ETH directly. You need a "load switch" for that.

elessargr , 03-15-2026, 10:47 AM

I see,
Any recommendation how I can do this or schematic?
I assume I need to use the pwrctl pin and something close to this.

QDrives , 03-15-2026, 04:08 PM

Pick one from these: https://www.digikey.nl/en/products/filter/power-management-pmic/power-distribution-switches-load-drivers/726

elessargr , 03-22-2026, 09:16 AM

Thanks! I was able to find some different options.
Now trying to see how this is designed or a schematic. My design doesnt have an EPROM so I am trying to avoid it.

QDrives , 03-22-2026, 05:20 PM

Can you make a patch on your existing board to test if it works before doing a redesign?

elessargr , 03-22-2026, 07:38 PM

I can definitely try depending on how hard is it to implement it. I am not an expert by any mean but I believe I can do some mods 🙂

Trying to see how a load switch works and what should be connected where. That is why I am looking for a schematic.

elessargr , 04-18-2026, 10:15 AM

okay so I think I figured it out. Maybe? 😅

elessargr , 04-18-2026, 10:16 AM

I used the https://www.ti.com/lit/ds/symlink/tps22918.pdf as its accepting the 5v main power rail for the ETH IC.

elessargr , 04-18-2026, 10:18 AM

and updated the 5v power rail of the ETH IC to be from the power switch +5V_RT

elessargr , 04-18-2026, 10:18 AM

Does the first schematic looks correct?

elessargr , 04-18-2026, 10:24 AM

or this works actually?

QDrives , 04-18-2026, 01:57 PM

If it is a single power (input) pin, why not test that first on an existing board?
Just cut the connection, solder some wires, and when the wires touch, the ETH interface should work.

elessargr , 04-18-2026, 06:23 PM

I did this with cutting the power to the ETH IC with some wires and connecting it carefully after a cold start. It seems to be working.

One thing I noticed is that every time I energize the ETH IC the whole board seems to be resetting. So not sure if this is normal or not.

QDrives , 04-18-2026, 08:05 PM

You might want to reduct the capacitance of C10 to <= 1uF and increase that of C7 to 10uF.

QDrives , 04-18-2026, 08:05 PM

A reset of the hub (use USB Event viewer) would invalidate this whole test.

elessargr , 04-18-2026, 09:05 PM

C7? Aaa from the PCB you saw it.
Its part of the MUX I have for the USB-C.
I used to bridge the two capacitors just to use them for easy connection of the 5v rail as I cut it for the ETH IC.

elessargr , 04-18-2026, 09:05 PM

The USB Event viewer you mean of the windows right?
The reset its happening automatically when I connect the +5v rail of the ETH IC to get power from the cable you see between the two capacitors.

QDrives , 04-19-2026, 01:24 AM

A reset can be caused because the capacitances are wrong.
The 'powered' (the hub in this case) should have a (much) higher capacitance than the 'switched' load (ETH bridge in this case - C10?)
If the hub resets, you should get a message also in the USB event viewer (tool on windows) or some other USB logging tool.

elessargr , 04-19-2026, 09:39 AM

Added some more details here to help understand the PCB.

The 5v rail that I have for the MUX passes through the CAPs or the ETH IC (C51 & C53) plus some resistors for LEDs that I have.
The LED resistors I dont think, I guess, causes any issues.
Yellow X are the cuts to the trace on the PCB.
Light blue is the wires I have placed.

elessargr , 04-19-2026, 09:39 AM

Also doesnt the capacitors "clean" the 5V signal?
If this is an issue I can try to find a way to get another 5v rail without any capacitors in the path to the MUX.

elessargr , 04-19-2026, 09:39 AM

The HUB IC gets 3v3 and 1v1 from a power regulator.

elessargr , 04-19-2026, 09:50 AM

PS: just disclaimer, I am not by any stretch great at PCBs or electronics. Still learning and trying to understand how things works with my first PCB 🙂

elessargr , 04-19-2026, 10:21 AM

okay so I did same test again.
From a cold start (powering up the PC) with the PCB connected to a USB2.0 port.
The USB HUB is detected.
Then bridging the ETH IC 5v and actually it did detect it.

Not sure what was the reset happening previously with the HUB.
From the USBLogView it seems that the reset was actually happening to the second downstream IC I have and not actually the HUB.

So I guess the load switch "should" fix the issue?

QDrives , 04-19-2026, 02:15 PM

With C7 and C51 being 100nF each and C53 being 10uF, that gives you a 50:1 capacitive divider.
So C7 and C51 are charged to 5V. C53 is 0V.
Touch C53 to 'blue line' and you charge C53. However, the impedance of the circuit, makes it drain C7 and C51.
So you end up with 5V * (C7+C51) / (C7+C51+C53) = 200nF / 10.2uF ~ **0.1V !**

That is why I said you need to decrease the capacitance of C53 to less than 1uF (or equal) and increase the capacitance of C7 and C51 to be (combined) >= 10uF.
A simple swap of C53 with either C51 or C7 would do the trick.
In that case you get 5V * (10u + 100n) / (10u + 200n) ~ **4.95V**

QDrives , 04-19-2026, 02:17 PM

Also do not place a trace between pads as pointed to by the green arrows.

QDrives , 04-19-2026, 02:18 PM

Or you will end up 'fixing' shorts as shown in this photo.

elessargr , 04-19-2026, 03:55 PM

I guess the swap of C53 with the C51 its only needed for the test right?
As afterwards there is no difference on the design as they are part of the same power rail.
Or I need the MUX to get a different 5v rail so its C7 doesnt have anything prior to it?

elessargr , 04-19-2026, 03:55 PM

Thanks for the insight of the trace between the pads!
Question on this, does it make difference as those bridged pads are GND or its just best to extend their pads to the main GND plane below the IC?

elessargr , 04-19-2026, 03:59 PM

As I am trying to find out how the ETH IC can be power up after the HUB since the main issue is that with USB2.0 cold start the ETH IC is not detected. if everything is connected as normal.

QDrives , 04-19-2026, 04:05 PM

"*I guess the swap of C53 with the C51 its only needed for the test right? *" -- Correct.
"*As afterwards there is no difference on the design as they are part of the same power rail.*" -- No, it is a different power rail as the ETH bridge needs to be powered through a load switch.
For that you still need to keep an eye out for the capacitive divider.
Also note, that if it works 'correctly' through a load switch, that there is a flaw in the design of the ETH bridge to begin with as it should not need to be working only when switching on.

QDrives , 04-19-2026, 04:07 PM

As you can see in the photo there are 3 shorts between pins. I spend several hours to 'fix' those only to discover once I checked the layout that they are supposed to be connected together.
Functional wise, there is no difference if you connected all 4 pins directly to the thermal pad. It is even slightly better to do so.

elessargr , 04-19-2026, 04:13 PM

> No, it is a different power rail as the ETH bridge needs to be powered through a load switch.
A true yes. If I use a load switch I probably need to check what is after the Load switch?
Like C10, C51, C53, C54, C55 correct?

elessargr , 04-19-2026, 04:16 PM

> Also note, that if it works 'correctly' through a load switch, that there is a flaw in the design of the ETH bridge to begin with as it should not need to be working only when switching on.
While another person also mentioned this that it may be a flaw on the ETH IC it self, I cant change it and so I can only work around it. I agree it should be "working" also when the HUB IC is not active but I dont think its matters as when the HUB IC if off there is no connectivity between the PC and the ETH.

Now that I am thinking about maybe the WakeOnLan would not be working if the ETH IC is not on... Hmm 😫 Good point. Need to think more on that as I would like to get this feature available on my PCB.

elessargr , 04-19-2026, 04:18 PM

aa yes, No the pad that I have bridged there is a soldermark between them so the connection is clean.
Regardless if its a good practice to make the connection directly to the GND main plain below the IC I can do this easily 🙂

QDrives , 04-19-2026, 07:06 PM

Not C54, that is before the switch.
I noted that your selected load switch does not have a current limit. May be wise with the capacitors to charge.
And in the U7 schematic part - note that the input capacitor (C54) is 1uF while the output is 22uF (C55). This creates a capacitive divider.

QDrives , 04-19-2026, 07:07 PM

The load switch is driven from the PWRCTL1 of the hub. So your ETH bridge is powered as soon as the PC gives it OK to do so.

QDrives , 04-19-2026, 07:09 PM

Solder mask in your design -- but is there solder mask every time it is being produced?
The design on the photo I presented also has solder mask between the pins, yet often removed in production.

elessargr , 04-19-2026, 08:34 PM

but if I wish to get the WakeOnLan feature of the ETH IC then I need to remove the loadswitch or think about it more carefully .
Right?

QDrives , 04-20-2026, 01:26 AM

The load switch is controlled by the PWRCTL(1) pin. That is controlled by the hub. That is controlled by the PC.
So no, the ETH bridge is powered as soon as the hub is ready.
It does allow the power save feature, but that depends on the PC again.