ESD protection for 2 USB port connectors

[LMorr.3]

Hello,

I setup a USB virtual com port which I use for serial comm.  

My design has 2 boards; a panel-board and main-board.  The main-board has an STM32F373 with a USB-C connector as well as ESD321DYAR diodes for DM and DP USB line protection.

The main-board can be used on its own, but can also be optionally connected into the panel-board ( a front panel ).  For convenience, I would like the panel-board to also have a USB C connector so that the user can plug a USB cable into the front panel instead of the main-board.  Both USB C connectors lead to the same USB virtual com port on the main board.

I would like to add ESD321 protection close to both connectors, but noticed having 2 ESD on the DM/DP lines instead of 1 ESD resulted in failed USB communiation.  I'm assuming the 1.4V drop ( instead of 0.7V ) caused by the diodes is too much and causing the failed comm.  

What is the best way to ESD protect both connectors?  I currently only have ESD diodes close to the main board connector, but am concerned they are too far away from the panel-board connector to protect adequately.   Should I add jumpers to allow the user to disable the main board USB connector if/when the panel-board is used?

( Sorry if this is a bit off-topic )

Thank you,

[gregstm]

I haven't done much ESD/EMI stuff for a fair while, but when I did, I found the application notes and products at Murata very useful and interesting. Maybe have a browse through their stuff. I used a lot of their 603 sized ferrite beads in my design (mainly for EMI purposes)

https://www.murata.com/en-global/products/emc/esd/overview/basicesd/chapter1

[LMorr.3]

To answer my own question:  If the USB pcb traces have no other components in the circuit between the panel-board and the main-board, one set of ESD protection will suffice. 

[LCE · Accepted solution]

---

@LMorr.3 wrote:

I would like to add ESD321 protection close to both connectors, but noticed having 2 ESD on the DM/DP lines instead of 1 ESD resulted in failed USB communiation.  I'm assuming the 1.4V drop ( instead of 0.7V ) caused by the diodes is too much and causing the failed comm.  

---

This is strange, about 1 pF more on the USB lines should not affect functionality, unless it was close to breaking down in the first place. So better check your routing.

And a diode voltage drop shouldn't have to do anything with that problem, I hope you connect the ESD diodes from signal to GND, and not serial?

[LMorr.3]

Thanks for the additional info.  I suspect I may have connected the ESD diode from GND to Signal by mistake.  I removed 1 set of ESD and it then worked.  I just assumed having two sets of ESD diodes caused the issue but now realize it may have been something else.  I had the first set of ESD connected in the same direction as the second set, and it worked after removing the first set so I'm guessing they are in the right orientation.  

The ESD321 part is 0403 and I needed my digital microscope to determine pin 1.  The datasheet states the package I used also 'swaps' pin 1 with pin 2 making things more confusing.  I may be making a mistake on pin orientation... I will investigate further...

[LCE]

Oh yeah, 0402, not possible without magnifying glass anymore - 12 years ago I could solder 0201 without - my eyes age badly, it seems... :D

[LMorr.3]

It turns out I had the pins swapped, thanks for the heads-up.  

Only 1 set of ESD321 parts are recommended in my case, but it's good to know my original issue was caused by swapped pins.  The second set of ESD did NOT have pins swapped which explains why it worked when I removed the first set.

Thank you!

[MROUV.1]

Hello,

To protect USB port, I recommend USBLC6-2P6 (ESD only) or ECMF2-40A100N6 (ESD + CMF) to place as close as possible to the connectors.

Your issue may be not link to Vf voltage of the ESD protections when placed in parallel.

I suspect an impedance mismatch when both boards are connected to the same USB transceiver of STM32F373, or stub effect due to the unconnected port.

I recommend disconnecting the unused D+/- line to grant 90Ω differential impedance on the other line all long the differential pair.

The best solution is to use 0 Ohm resistors else jumpers.

Best regards,

Mathieu

[LMorr.3]

Thanks for pointing out the stub effect.   I'll include jumpers to disconnect D+/-.  Worse case if jumpers are left enabled by the user is the stub effect making the comm less reliable and no 'damage' to hardware.