USB connector design

[hsn93]

hello,

this is from XPlained-Pro evaluation board



and i have two questions:

1. why the shield is connected to ground through capacitor .. and how its calculated exactly


2. the TVS is connected to GND -> is this right ? i can see other design guide from FTDI:
http://www.ftdichip.com/Support/Documents/AppNotes/AN_146_USB_Hardware_Design_Guidelines_for_FTDI_ICs.pdf
they put connect TVS to the Shield not the GND direct

[ataradov]

1 .That's how USB standard recommends doing it. There are no exact calculations, the exact value is not all that important and you can drop the whole thing entirely for most designs.
2. Most of the designs I've seen use the ground. But realistically, I doubt it makes a difference.

[Chriss]

The C and R are a RC filter-noise reduction stuff.

Practically I would connect the TVS to shield, but to gnd is also almost the same.

Sent from my GT-I8260 using Tapatalk

[wraper]

Practically I would connect the TVS to shield, but to gnd is also almost the same.

By doing so you cripple ESD protection of IC connected to USB data lines and also increase risk damaging capacitor between shield and GND during ESD event (yes capacitors can be damaged by ESD too). As of placing something between connector shield and GND, IMO it should be done only if connector shield is connected to GND by some other means like conductive gasket between connector and grounded metal enclosure. FWIW, the vast majority of micro USB cables have shield and GND wire sorted together at connector. And the strangest is that placing ferrite bead instead of capacitor seem to be just as popular, despite that it has completely opposite effect.

2. Most of the designs I've seen use the ground. But realistically, I doubt it makes a difference.

And where goes return path for current from ESD discharge? If your intention is protecting usb connector from discharging between it's pins and shield, then sure, you can connect protection device to connector shield. But you are not protecting IC. It will mean that voltage will rise on both data pins and shield in reference to GND to which IC is connected, and IC will still receive hit on it's USB pins. However if metal enclosure which is connected to connector shield is used, then sure you could connect ESD protection to connector as it should act as current return pant.

[wraper]

Just look at part of Arduino schematic. They nullified the effect of ESD varistors by connecting them through ferrite bead 

[wraper]

I'd say ESD protection device manufacturer knows better.
http://www.st.com/resource/en/datasheet/usblc6-2.pdf

[Neilm]

There is no generic correct way to do this - I have designed with several different ways depending on the requirements. If there is an earth path, I would directly connect to that (assuming the 0V was star pointed or isolated from Earth). Otherwise connecting to 0V via the RC is usually sufficient. The cap shunts the high frequency and the resistor slowly discharges it.

[wraper]

And here is a document from TI which suggests to place ferrite bead to GND pin isolate it at high frequency and allow low resistance path for DC. The catch is that shell shall be connected directly to metal enclosure. Or if there is no metal enclosure, suggests connecting shell directly to the ground plane.
https://www.ti.com/sc/docs/apps/msp/intrface/usb/emitest.pdf

The shell should be provided with a low impedance connection to signal ground and, in cases where
a metal chassis is present, a means of accepting a metal EMI gasket clip.

[T3sl4co1l]

The catch is that shell shall be connected directly to metal enclosure.

[Ed: emphasis]

Basically no appnote gets this.

So you end up with everyone using literally the flimsiest possible ground connection.  It's embarassing.

For 99% of what you're doing with USB:

GROUND THE SHELL!

Ground it hard, in multiple points, to the ground plane.  And obviously, use a ground plane, too!

Tim

[ataradov]

GROUND THE SHELL!

On the host side only. If you ground it on the device side, then the current will flow through the shield, eliminating any positive EMI effects it may have.

Also, it appears that there is really no consensus on this at all. And whatever opinion is expressed here by anyone can't be 100% true without a really good justification.

On the other hand, even with all the different implementations, things still works fine.

[Karel]

Basically no appnote gets this.

So you end up with everyone using literally the flimsiest possible ground connection.  It's embarassing.

For 99% of what you're doing with USB:

GROUND THE SHELL!

Ground it hard, in multiple points, to the ground plane.  And obviously, use a ground plane, too!

Tim

I can confirm this. I burned myself once. Now I always connect the shell to the groundplane. No resistors, no capacitors, no ferrites.

[T3sl4co1l]

On the host side only. If you ground it on the device side, then the current will flow through the shield, eliminating any positive EMI effects it may have.

 

Dude, that's precisely what it's there to do!

Also, it appears that there is really no consensus on this at all. And whatever opinion is expressed here by anyone can't be 100% true without a really good justification.

I haven't read a single book, since vacuum tubes were in vogue, that ever claimed open shields was a good thing.

Do you connect your cable TV without the shield?  Audio/video cables?  HDMI cables?

Do you connect your Ethernet with one wire in each pair floating?

Tim