ESD damage to XMOS U6

Technical discussions around xCORE processors (e.g. xcore-200 & xcore.ai).
CMJ
Member
Posts: 8
Joined: Thu Mar 03, 2016 6:09 pm

ESD damage to XMOS U6

Post by CMJ »

We have a custom board which uses the XMOS XS1-U6. On at least 2 boards the U6 has died and we suspect ESD damage from the USB port. We have heard reports of other similar failures in this part. Two questions:

1) Can you please provide a robust ESD protection mechanism for the USB interface signals?
2) Our board is self powered. Is it OK to not attach VBUS and simply reference it to local ground, or does the U6 require this signal for proper operation?


User avatar
Ross
XCore Expert
Posts: 962
Joined: Thu Dec 10, 2009 9:20 pm
Location: Bristol, UK

Post by Ross »

CMJ wrote: 2) Our board is self powered. Is it OK to not attach VBUS and simply reference it to local ground, or does the U6 require this signal for proper operation?
You should attach it - the device needs to know when vbus is valid to turn on its pull-up. This is part of compliance testing.
CMJ
Member
Posts: 8
Joined: Thu Mar 03, 2016 6:09 pm

Post by CMJ »

Ross wrote:
CMJ wrote: 2) Our board is self powered. Is it OK to not attach VBUS and simply reference it to local ground, or does the U6 require this signal for proper operation?
You should attach it - the device needs to know when vbus is valid to turn on its pull-up. This is part of compliance testing.
OK thanks for that info Ross. As for the first part of my post, can you please point me towards a proven robust ESD protection circuit for all the USB signals including VBUS?
User avatar
Caleb
Experienced Member
Posts: 82
Joined: Thu Apr 04, 2013 10:14 pm
Contact:

Post by Caleb »

Ross wrote:
CMJ wrote: 2) Our board is self powered. Is it OK to not attach VBUS and simply reference it to local ground, or does the U6 require this signal for proper operation?
You should attach it - the device needs to know when vbus is valid to turn on its pull-up. This is part of compliance testing.
Please elaborate and link to documentation on this. Using a U8 as a USB2 device we've operated wtih the Vbus pin unconnected. We haven't encountered any problems so far.
Datasheet says:
For self-powered systems, a bleeder resistor may be required to stop VBUS from
floating when no USB cable is attached...
A The VBUS pin is used for measuring the VBUS voltage only.
I don't recall now what motivated us to try leaving it unconnected. XUD still recognizes when USB host attaches.
User avatar
Caleb
Experienced Member
Posts: 82
Joined: Thu Apr 04, 2013 10:14 pm
Contact:

Post by Caleb »

We've used this TI part for several years and a couple thousand units with no apparent ESD related failures: TPD2E001DZDR
It looks like a very similar product to the ST part you're using.

We have had a few mystery problems that we've never taken time to sort-out. Out of a about 500 units there's maybe 8 that don't work correctly when configured for USB 2 but work fine for USB 1. We tried everything we could think-of short of changing the XMOS - including removing the protection device, change the USB connector, etc. Someday we'll send the batch out to have the XMOS replaced.
CMJ wrote: 1) Can you please provide a robust ESD protection mechanism for the USB interface signals?
User avatar
segher
XCore Expert
Posts: 844
Joined: Sun Jul 11, 2010 1:31 am
Contact:

Post by segher »

[ This is a user forum, we are not XMOS. Some people here work for XMOS though;
but for official support, you need to look elsewhere. ]

Allowed ESD stress under the human body model is 2kV. It is much recommended
you use a separate USB protection device, as the datasheets show; do you?

VBUS is not used as supply, only for detecting connect / disconnect.
CMJ
Member
Posts: 8
Joined: Thu Mar 03, 2016 6:09 pm

Post by CMJ »

Hi Segher

Thanks for your help and the intro -- this is my first post so I wasn't sure who ran this forum. I appreciate the assistance! I actually first submitted a support ticket to XMOS and they told me to post my question here :)

We are using a USBLC6-02 for protection connected as recommended in that part's datasheet. However we have still had 2 failures during testing of our prototypes, so it seems likely this is insufficient. Hence I was hoping to get some expertise on a recommended ESD protection circuit that has been proven to be robust.

Do you know if a reference exists?

Thanks,

Chris
User avatar
mon2
XCore Legend
Posts: 1913
Joined: Thu Jun 10, 2010 11:43 am
Contact:

Post by mon2 »

Hi Chris.

Can you share what are the symptoms of your non-working boards ?

Have you confirmed that the voltages on the non-working boards are compliant to the recommended specs ?

Could the fault be some other issue ? Can you share partial schematics and/or PCB layout details ?

Does your device simply plug into a standard PC ? How were these protos tested ? They failed how and when during your testing ? If you replace the BGA XMOS device, the board works again ?

What is your capacitive load on VBUS ?

The ESD device you are using is from ST and openly have a hard time in believing that ST would build in high volumes and ship the referenced component that does not meet the posted specs. You most certainly have many options for ESD protection and also EMI filtering. We can share some vendor details if there is an interest. Our company is about to roll out a list of widgets based on USB 2.0 Full Speed, High Speed and also USB 3.1 Super Speed. For each, we are using different ESD protection devices. In addition, we have a polymer fuse in line with VBUS. To remain competitive, we source these devices direct from the manufacturers in China. You should be able to do the same with even single T&R qty.

Check your power supply rails since this XMOS device contains an internal switching power supply. Many moons ago, we beta tested the Oxford (UK) PCI Express UART line before they went into production and we raised a similar power supply observation in that their power supply did not fire up every time. The comment was not taken as a serious concern. Fast forward about 8 years later and PLX Semiconductor (they bought Oxford Semiconductor) raised an errata note on exactly the same fault we noted at the start of the review. Their work around for the Oxford device was to use an external power supply for the 1v2 rail :(

If you suspect the ST ESD part to be not suitable then you should be able to ping your local FAE at ST and raise the case. They should then request to see the parts on the non-working board for a root cause analysis. If you are using a contract manufacturer, they may (should) have an X-RAY inspection system which would allow to review the insides of a 'good' XMOS device to compare against a 'defective' XMOS device. Will be really interesting to see where the issue lies for your boards. Please proceed till you find a suitable answer else you will have the wrath of RMAs.

================

We are using the following device in our pending projects for USB 2.0 & USB 3.1 Gen 1 (USB 3.0):

http://www.socay.com/upfile/201407020248027678.pdf

Notes:

1) the Vbr value is 6 volts as compared to the TI part noted by Caleb @ 11v. Lower is better here.

2) Capacitance is 0.3 pf as compared to the TI part @ 1.5 pf. This part is suitable for USB 2.0 HS (480 Mbps) and also USB 3.1 Gen 1 (aka USB 3.0) 5 Gbps. Lower capacitance is better.

3) Socay part is IEC61000-4-2 (ESD); IEC61000-4-4 (EFT); IEC61000-4-5 (Lightning) compliant vs. TI part being IEC61000-4-2 (ESD) only. More the better.

Socay is a true manufacturer of the parts and they often supply the many brand names seen in the catalog shops like Digikey and Mouser. We have met with this manufacturer in HK at past trade shows and their products match the specs they advertise. We have verified this against their line of polymer fuses and where others failed during our testing, Socay met all of our requirements. They are certified against the required specs including UL, CE, CSA, etc.

They are flexible on qty and the parts are very low cost as compared to their landed branded versions.

For example:

Part Number: ULC0524P
Quantity: 3 K
Unit Price: 0.042 USD/ pc

Same as Littlefuse SP3010 device:

http://www.littelfuse.com/~/media/elect ... et.pdf.pdf

Contact details:

Cindy
socay012@socay.com <socay012@socay.com>

We also encourage you to place an in-line polymer fuse on the Vbus line to limit excessive currents. Socay also offers these devices in assorted current / trip values. Hope this helps.

Kumar
Last edited by mon2 on Fri Mar 04, 2016 6:42 pm, edited 1 time in total.
CMJ
Member
Posts: 8
Joined: Thu Mar 03, 2016 6:09 pm

Post by CMJ »

Caleb, thanks for your response. We are finalizing schematics for our next prototype build and we will leave provision for trying VBUS disconnected as an option.
CMJ
Member
Posts: 8
Joined: Thu Mar 03, 2016 6:09 pm

Post by CMJ »

Hello Kumar

Thanks for all the detailed information and advice. I am reviewing it with my team and will investigate more fully. Here is a bit more detail.

When the units fail, we can no longer attach to XMOS via XTAG, and it stops being visible to the host computer.

The board is designed to simply plug in to a standard PC. At some point during use, we noticed that they stopped working with the PC. Nothing unusual had been done to the units, but the USB cable is unplugged and reinserted frequently so we suspected ESD.

We have not yet tried replacing XMOS but that is a good idea.

We have done design verification around the part and we are within tolerance, but will review again. Since the parts stopped responding we have not yet done a thorough investigation of all voltages, etc. on the broken boards. That is an interesting history on the power supplies, thanks for that.

VBus has a 4.7uF capacitor attached on our board.

The Socay part you are using looks interesting and I think that it would offer better protection than the ST part we are using. We will look into switching to this part for our next board rev.

Our first thought was that something is not right with our VBUS connection since there is nothing in series with the line, and thus it relies 100% on the ST device for protection. We will definitely look into adding a polymer fuse.

Thanks again,

Chris
Post Reply