hot- or cold-plugging altera cyclone-3 LVDS inputs causing damage?

Hi!

We have a design where an Altera Cyclone-3 (EP3C5) with LVDS inputs is
connected to a sender on another board through a cable.

In a number of cases on our prototype boards, the LVDS-inputs have
been fried from this setup.

Apart from any misdesign in the sender-board, does anyone have any
suggestions on possible causes? Do these inputs have lesser ESD-
protection? The cables used are shielded RJ45 TP-cables (ethernet
cables) and I think the shield touches the connector before the
signals, thus grounding away any accumulated potentials during
assembly.

I guess we'll put on external protection for the next board-spin, but
I would just like some hints on if this is a common problem with these
devices (if external protection is not used).

Best regards,
Bjorn W

Hi,

funny, we have a very similar setup here and also fried an LVDS-input
on a prototype... (Now, the n-input of the pair is low impedance to
ground). We came to the same conclusion (add ESD-protection in the re-
design)... So I am curios if there is an answer to your question.

Regards,

Thomas

www.entner-electronics.com

On 11/17/2010 6:26 PM, BW wrote:
> Hi!
>
> We have a design where an Altera Cyclone-3 (EP3C5) with LVDS inputs is
> connected to a sender on another board through a cable.
>
> In a number of cases on our prototype boards, the LVDS-inputs have
> been fried from this setup.
>
> Apart from any misdesign in the sender-board, does anyone have any
> suggestions on possible causes? Do these inputs have lesser ESD-
> protection? The cables used are shielded RJ45 TP-cables (ethernet
> cables) and I think the shield touches the connector before the
> signals, thus grounding away any accumulated potentials during
> assembly.
>
> I guess we'll put on external protection for the next board-spin, but
> I would just like some hints on if this is a common problem with these
> devices (if external protection is not used).
>
> Best regards,
> Bjorn W
>

I can imagine situations where a difference in ground levels between the 
two modules would cause trouble.  How are the grounds of the two modules 
related?  Are the modules powered from the same supply? (Like two boards 
in the same backplane, for example.)  If you disconnect the data cable 
between the two and connect a multimeter in current mode between the 
grounds is there any current flowing?

Probably not the issue, but easy to check.

Chris Abele