What is the impact of having some VSS pins not connected to ground?

[GePa_4776901]

In one board we have a BGA PSoC 6, and three of the VSS pins (ground) are floating, and since it is a BGA, we are out of luck... I want to know the impact.

In the datasheet I can only find which ports are connected to which VDD signals.. but I cannot find any information on which of the VSS affect which ports.

It's a CY8C62x6/7, 124-BGA package and we only have pins B12,C3,D4 connected to ground... pins D10,K4,K10 are floating...

We see some unpredictable behavior in some outputs, but I don't have a clear image of the impact.

Is there a diagram or datasheet with that information?

Thanks!

[DaBl_4548126 · Accepted solution]

Assuming the VSS balls are all tied together (usually true, you can verify with a part not soldered down), the main effect is that you will get a lot more ground bounce due to higher inductance in the return path. As a possible workaround, program all your outputs to the lowest drive strength and slowest switching speed (slew rate) as you can, and try to avoid outputs simultaneously switching - stagger them if possible. Even staggering them by a gate delay will help.

If you have an output in the vicinity of the non-connected balls that can be held low, put a scope on it and see how high it might be bouncing. High ground bounce can cause false clocking of registers, and can make inputs that should be high, be read as low.

If you have bare boards, it might be possible to have a blind, laser-drilled via put into those unconnected pads, to connect them to a ground plane (if you have a ground plane one or two layers below the surface).

Slowing down the PSOC clock frequency might help some, although this is mostly a "switching edge" problem.

Otherwise, just do as much testing and debug of your other circuitry as you can before you respin the PWB to fix this.

[GePa_4776901]

Thank you for your answer. Three of the six Vss pins are floating, not because of the BGA solder, but a board error... they are simply not connected.

My question is if each Vss pin is for a particular function or a particular port, or if they are internally connected.

I've only found one output not working so far, and one that works intermittently but I don't know if there are more pins affected, or if it will only affect outputs but no inputs.. that sort of thing. The datasheet only mentions that they should all be connected to gnd.

[DaBl_4548126]

Yes, that's what I assumed. Only Cypress can give exact details, but I suspect they may just say "Don't leave them floating, it will cause all kinds of bad things." (And I can see why they might say that. They have never tested the chip with those pins floating, so they would be reluctant to tell you something, and then it could be wrong, because you could sue them and say "but you said it would work!"  YOU might not do that, but some people would, so that's why some companies just say "Don't do that.")

But as I said, if you have an unsoldered part, you can measure those balls with an ohmmeter and see if they are connected to the other VSS balls. If they are not, you really are out of luck, as all kinds of misbehavior are possible. It is uncommon (but not unheard of) to have some VSS pins isolated from others, and be tied together on the PWB. It is usually done on chips like switching regulators or other analog parts, where they want to keep switching noise out of the more sensitive analog circuitry. (Although then they usually tell you in the data sheet, "these pins are not connected on the die, and MUST be tied together on the PWB.")

If you can't do a laser-drilled via, there are companies that can add traces to a bare PWB (search for "additive traces"), I think it is a silk-screen process using conductive ink. It might be possible that this could solve your problem, at least to allow you to finish your development. I would still respin the PWB before production, because this is still not as good as a via directly to the ground plane. Good luck.