MUX508: intermittent short or abnormal channel behavior

Hi Kelvin,

There shouldn't be a condition that causes S8 to latch to the drain. Can you check the control pins when this happens? Wondering if A0,A1,A2 pull high momentarily.

How long is this period of time? And does the S8 path then connect directly and short to the other selected channels?

Thanks,
Rami

Hi

Thanks for the reply.

I’ve reviewed the truth table of the MUX508 and understand that it shouldn’t be possible for more than one channel to be active at the same time.

In my case, I’m cycling through the inputs using A0–A2 in order:
S1 → S2 → S3 → … → S8 → S1 …, switching every 1 millisecond.

I probed S8 with an oscilloscope.

• During normal operation, I observe the expected signals on S8 — everything looks correct and stable.

• However, during the abnormal condition, I see unexpected signals on S8 — it seems to reflect the signals of the currently selected channels in sequence.

This abnormal behavior can persist for several minutes before recovering by itself.

Importantly, S8 does not short to the other Sx lines in my circuit. It just seems to stay connected to D, even when the MUX has moved on to other channels.

That’s why I suspect some kind of internal short or conduction fault on the S8 switch path.

So my question is:
Could an external pulse, surge, or ESD cause an internal short like this in the MUX?

Or are there other possible root causes we should consider?

Thanks again!

Hi Kevlin,

I'm a bit confused though how S8 is staying connected but isn't connecting to the other signal line that is connected. They should both be connected to the drain pin, effectively shorting them there. 
That being said, I've never seen behavior like this. How are you verifying the S8 is always connected. It looks like you're scoping S8 but what behavior makes you think that it stays connected? Should you not be probing the drain instead since it's the output? 
There certainly could be some internal fault or esd damage that could damage the S8 FET. How repeatable is this behavior showing up? As in across multiple devices do you see this, or only one, or a small handful?

Thanks,
Rami

Hi

By MUX design, only one channel should be connected to the drain at any given time.

The waveform below was captured directly from the S8 pin.

During the abnormal condition, I observed signals from other channels appearing on S8, even though those channels were actively selected and S8 should have been disconnected.

This behavior suggests that S8 remains internally connected to the drain, even when it is not selected.

That’s why I suspect an internal short or conduction fault on the S8 switch path to the drain.

I also suspect that this issue might be caused by ESD or surge, possibly creating a recoverable fault in a small number of devices.  

Hello Kelvin,

Rami is out of office today.
Please expect a response early next week.

Thanks,

Nir 

Kevlin,

Kelvin Lai said:

By MUX design, only one channel should be connected to the drain at any given time.

Yes, that is the point I am making. Only one channel should be connected. But if 2 are selected they should be shorted together. Even in your images you show this:

Your abnormal conditions doesn't show to me that they're shorted. Where are you drawing the conclusion that they're shorted from? 

Can you share actual screenshots? To me it looks like your logic is switching when it shouldn't be switching. Maybe I'm missing something but your normal condition looks like it's always signal 8. But your abnormal looks like it's switching between states. Are these images mixed around maybe? Or can you explain why you see the abnormal and get to the conclusion that 8 is always on?

-Rami 

Thanks again for your reply and for reviewing the waveform captures.

Let me clarify a few points to avoid confusion:

• Only one channel is selected at a time by the logic (A0–A2). I am not intentionally selecting two channels simultaneously.

• During normal operation, S8 shows a signal only when it is selected, as expected.

• During the abnormal condition, even when other channels (e.g., S1, S2, etc.) are selected, I can still observe their signals (S1, S2, etc.) appearing on the S8 pin using an oscilloscope. This strongly suggests that S8 remains internally connected to the drain, even when it is not selected.

• I'm not suggesting that multiple channels are simultaneously selected and shorted. Rather, it appears that S8 is stuck on, remaining connected to the drain while the logic control continues switching through other channels.

This behavior does not occur during normal operation and sometimes persists for several minutes before recovering on its own.

Hi Kelvin,

I understand what the situation is but your conclusion doesn't make sense. If S8 is stuck on and you are still cycling through other channels, it means that S8 is shorted to other channels. S8 isn't stuck on according to the pictures you sent. There is only 1 drain that all the channels share. If you see the other channels on the drain, and they look as expected, then S8 isn't also selected. Rather S8 likely is getting disconnected, and your logic is selecting other channels. is it possible to get actual probe shots of the control pins and the drain while this event happens? 

Thanks,
Rami

Thanks again for your patience.

In simple terms, I can observe the signal from S1 appearing on the S8 pin.
That’s why I suspect that S8 is still internally connected to the drain — in other words, S8 might be stuck on.

Hi Kelvin,

Oh I see, you're probing on the S8 not on the drain. Sorry for my oversight on that.
That's interesting and I've never seen that occur. How repeatable is it? As in is this the only occurrence of this behavior and have you tried replacing it with a new device? Could very well be ESD damage or some form of EOS. But it's strange that it seems to happen after some time and then fixes itself.

Thanks,
Rami

Thanks for your support.

While continuing to investigate, I came across this TI FAQ about latch-up in CMOS devices, which seems highly relevant to the issue I’m facing.

Given that MUX508 is CMOS-based, and the symptoms I'm seeing involve a channel that behaves as if it's conducting when not selected — and the issue is intermittent and self-recovering — this sounds consistent with the kind of latch-up or internal conduction path described in the article.

Would you agree that this is a possible failure mode?
Do you recommend switching to one of TI’s latch-up immune multiplexer families for better robustness?

Hi Kelvin,

It seems similar in some ways but not really the same thing here. If you're switch channel is always on then there is likely either damage on the gate driver or the gate oxide itself. If it was latch up the path would be from VDD to GND and it could impact your switching but more because you likely wouldn't be able to switch your channels at all.
You can check the supply current during the event to offer us a hint of what is happening.
I'm still looking an answer to if the results are repeatable across multiple MUX508 devices or can we resolve this with replacing the device for a new IC.
Scope shots of your S8 and D with the control logic would be helpful here too.

-Rami