Why will BAV99 not clamp to Vcc?

[MikeMl]

As a retired CMOS IC Chip designer, I can assure you that all CMOS makers try to design their input pad protection devices to minimize latch-up, but, that is the goal, not the reality.

It is important to look at the publication dates of AN521 (1997) and TB3009 (2008). I suspect that Microchip started getting some complaints about their devices latching up, and published TB3009 to cover their backsides.

Also note that AN521 talks about injecting a few tens of uA into a PIN; not 10s of mA!!!!!

 

[audioguru]

As a retired CMOS IC Chip designer, I can assure you that all CMOS makers try to design their input pad protection devices to minimize latch-up, but, that is the goal, not the reality.

Good. Then do you recommend that we clamp the input with a Schottky diode?

 

[MikeMl]

Good. Then do you recommend that we clamp the input with a Schottky diode?

It is a matter of degree. I would begin to worry if >10mA is ever injected into a CMOS input. If you are hooking a CMOS input to a CMOS output (powered by the same power supply), no problem. If you have a CMOS input that is returned to a power supply that can be On before the CMOS supply is, then you need to consider what might happen. If a CMOS input comes from an external interface that you touch (like an keypad), then I would always have some sort of series resistor with shunt diodes.

 

[Nigel Goodwin]

As a retired CMOS IC Chip designer, I can assure you that all CMOS makers try to design their input pad protection devices to minimize latch-up, but, that is the goal, not the reality.

Have you ever heard of any such problem on a PIC?.

It is important to look at the publication dates of AN521 (1997) and TB3009 (2008). I suspect that Microchip started getting some complaints about their devices latching up, and published TB3009 to cover their backsides.

That's not the only one, it's just one I remembered off hand, there are many similar examples - it's also NOT a 1997 Application note, it's FAR older than that, it was updated in 97, hence the 'modern' date on it. If I remember correctly, it probably pre-dated Windows, and certainly pre-dated Windows PIC Programmers (as I wrote the first one - well before MicroChip even had any Windows tools).

Also note that AN521 talks about injecting a few tens of uA into a PIN; not 10s of mA!!!!!

I've never suggested injecting high currents - if you're not careful the PIC will be powered from the input signal, rather than it's supply rail.