Hot plugging - what can happen and how to protect??

[Roque]

If I want to make sure a circuit is protected against hot plugging, what do
I do? changing the length of power pins isn't an option so I'm thinking
along the lines of a solution with series limiting elements for any I/O
lines.

This isn't an address/bus type of situation, it's more like discreet I/O
lines with transistor or buffer gate drivers chips. I have TVS diodes on
the inputs for ESD protection on the 5 volt logic, and I'm afraid they may
blow up during hot plugging. The system has 5 volt and 24 volt power, along
with the I/O on the connector in question.

What CAN go wrong in such a circuit when hot plugged, I don't really
understand what it is that could happen if certain pins make contact before
others...that's why I don't know how to design for it..

Thanks

 

[N. Thornton]

If I want to make sure a circuit is protected against hot plugging, what do
I do? changing the length of power pins isn't an option so I'm thinking
along the lines of a solution with series limiting elements for any I/O
lines.

This isn't an address/bus type of situation, it's more like discreet I/O
lines with transistor or buffer gate drivers chips. I have TVS diodes on
the inputs for ESD protection on the 5 volt logic, and I'm afraid they may
blow up during hot plugging. The system has 5 volt and 24 volt power, along
with the I/O on the connector in question.

What CAN go wrong in such a circuit when hot plugged, I don't really
understand what it is that could happen if certain pins make contact before
others...that's why I don't know how to design for it..

Thanks

Hi


your risks are:
any voltage on the connector getting onto any other pin
static discharge to any pin
with unearthed equipment, you may get mains on the pins via the
external unit's mains to chassis caacitance and any leakage.

R in series with each pin, Z across pin/chassis, and diodes to rails
will generally take care of all this I believe. But I'd wait for
confirmation or not on that before going ahead.

Regards, NT

 

[Roque]

[email protected] (N. Thornton) wrote in

your risks are:
any voltage on the connector getting onto any other pin
static discharge to any pin
with unearthed equipment, you may get mains on the pins via the
external unit's mains to chassis caacitance and any leakage.

R in series with each pin, Z across pin/chassis, and diodes to rails
will generally take care of all this I believe. But I'd wait for
confirmation or not on that before going ahead.

Thanks for the info...it all makes sense to the bits and pieces of general
protections I have seen but didn't know how to justify.

I have one question, just by hot plugging, how can voltage from another pin
far away on the connector, like 24volts, end up on my 5volt logic input
located on the far side of the connector? If I could explain that one I
think I'd have all my puzzle solved.
Thanks again

 

[MarkyMark]

N
In my experience the major problem occurs when the card mates at a slight
angle, such that the ground is one of the last to connect for a brief time.
During this time the supply rail is connected and power finds its way around
the board via any gates (usually with input protection diodes.... CMOS etc)
that mate prior to main ground, particularly bus drivers and deliberately
grounded backplane pins (card ID etc). Major current temporarily flows to
ground via this mechanism rather than to the ground plane/rail and take out
the driver or gate/s.

As you said extended power/ground pins are the traditional method of
overcoming this, but not possible for your problem. Given that, I suggest
ground pins at either end (borrow spare ones if any) to give more chance of
a true ground during power up. Failing that, intercept the 24V rail with a
decent P channel FET - Source to 24V input, Drain to 24V rail on the board
and gate to the genuine ground (via a divider to limit the gate drive for
24V input). This will prevent power up until the card mates completely.

Hope this helps
Regards Mark

 

[Jim Thompson]

If I want to make sure a circuit is protected against hot plugging, what do
I do? changing the length of power pins isn't an option so I'm thinking
along the lines of a solution with series limiting elements for any I/O
lines.

This isn't an address/bus type of situation, it's more like discreet I/O
lines with transistor or buffer gate drivers chips. I have TVS diodes on
the inputs for ESD protection on the 5 volt logic, and I'm afraid they may
blow up during hot plugging. The system has 5 volt and 24 volt power, along
with the I/O on the connector in question.

What CAN go wrong in such a circuit when hot plugged, I don't really
understand what it is that could happen if certain pins make contact before
others...that's why I don't know how to design for it..

Thanks

Do a little research into "Hot Plug" Controllers. This is a
well-documented problem area. LT makes lots of chips to alleviate
this problem.

...Jim Thompson

 

[N. Thornton]

[email protected] (N. Thornton) wrote in


Thanks for the info...it all makes sense to the bits and pieces of general
protections I have seen but didn't know how to justify.

I have one question, just by hot plugging, how can voltage from another pin
far away on the connector, like 24volts, end up on my 5volt logic input
located on the far side of the connector? If I could explain that one I
think I'd have all my puzzle solved.
Thanks again

No problem. Consider just 4 pins,
1: 0v,
2: +24v,
3: a TTL logic level, at 0v
4: an analog signal currently at 24v.

Now, when you plug in, lets say pin 4 connects first. The remote
device's analog line is at +24v, and the near device's input on pin 4
has a resistor to ground, so it sits at 0v relative to near-device
chassis. Because there is no circuit completed, ie no ground
connected, no current flows and the Vs stay unchanged. The V on pin 4
is seen as 0v by the near device, and as 24v by the far device. I hope
that's clear-ish.

Now pin 3 connects. Still no ground connection. The remote device's
chassis is 24v lower than the near device's chassis, due to the pin 4
connection. The pin 3 voltage is 0v relative to remote device chassis,
but of course that chassis is -24v relative to near chassis. So the
near unit sees -24v on its TTL line, pin 3. This is how your power
voltages get onto the signal pins.


Regards, NT

 

[Roque]

[email protected] (N. Thornton) wrote in

1: 0v,
2: +24v,
3: a TTL logic level, at 0v
4: an analog signal currently at 24v.

I think I got it now, with one more question, does this all hinge on the
fact that the analog 24v on pin 4 has the resistance to ground in the near
chasis, which is what causes the near chasis to be sitting at 24v with just
pin 4 connected, which then puts 24v on the ground for pin 3 making
contact, ending up in -24v on TTL?

I sat on it for 10minutes and finally saw that picture with that resistance
on pin 4 to chassis. Do we assume that this resistance or other sort of
path will exist in all or most circuits, from 24v analog pin to chassis,
when there is no ground? If so, then I have it all figured out now...and
thanks again!

 

[Robert Baer]

[email protected] (N. Thornton) wrote in


I think I got it now, with one more question, does this all hinge on the
fact that the analog 24v on pin 4 has the resistance to ground in the near
chasis, which is what causes the near chasis to be sitting at 24v with just
pin 4 connected, which then puts 24v on the ground for pin 3 making
contact, ending up in -24v on TTL?

I sat on it for 10minutes and finally saw that picture with that resistance
on pin 4 to chassis. Do we assume that this resistance or other sort of
path will exist in all or most circuits, from 24v analog pin to chassis,
when there is no ground? If so, then I have it all figured out now...and
thanks again!

Sounds like at least a protection diode is in order...

 

[N. Thornton]

[email protected] (N. Thornton) wrote in

I think I got it now, with one more question, does this all hinge on the
fact that the analog 24v on pin 4 has the resistance to ground in the near
chasis,

for this specific example yes, but IRL no because there will be
numerous paths all over the place in practice.

which is what causes the near chasis to be sitting at 24v with just
pin 4 connected, which then puts 24v on the ground for pin 3 making
contact, ending up in -24v on TTL?

I think you're not defining things with clarity here. All voltages are
relative, and in this case they're all relative to 2 different
chassis. Hopefully if you redo what you said with this level of
complexity/clarity, full clarity should be there.

I sat on it for 10minutes and finally saw that picture with that resistance
on pin 4 to chassis. Do we assume that this resistance or other sort of
path will exist in all or most circuits,

yes, most.

from 24v analog pin to chassis,
when there is no ground? If so, then I have it all figured out now...and
thanks again!

Tc, NT