keyboard doesn't work, then does

[mm]

Keyboard doesn't work, then does WinXP

Hi,

I don't know what group would be best for this, but it's hardware
related so I chose you. (I hope you feel honored.)

Yesterday was a heavy rainstorm, no thunder that I recall, and no
lighting that I saw out the window, but two very short power failures.
One second each.

I'm temporarily without a UPS.

Everything else I've tried to use works, tv, radio, cordless
telephone, speakers, fan, and in the other room DVDR, RF mod, TV,
radio. The breaker in the bathroom tripped, a GFI breaker in the
basement, but I'll bet the tv and radio connected to that still work
fine.

After each failure, the green light on the cmputer was flashing and
neither pushing the reset button or the on/off button (for a moment or
two) did anything. So I unplugged the cmputer, waited until the
light stopped flashing, which took at least 5 seconds (why?) and
plugged it in again.

******The second time, the PS-2 keyboard didn't work. No key afaict
and in no program. I plugged in a USB keyboard and continued. After
an hour or so I exited winXP, which turned off the computer.

Later I restarted and again the PS-2 connected keyboard didn't work
but the USB one did. (Unfortuately, the group of 6 keys are arranged
differently. ugh.)

Device manager said there was no problem with the keyboard.

Later I exited and then started again and the third time the keyboard
worked fine, everywhere, and has for the last 5 hours elapsed time.

So what's going on? It seems like the keyboard is far away
electrically from the power cord and the DSL line, too far to get
damaged without the mobo being damaged, and anyhow, if it were
damaged, how come it's working now?

Any ideas? <sheepish grin>

 

[Paul]

Keyboard doesn't work, then does WinXP

Hi,

I don't know what group would be best for this, but it's hardware
related so I chose you. (I hope you feel honored.)

Yesterday was a heavy rainstorm, no thunder that I recall, and no
lighting that I saw out the window, but two very short power failures.
One second each.

I'm temporarily without a UPS.

Everything else I've tried to use works, tv, radio, cordless
telephone, speakers, fan, and in the other room DVDR, RF mod, TV,
radio. The breaker in the bathroom tripped, a GFI breaker in the
basement, but I'll bet the tv and radio connected to that still work
fine.

After each failure, the green light on the cmputer was flashing and
neither pushing the reset button or the on/off button (for a moment or
two) did anything. So I unplugged the cmputer, waited until the
light stopped flashing, which took at least 5 seconds (why?) and
plugged it in again.

******The second time, the PS-2 keyboard didn't work. No key afaict
and in no program. I plugged in a USB keyboard and continued. After
an hour or so I exited winXP, which turned off the computer.

Later I restarted and again the PS-2 connected keyboard didn't work
but the USB one did. (Unfortuately, the group of 6 keys are arranged
differently. ugh.)

Device manager said there was no problem with the keyboard.

Later I exited and then started again and the third time the keyboard
worked fine, everywhere, and has for the last 5 hours elapsed time.

So what's going on? It seems like the keyboard is far away
electrically from the power cord and the DSL line, too far to get
damaged without the mobo being damaged, and anyhow, if it were
damaged, how come it's working now?

Any ideas? <sheepish grin>

I'll try the easy one first.

"So I unplugged the computer, waited until the light stopped flashing,
which took at least 5 seconds (why?) and plugged it in again."

The light could be tied to +5VSB. The power supply has a large capacitor
on the primary side, which holds a lot of energy. Part of the function
of that capacitor, is to provide "hold up time" during a UPS switching
event. The capacitor is guaranteed to hold up the supply for 16 milliseconds
at full power supply load. In cases where only the +5VSB part of the
supply is running, and the load on that rail isn't really that significant,
the circuit can run for 30 seconds or longer, using juice stored in the
capacitor. For more info, you can examine this schematic for an ATX PSU
(C5 and C6 are the ones storing the energy - R2 and R3 are bleeder resistors,
for making the capacitors "safe" when the unit is unplugged and opened up.
Don't touch the terminals on the bottom of C5 and C6... ).

http://www.pavouk.org/hw/en_atxps.html

*******

Theorizing about your keyboard, would require the "purest speculation".
I'm not sure I can make up a satisfying hypothesis for you. But I'll try
relating a story.

As a new designer at my first employer, I built a circuit, one of many
modules used in a tall piece of equipment. I was getting complaints that
my circuit was freezing (with some small probability of it happening).
Most of the time, you'd plug my module in and it worked. But occasionally
it would freeze and do nothing.

I got a sample unit on my bench, and decided it would be fun to build an
automated torture machine. I built a series of timers and relays, to turn
the equipment on and off every 15 seconds, and evaluate whether it was
working or not. The idea was, I could go home, and leave the equipment doing
the tests all by itself.

Sure enough, I caught the chip on the module, freezing. Power was there,
but the chip would not do anything.

I happened to stick the probe of my storage scope on the supply rail. The chip
was supposed to run from a 5V supply. The scope recorded an 8 volt high transient
when the power was turned on. Bingo! The high voltage of that transient,
correlated with the freezing.

It turned out, that as in every large engineering project, common infrastructure
is designed by one engineer, and "given" to others. In this case, a filter
network had been placed on *every* circuit pack (for reasons that were never
fully explained). That filter was a "thing of its generation". I'd seen
it used in some equipment I tore apart as a student, so I'd seen it before.
It was an LC entry filter.

My memory is foggy now, but I think my quick fix was to place a Zener after
the filter, to clamp the amplitude of the overshoot, when the power comes
on. No more freezes. I would have been happier to remove the damn filter,
but the politics of the situation didn't make that possible.

Something like that could happen to your keyboard, but then we'd have the
messy detail, of trying to explain where a sharp-edged power transient would be
coming from on the 5V feeding the keyboard. That is the part of the hypothesis
that doesn't make sense. We'd also need to open the keyboard, and see whether
an ancient LC filter was at play or not.

I doubt you're going to have this happen again, so it would be
pretty hard to prove or disprove what happened. At minimum, anyone
studying this, would need to start by opening up the keyboard,
and seeing how it was designed.

Another mechanism to "freeze" a circuit, is latchup (or micro-latchup).
I've seen micro-latchup first hand, and at 4PM in the afternoon, I
thought my circuit was dead. Turned off the power and went home. At
8AM the next morning, I turned on the power and it was working again.
I couldn't believe it, because I thought I'd fried an expensive
and valuable chip (we only had 25 prototypes at the time). There was a
solid technical reason for that happening, and I had a solution in hand
in a day or two, to stop it. But finding a solution wouldn't have happened,
without noting that the chip recovered from its failure. That was the
clue I needed, to focus my search in the right direction. In the case
of your keyboard, I'd be hard pressed to relate such a mechanism to
the keyboard controller. But it is another phenomenon that can freeze
circuits, and requires that all the stray charge be drained out of the
circuit before it will recover.

http://en.wikipedia.org/wiki/Latchup

Paul

 

[mm]

I'll try the easy one first.

"So I unplugged the computer, waited until the light stopped flashing,
which took at least 5 seconds (why?) and plugged it in again."

The light could be tied to +5VSB. The power supply has a large capacitor
on the primary side, which holds a lot of energy. Part of the function
of that capacitor, is to provide "hold up time" during a UPS switching
event. The capacitor is guaranteed to hold up the supply for 16 milliseconds
at full power supply load. In cases where only the +5VSB part of the
supply is running, and the load on that rail isn't really that significant,
the circuit can run for 30 seconds or longer, using juice stored in the
capacitor. For more info, you can examine this schematic for an ATX PSU
(C5 and C6 are the ones storing the energy - R2 and R3 are bleeder resistors,
for making the capacitors "safe" when the unit is unplugged and opened up.
Don't touch the terminals on the bottom of C5 and C6... ).

This makes sense.

http://www.pavouk.org/hw/en_atxps.html
Thanks.

*******

Theorizing about your keyboard, would require the "purest speculation".

Isn't that the best kind of speculation? IIRC it's the best kind of
orange juice.

I'm not sure I can make up a satisfying hypothesis for you. But I'll try
relating a story.

As a new designer at my first employer, I built a circuit, one of many
modules used in a tall piece of equipment. I was getting complaints that
my circuit was freezing (with some small probability of it happening).
Most of the time, you'd plug my module in and it worked. But occasionally
it would freeze and do nothing.

I got a sample unit on my bench, and decided it would be fun to build an
automated torture machine. I built a series of timers and relays, to turn
the equipment on and off every 15 seconds, and evaluate whether it was
working or not. The idea was, I could go home, and leave the equipment doing
the tests all by itself.

Sure enough, I caught the chip on the module, freezing. Power was there,
but the chip would not do anything.

I happened to stick the probe of my storage scope on the supply rail. The chip
was supposed to run from a 5V supply. The scope recorded an 8 volt high transient
when the power was turned on. Bingo! The high voltage of that transient,
correlated with the freezing.

It turned out, that as in every large engineering project, common infrastructure
is designed by one engineer, and "given" to others. In this case, a filter
network had been placed on *every* circuit pack (for reasons that were never
fully explained). That filter was a "thing of its generation". I'd seen
it used in some equipment I tore apart as a student, so I'd seen it before.
It was an LC entry filter.

My memory is foggy now, but I think my quick fix was to place a Zener after
the filter, to clamp the amplitude of the overshoot, when the power comes
on. No more freezes. I would have been happier to remove the damn filter,
but the politics of the situation didn't make that possible.

Something like that could happen to your keyboard, but then we'd have the
messy detail, of trying to explain where a sharp-edged power transient would be
coming from on the 5V feeding the keyboard. That is the part of the hypothesis
that doesn't make sense. We'd also need to open the keyboard, and see whether
an ancient LC filter was at play or not.

I doubt you're going to have this happen again, so it would be
pretty hard to prove or disprove what happened. At minimum, anyone
studying this, would need to start by opening up the keyboard,
and seeing how it was designed.

Another mechanism to "freeze" a circuit, is latchup (or micro-latchup).
I've seen micro-latchup first hand, and at 4PM in the afternoon, I
thought my circuit was dead. Turned off the power and went home. At
8AM the next morning, I turned on the power and it was working again.
I couldn't believe it, because I thought I'd fried an expensive
and valuable chip (we only had 25 prototypes at the time). There was a
solid technical reason for that happening, and I had a solution in hand
in a day or two, to stop it. But finding a solution wouldn't have happened,
without noting that the chip recovered from its failure. That was the
clue I needed, to focus my search in the right direction. In the case
of your keyboard, I'd be hard pressed to relate such a mechanism to
the keyboard controller. But it is another phenomenon that can freeze
circuits, and requires that all the stray charge be drained out of the
circuit before it will recover.

http://en.wikipedia.org/wiki/Latchup

Thanks for all this. very interesting. I used to buy keyboards at
hamfests for a dollar each, but now I like the ones with volume
control, etc. and they are two dollars each when I find them. So this
is just a curiosity, but I'm a very curious guy. I'll read the
webpages you give too.

FTR, it's a Microsoft keyboard with 16 extra buttons at the top. I
don't see a model number.

Thanks, Philo, also.

 

[mm]

Well, the keyboard that didn't work right has been working fine for
two weeks now. Except all of a sudden for a five-minute period where
the shift seemed to be on all the time, even for characters like
period and comma, which came out like < and >. Some keys didn't work
at all. I plugged in the USB keyboard and it worked like that too.

But after a couple minutes of pressing keys and trying different
programs, it went back to normal, and that was 5 days ago.

Life is strange.

Thanks again for your help.