Maker Pro
Maker Pro

Computer won't post with my favorite keyboard plugged in.

M

mike

The last "new" computer I bought was in 1989, so when I saw "new" I was
thinking it was actually a used computer new to the poster.

No, bypassing whatever current limiting there might be between the +5v
line and the keyboard is likely going to cause issues with warranty. Of
course, that only matters if you need the warranty, ie the computer
fails and you want to send it back.

Michael
Well, I built the fixture and tested it.
Keyboard draws 180 mA.
New keyboard is less than 10mA.
BIOS date is 2008. Probably the result of some
GREEN initiative to save the world from abusive
keyboard power waste.

I really don't want to work on the motherboard.
It's already installed and a logistic nightmare.

I'm thinking about leaving the 2.2K pullup resistors
connected to the motherboard 5V and running everything else
in the keyboard
off a wall wart. I think that reduces the risk of
anything in the computer latching up and smoking when the computer or
keyboard is powered off.

That sound rational?
 
P

petrus bitbyter

mike said:
Well, I built the fixture and tested it.
Keyboard draws 180 mA.
New keyboard is less than 10mA.
BIOS date is 2008. Probably the result of some
GREEN initiative to save the world from abusive
keyboard power waste.

I really don't want to work on the motherboard.
It's already installed and a logistic nightmare.

I'm thinking about leaving the 2.2K pullup resistors
connected to the motherboard 5V and running everything else
in the keyboard
off a wall wart. I think that reduces the risk of
anything in the computer latching up and smoking when the computer or
keyboard is powered off.

That sound rational?

Er... No. So your old keyboard requires 180mA. Did it work correctly on the
old machine that way? If so, can your new machine provide that 180mA? *If*
not, *then* your solution makes sense *otherwise* its only a maybe.

From your information it is not clear on what conditions the keyboard worked
all these years. If 5V/180mA does the trick and your new machine can provide
it, lack of power is not the main problem. The 5V/180mA is not all that bad.
In the old days the keyboards often were protected by a 1A fuse.

To check it out, you first connect the old keyboard to an old machine. You
will find the required current. Suppose that's the 180mA. Now connect your
new keyboard to the new machine. The 5V/10mA should do. Disconnect the new
one and connect the old one. Now if the machine cannot provide the power
required you should find the 5V dropping. If not, the machine does not seem
to have a problem providing the required current.

But if you new machine cannot provide enough power, your method is not 100%
save. Problems may arise when you shut down the computer while the wallwart
is still going on. I ever found printer ports blown when the computer was
powered down before the printer. So to stay on the safe side I'd use the
10mA from the machine to active a small relay that switches the wallwart
power to the keyboard. Suppose there's plenty of space in the old keyboard.

petrus bitbyter
 
M

mike

Er... No. So your old keyboard requires 180mA. Did it work correctly on the
old machine that way? If so, can your new machine provide that 180mA? *If*
not, *then* your solution makes sense *otherwise* its only a maybe.

From your information it is not clear on what conditions the keyboard worked
all these years. If 5V/180mA does the trick and your new machine can provide
it, lack of power is not the main problem. The 5V/180mA is not all that bad.
In the old days the keyboards often were protected by a 1A fuse.

To check it out, you first connect the old keyboard to an old machine. You
will find the required current. Suppose that's the 180mA. Now connect your
new keyboard to the new machine. The 5V/10mA should do. Disconnect the new
one and connect the old one. Now if the machine cannot provide the power
required you should find the 5V dropping. If not, the machine does not seem
to have a problem providing the required current.

But if you new machine cannot provide enough power, your method is not 100%
save. Problems may arise when you shut down the computer while the wallwart
is still going on. I ever found printer ports blown when the computer was
powered down before the printer. So to stay on the safe side I'd use the
10mA from the machine to active a small relay that switches the wallwart
power to the keyboard. Suppose there's plenty of space in the old keyboard.

petrus bitbyter
INteresting...
I follow and agree with your troubleshooting procedure.
But, I have a guiding principle...never ask a question if the answer
won't change the future.
I already know the old keyboard won't work on the new computer.
Plugging it in again just overloads the 5V again. Possible harm, no help.

I believe I have only one option...power the keyboard externally.
That either fixes it or doesn't. That experiment is far easier than
disassembling the machine and trying to reverse-engineer the keyboard
interface.
The risk/reward ratio for messing with the motherboard is unfavorable.

I share your concern about backfeed and originally considered the relay.
But, the only things attached to the computer are two 7406
open-collector outputs and 2.2K pullups.
The third is the reset into a 7406 input. I can probably disconnect that.

If I leave the pullups connected to the computer power, there's no
path from the wall wart power back into the computer????
Solves the problem? yes? no?

As soon as the portable scope charges up, I should have an answer to whether
it works.
 
P

petrus bitbyter

mike said:
INteresting...
I follow and agree with your troubleshooting procedure.
But, I have a guiding principle...never ask a question if the answer
won't change the future.
I already know the old keyboard won't work on the new computer.
Plugging it in again just overloads the 5V again. Possible harm, no help.

I believe I have only one option...power the keyboard externally.
That either fixes it or doesn't. That experiment is far easier than
disassembling the machine and trying to reverse-engineer the keyboard
interface.
The risk/reward ratio for messing with the motherboard is unfavorable.

I share your concern about backfeed and originally considered the relay.
But, the only things attached to the computer are two 7406 open-collector
outputs and 2.2K pullups.
The third is the reset into a 7406 input. I can probably disconnect that.

If I leave the pullups connected to the computer power, there's no
path from the wall wart power back into the computer????
Solves the problem? yes? no?

As soon as the portable scope charges up, I should have an answer to
whether
it works.

Guess you missed my point. So how do you know you overloaded the 5V? Of
course we already know the keybord did not work. But did the 5V drop when
the keyboard was connected? I not, the new machine looks like to be able to
provide that 180mA.

As for the interface, I ever got a printer interface that was blown by just
the 7406 you mentioned. It was not a single accident but structural for that
series of PCs. The (CMOS) chip involved could not handle the 5V on the I/O
when the chip got no power. We may hope todays interfaces are able to handle
this but I'd not be too sure. So yes, there is a path from the wallwart into
the computer and it is through the pullups.

A second path may exist through the keyboards inputs. These inputs may be
74LS14 often used along with 7406 outputs. If so, the computers outputs will
"get" 5V even when not powered.

While working these outputs should be able to sink 0.4mA plus the current
through the pullups. So if the output is driven low but cannot sink enough
current, it may not go low enough to make the keyboard see it as a low.
(FAIK this is a possible cause for the problem not mentioned so far.)

Using a scope you'd be able to figure out not only the power levels but also
the shapes of the signals. I'd like to hear the results.

petrus bitbyter
 
M

mike

Guess you missed my point. So how do you know you overloaded the 5V?

Actually, I don't. But if that's not the problem, I don't have any
other options. I can't do anything about the signal levels or the timing.

I just realized another possible option. As I recall, the keyboard
runs off the "always-on" auxiliary 5V supply. If that's got a problem,
that would explain why dragging down the keyboard power makes it refuse
to boot. That's relatively easy to check.
Of
course we already know the keybord did not work. But did the 5V drop when
the keyboard was connected? I not, the new machine looks like to be able to
provide that 180mA.

As for the interface, I ever got a printer interface that was blown by just
the 7406 you mentioned. It was not a single accident but structural for that
series of PCs. The (CMOS) chip involved could not handle the 5V on the I/O
when the chip got no power. We may hope todays interfaces are able to handle
this but I'd not be too sure. So yes, there is a path from the wallwart into
the computer and it is through the pullups.

A second path may exist through the keyboards inputs. These inputs may be
74LS14 often used along with 7406 outputs. If so, the computers outputs will
"get" 5V even when not powered.

I tested it on a different computer.
The diode isolates the 5V, but the data line does have an input on it.
Worst case back into the computer is 1.6V. Short circuit currrent to ground
is under 2mA.
Keyboard works fine with or without the wall wart.
Now, to test it on the problem computer.

While working these outputs should be able to sink 0.4mA plus the current
through the pullups. So if the output is driven low but cannot sink enough
current, it may not go low enough to make the keyboard see it as a low.
(FAIK this is a possible cause for the problem not mentioned so far.)

Using a scope you'd be able to figure out not only the power levels but also
the shapes of the signals. I'd like to hear the results.

I can't take this computer down till tonite, but I'll have more info later.
 
P

petrus bitbyter

mike said:
snip

Ok,
I got paranoid about the 5V standby supply and took the machine apart.
I had one power supply with a bad cap in the 5V standby switcher
and was letting 25V spikes thru. Wanted to verify that this
one didn't have that problem.

Turns out that the keyboard does not run off the 5V standby
supply.
And all the power supplies check out ok.
And the keyboard has 5V.
Adding the wall wart didn't help.

But it still wouldn't post.

I got distracted and discovered that the new computer
with the old keyboard does indeed boot after about a minute or so.
Also discovered by accident that if I press a key, the computer
will POST immediately.

I'd cut the reset wire, so I experimented with that.
If I press scroll lock, the light lights.
If I ground the reset pin, nothing happens.
Keyboard still working, light still on.
I infer that grounding the reset pin didn't reset
anything.

I put it all back together and called it done!

As luck would have it, I found a keyboard at a garage sale
today that I like pretty well. I miss the function keys
down the left side, but having F12 and right-control for
virtualbox is a plus. I'll play with that one for a while.

Thanks for the help.
mike

Well, if you only have to hit a key to post I think the problem has been
solved.

I did not expect to reset line to be the cause but you had to check it.

I neither expected the 5V to be the cause anymore. Not after what you told
about it.

I also did not expect what you eventually found but I'm not astonished
either. At post, the BIOS looks for a keyboard by trying to communicate with
it. I suppose the old and relatively slow keyboard missed the initial
message and did nothing while the PC was waiting for a reaction until it
timed out.

I am astonished about the way you found the results at last. In my
experience operators (and other users) tend to hit the keyboard if nothing
seems to happen. Often hit it very hard.

petrus bitbyter
 
M

mike

If I plug in something new and the system won't POST, the very first thing
I do is pull the plug. Don't wanna let the smoke out.

I've had systems put up the BIOS splash screen and hang.
This was a new symptom for me.
Interesting, may explain why the BIOS has a setting where it can ignore
the keyboard. Completely forgot about the POST interrogating the
keyboard part...
I'm pretty sure this bios doesn't have that. Would have been the first
thing I checked.
Can't check while I'm typing this ;-)
John :-#)#
Yep, problem solved. Smoke still intact ;-)
Thanks, mike
 
Top