At a guess, in the absence of any additional details, you have a
couple ports wired as DTEs (Data Terminal Equipment, most common
config for printers and other output-only devices) and a couple ports
wired as DCEs (Data Communication Equipment). Ordinarily when a
computer has these ports wired differently, they will also have
connectors of different gender, so you shouldn't be able to just take
the cable from one port and plug it straight into the other port.

You probably need to get a null-modem to go between your current cable
and the other port when you swap the cables around. (A null-modem is
just a pair of connectors wired together, with Transmit and Receive
lines crossed over. They should be pretty cheap, though perhaps a bit
rarer these days now that USB is replacing RS232...)

On Aug 12, 9:51 pm, asham...@yahoo.com wrote:
> Hi,
>
> I have a question which I'm pretty sure is about serial ports. I work
> at a supermarket helpdesk, and one of our most common problems is when
> the EFTPOS terminals (don't know if they're called this outside of
> Australia, what I mean is the card swipe things you use when you pay
> for your groceries on your card) stop communicating with the register
> (a linux box).
>
> More often than not, this is because the serial port the EFTPOS
> terminal attaches to has stopped working. At the moment though, we
> have to run a whole bunch of tests to rule everything else out before
> we know one way or another.
>
> What we'd like to do is take the customer display (just an array of
> LEDs capable of displaying two lines of text at a time), plug it into
> the EFTPOS serial port and send a line of text to it, just to test the
> port.
>
> However, this doesn't work.
>
> There are four devices attached to the register: the customer display,
> the receipt printer, a scanner/scale unit and the EFTPOS terminal.
>
> The customer display and the receipt printer can be interchanged. The
> receipt printer will print out the display text if attached to the
> customer display serial port, and vice versa.
>
> If the customer display is plugged into the scanner/scale port, it
> will print using an alphabet of symbols when text is sent to it.
>
> But the customer display won't respond at all when plugged into the
> serial port for the EFTPOS terminal.
>
> This might be a really dumb question, but I work in an office with
> five other guys, all with tertiary education in computing and nobody
> here has any idea what the solution might be. I assume the serial
> port needs to be reconfigured for a different I/O, but don't know how
> to go about doing this.
>
> Any help would be greatly appreciated.
>
> -Mark

Thanks. Is there any easy way to check how a serial port is wired?

On Aug 13, 3:14 pm, hyc wrote:
> At a guess, in the absence of any additional details, you have a
> couple ports wired as DTEs (Data Terminal Equipment, most common
> config for printers and other output-only devices) and a couple ports
> wired as DCEs (Data Communication Equipment). Ordinarily when a
> computer has these ports wired differently, they will also have
> connectors of different gender, so you shouldn't be able to just take
> the cable from one port and plug it straight into the other port.
>
> You probably need to get a null-modem to go between your current cable
> and the other port when you swap the cables around. (A null-modem is
> just a pair of connectors wired together, with Transmit and Receive
> lines crossed over. They should be pretty cheap, though perhaps a bit
> rarer these days now that USB is replacing RS232...)
>
> On Aug 12, 9:51 pm, asham...@yahoo.com wrote:
>
> > Hi,
>
> > I have a question which I'm pretty sure is about serial ports. I work
> > at a supermarket helpdesk, and one of our most common problems is when
> > the EFTPOS terminals (don't know if they're called this outside of
> > Australia, what I mean is the card swipe things you use when you pay
> > for your groceries on your card) stop communicating with the register
> > (a linux box).
>
> > More often than not, this is because the serial port the EFTPOS
> > terminal attaches to has stopped working. At the moment though, we
> > have to run a whole bunch of tests to rule everything else out before
> > we know one way or another.
>
> > What we'd like to do is take the customer display (just an array of
> > LEDs capable of displaying two lines of text at a time), plug it into
> > the EFTPOS serial port and send a line of text to it, just to test the
> > port.
>
> > However, this doesn't work.
>
> > There are four devices attached to the register: the customer display,
> > the receipt printer, a scanner/scale unit and the EFTPOS terminal.
>
> > The customer display and the receipt printer can be interchanged. The
> > receipt printer will print out the display text if attached to the
> > customer display serial port, and vice versa.
>
> > If the customer display is plugged into the scanner/scale port, it
> > will print using an alphabet of symbols when text is sent to it.
>
> > But the customer display won't respond at all when plugged into the
> > serial port for the EFTPOS terminal.
>
> > This might be a really dumb question, but I work in an office with
> > five other guys, all with tertiary education in computing and nobody
> > here has any idea what the solution might be. I assume the serial
> > port needs to be reconfigured for a different I/O, but don't know how
> > to go about doing this.
>
> > Any help would be greatly appreciated.
>
> > -Mark

ashamael@yahoo.com wrote:

> Hi,

Hi,

> I have a question which I'm pretty sure is about serial ports. I work
> at a supermarket helpdesk, and one of our most common problems is when
> the EFTPOS terminals (don't know if they're called this outside of
> Australia, what I mean is the card swipe things you use when you pay
> for your groceries on your card) stop communicating with the register
> (a linux box).

Here in the Netherlands they are called EFT terminals or PIN pads.
Pretty much the same everywhere, I guess.

> More often than not, this is because the serial port the EFTPOS
> terminal attaches to has stopped working. At the moment though, we
> have to run a whole bunch of tests to rule everything else out before
> we know one way or another.

RS-232 serial ports are known to stop 'working' temporarily or even
permanently
(i.e. blow up!) when the attached device is (un)plugged or power-cycled.
Serial devices in general are not 'hot-pluggable' as opposed to USB devices.
Some motherboards/chipsets are very forgiving to RS-232 'hot-plugging'
whereas
others are very sensitive.
Did you try to find out whether anyone unplugged or power-cycled the EFT
terminal prior to calling for your help?
Check the output of the 'dmesg' command for any reference to the
word 'UART'.
If the Linux device driver detects a shutdown of the serial port, it will
display a message (I forgot the exact message phrase).


> What we'd like to do is take the customer display (just an array of
> LEDs capable of displaying two lines of text at a time), plug it into
> the EFTPOS serial port and send a line of text to it, just to test the
> port.
>
> However, this doesn't work.

Possible explanation below...

> There are four devices attached to the register: the customer display,
> the receipt printer, a scanner/scale unit and the EFTPOS terminal.

Ok, seems like a pretty standard POS configuration.

> The customer display and the receipt printer can be interchanged. The
> receipt printer will print out the display text if attached to the
> customer display serial port, and vice versa.
>
> If the customer display is plugged into the scanner/scale port, it
> will print using an alphabet of symbols when text is sent to it.
>
> But the customer display won't respond at all when plugged into the
> serial port for the EFTPOS terminal.

POS vendors have done someting 'interesting' to extend the RS-232 standard
to provide DC power (5V or 12V) on one of the 'unused' pins in the RS-232
connector. In most POS boxes this so-called 'power-over-serial' feature is
controlled through parameters in the BIOS.
A typical POS customer display (2x20 fluorescent display) is powered through
the RS-232 cable using this feature; it doesn't require a separate power
supply.
The behaviour you're observing could be caused by the fact that some
serial ports on your POS box are configured for 'power-over-serial' and
others are not.

Be very careful about the proper cabling when connecting devices to a serial
port carrying 12V DC on one of its pins; you could blow up your precious
serial device...


> This might be a really dumb question, but I work in an office with
> five other guys, all with tertiary education in computing and nobody
> here has any idea what the solution might be. I assume the serial
> port needs to be reconfigured for a different I/O, but don't know how
> to go about doing this.

Like described above, check the BIOS. What brand of POS hardware are you
using?


Regards,

Ton.
--
--------------------------------------------------------------------------------
//
// Ton Nijkes Email: ton@murphy.nl
//|\ || Murphy Software BV,
//||\\|| P.O. Box 285, Voice: +31 (0)53 4320055
|| \|| 7500 AG Enschede, The Netherlands Fax : +31 (0)53 5360448
--------------------------------------------------------------------------------

Ton Nijkes wrote:
> ashamael@yahoo.com wrote:
>
>> Hi,
>
>Hi,
>
>> I have a question which I'm pretty sure is about serial ports. I work
>> at a supermarket helpdesk, and one of our most common problems is when
>> the EFTPOS terminals (don't know if they're called this outside of
>> Australia, what I mean is the card swipe things you use when you pay
>> for your groceries on your card) stop communicating with the register
>> (a linux box).
>
>Here in the Netherlands they are called EFT terminals or PIN pads.
>Pretty much the same everywhere, I guess.
>
>> More often than not, this is because the serial port the EFTPOS
>> terminal attaches to has stopped working. At the moment though, we
>> have to run a whole bunch of tests to rule everything else out before
>> we know one way or another.
>
>RS-232 serial ports are known to stop 'working' temporarily or even
>permanently
>(i.e. blow up!) when the attached device is (un)plugged or power-cycled.

That is not true.

>Serial devices in general are not 'hot-pluggable' as opposed to USB devices.

Only in the sense that the OS does not support any kind
of auto initialization for RS-232 devices. That is
simply because there is no way to determine
automatically what configuration it should be configured
for.

As for removing, power cycling, or reconnecting RS-232
devices and then manually starting whatever process is
required to initialize the device, that *is* *very*
commonly done.

>Some motherboards/chipsets are very forgiving to RS-232 'hot-plugging'
>whereas
>others are very sensitive.

I've never seen even one that was "sensitive". I've
been using RS-232 devices for literally decades, on all
sorts of equipment.

>Did you try to find out whether anyone unplugged or power-cycled the EFT
>terminal prior to calling for your help?
>Check the output of the 'dmesg' command for any reference to the
>word 'UART'.
>If the Linux device driver detects a shutdown of the serial port, it will
>display a message (I forgot the exact message phrase).

I don't think it does that...

....


>Be very careful about the proper cabling when connecting devices to a serial
>port carrying 12V DC on one of its pins; you could blow up your precious
>serial device...

Not with 12 VDC you won't. Every pin on an RS-232
interface can deal with significantly more voltage than
that.

But you advice that the cabling is critical is indeed
correct. There are approximately 2 more ways to cable
RS-232 than there are design engineers to decide which
way it is, so not many are ever the same. :-)

It isn't going to blow anything up, but it isn't going
to work either.

--
Floyd L. Davidson
Ukpeagvik (Barrow, Alaska) floyd@apaflo.com

Floyd L. Davidson wrote:

> Ton Nijkes wrote:
>> ashamael@yahoo.com wrote:
>>
[snip]
>>
>>RS-232 serial ports are known to stop 'working' temporarily or even
>>permanently
>>(i.e. blow up!) when the attached device is (un)plugged or power-cycled.
>
> That is not true.

Is is not true because you never experienced it yourself or because
it is...well...not true?

>>Serial devices in general are not 'hot-pluggable' as opposed to USB
>>devices.
>
> Only in the sense that the OS does not support any kind
> of auto initialization for RS-232 devices. That is
> simply because there is no way to determine
> automatically what configuration it should be configured
> for.

True.

> As for removing, power cycling, or reconnecting RS-232
> devices and then manually starting whatever process is
> required to initialize the device, that *is* *very*
> commonly done.

I know, I do it myself _all_ the time :-) It works well _most_ of the
time :-(

>>Some motherboards/chipsets are very forgiving to RS-232 'hot-plugging'
>>whereas
>>others are very sensitive.
>
> I've never seen even one that was "sensitive".

I have. I have seen specific brand PCs (Wincor) lock up hard consistently
when
hot-plugging RS-232 devices.
I have seen other brand PCs (NCR) have their serial ports toasted after
hot-plugging RS-232 devices. Not just one, but many.
Note that this all applies to POS (Point-Of-Sale) PCs and peripherals.
Bad hardware design on some component is entirely imaginable.

Never seen sparks coming from an RS-232 connector? I thought that's why
they call it 'hot' plugging ;-)

> I've been using RS-232 devices for literally decades, on all
> sorts of equipment.

Same here.

>>Did you try to find out whether anyone unplugged or power-cycled the EFT
>>terminal prior to calling for your help?
>>Check the output of the 'dmesg' command for any reference to the
>>word 'UART'.
>>If the Linux device driver detects a shutdown of the serial port, it will
>>display a message (I forgot the exact message phrase).
>
> I don't think it does that...
>
> ...

Well it doesn't really 'detect' a shutdown of the port. In fact, the kernel
complains that no UART chip is found at a given I/O-address, although it was
working perfectly fine just moments before... So probably the UART decided
to call it quits. The particular kernel message to look for is:

ttyS%d: LSR safety check engaged!


>>Be very careful about the proper cabling when connecting devices to a
>>serial port carrying 12V DC on one of its pins; you could blow up your
>>precious serial device...
>
> Not with 12 VDC you won't. Every pin on an RS-232
> interface can deal with significantly more voltage than
> that.

Ooops, my bad.
You are right. RS-232 typically operates at signal voltages of 3V or 5V, but
it
is supposed to be able to handle voltages between 3V and 25V (+/-).
So 12V should be fine...i.e. not break anything.

> But you advice that the cabling is critical is indeed
> correct. There are approximately 2 more ways to cable
> RS-232 than there are design engineers to decide which
> way it is, so not many are ever the same. :-)

:-(

> It isn't going to blow anything up, but it isn't going
> to work either.

Rule #1 wrt RS-232: Never expect it to 'just work' :-(


Regards,

Ton.
--
--------------------------------------------------------------------------------
//
// Ton Nijkes Email: ton@murphy.nl
//|\ || Murphy Software BV,
//||\\|| P.O. Box 285, Voice: +31 (0)53 4320055
|| \|| 7500 AG Enschede, The Netherlands Fax : +31 (0)53 5360448
--------------------------------------------------------------------------------

Ton Nijkes wrote:
>Floyd L. Davidson wrote:
>> Ton Nijkes wrote:
>>> ashamael@yahoo.com wrote:
>>>
>[snip]
>>>
>>>RS-232 serial ports are known to stop 'working' temporarily or even
>>>permanently
>>>(i.e. blow up!) when the attached device is (un)plugged or power-cycled.
>>
>> That is not true.
>
>Is is not true because you never experienced it yourself or because
>it is...well...not true?

Let me repeat that one more time: it is *not* true.

The whole idea that power cycling a device will cause an
RS-232 port to "blow up" is hilarious.

>>>Serial devices in general are not 'hot-pluggable' as opposed to USB
>>>devices.
>>
>> Only in the sense that the OS does not support any kind
>> of auto initialization for RS-232 devices. That is
>> simply because there is no way to determine
>> automatically what configuration it should be configured
>> for.
>
>True.

I hope you understood the use of "only" in that
statement! :-)

>> As for removing, power cycling, or reconnecting RS-232
>> devices and then manually starting whatever process is
>> required to initialize the device, that *is* *very*
>> commonly done.
>
>I know, I do it myself _all_ the time :-) It works well _most_ of the
>time :-(

Exactly. The difference is that you don't know why it
fails now and then... and those who actually work with
this stuff day and and day out for decades do (or
should).

In three words: Electro static discharge. It has
nothing to do with an RS-232 port as such. It's a
simply case of a person walking over and grabbing the
connectors/cords/devices without taking appropriate
steps to discharge any static build up. The same damn
fool thing happens with circuit boards of any kind,
RS-232 or not. (In fact, I've seen circuit cards blown
by people just *walking* past them... and I don't
mean on poorly designed sysytems either.)

>>>Some motherboards/chipsets are very forgiving to RS-232 'hot-plugging'
>>>whereas
>>>others are very sensitive.
>>
>> I've never seen even one that was "sensitive".
>
>I have. I have seen specific brand PCs (Wincor) lock up hard consistently
>when
>hot-plugging RS-232 devices.

So a faulty mother board design is used to condemn
RS-232 ports?

(To be honest though, I'm not willing to credit your
ability to correctly observe and report, given the
incredibility of the other statements in these two
articles. But I have never seen a "Wincor" motherboard
and cannot vouch for it either way, so I cannot claim
you are necessarily wrong.)

>I have seen other brand PCs (NCR) have their serial ports toasted after
>hot-plugging RS-232 devices. Not just one, but many.
>Note that this all applies to POS (Point-Of-Sale) PCs and peripherals.
>Bad hardware design on some component is entirely imaginable.

Sounds more like poor maintenance practices that need
improvement.

>Never seen sparks coming from an RS-232 connector? I thought that's why
>they call it 'hot' plugging ;-)

Incredible. Sparks from an RS-232 connector???? Either
you need to learn more about static discharge, or you
need to learn more about what is and what is *not*
RS-232. You cannot get an RS-232 lead to spark. The
voltages and impedances make it virtually impossible.

That is, assuming proper ESD precautions...

>> I've been using RS-232 devices for literally decades, on all
>> sorts of equipment.
>
>Same here.
>
>>>Did you try to find out whether anyone unplugged or power-cycled the EFT
>>>terminal prior to calling for your help?
>>>Check the output of the 'dmesg' command for any reference to the
>>>word 'UART'.
>>>If the Linux device driver detects a shutdown of the serial port, it will
>>>display a message (I forgot the exact message phrase).
>>
>> I don't think it does that...
>>
>> ...
>
>Well it doesn't really 'detect' a shutdown of the port. In fact, the kernel
>complains that no UART chip is found at a given I/O-address, although it was
>working perfectly fine just moments before... So probably the UART decided
>to call it quits. The particular kernel message to look for is:
>
> ttyS%d: LSR safety check engaged!

Unplugging or power cycling of a connected RS-232 device
does not cause the port to shutdown, the UART to
disappear, or a kernel error.

>>>Be very careful about the proper cabling when connecting devices to a
>>>serial port carrying 12V DC on one of its pins; you could blow up your
>>>precious serial device...
>>
>> Not with 12 VDC you won't. Every pin on an RS-232
>> interface can deal with significantly more voltage than
>> that.
>
>Ooops, my bad.
>You are right. RS-232 typically operates at signal voltages of 3V or 5V, but

RS-232 does not typically operate at 3-5V. They
typically operates at 10 to 15 volts, and on virtually
all PC's they use +/- 12 VDC.

>it
>is supposed to be able to handle voltages between 3V and 25V (+/-).
>So 12V should be fine...i.e. not break anything.
>
>> But you advice that the cabling is critical is indeed
>> correct. There are approximately 2 more ways to cable
>> RS-232 than there are design engineers to decide which
>> way it is, so not many are ever the same. :-)
>
>:-(
>
>> It isn't going to blow anything up, but it isn't going
>> to work either.
>
>Rule #1 wrt RS-232: Never expect it to 'just work' :-(

Rule #2, when it doesn't, it didn't blow up.

--
Floyd L. Davidson
Ukpeagvik (Barrow, Alaska) floyd@apaflo.com

On 2007-08-15, Floyd L. Davidson wrote:

>>Ooops, my bad.
>>You are right. RS-232 typically operates at signal voltages of 3V or 5V, but
>
> RS-232 does not typically operate at 3-5V. They
> typically operates at 10 to 15 volts, and on virtually
> all PC's they use +/- 12 VDC.

You guys just caused a major flashback, here. Major. :-)


I studied the CCITT norm on RS-232 signaling, some 20-odd years ago. At that
time, I also looked into the hardware side of it. In those days, I used it
to connect PC-XT's to a Intergraph Vax and later to some microcomputers,
like the Synclair QL. Oh, and the pen plotters and first laser printers used
RS-232, too.

From memory (allways dangerous, but that's all I have here) an original
RS-232 was expected to signal using +/- 12V. It had to be able to 'endure'
24V at least, maybe 32V. And it had to be able to get a signal out of a
tension as low as +/- 3V. That was for the old DB25 connector.

When the DB9 came into use, the typical voltage dropped to +/-5V, probably
because PC's motherboards dropped in tension, too. But the high and low
limits still had to be respected. This drop in tension might not have been
part of any standard, but was common practise, especially on luggeables,
portables and laptops.



Groan... I'm getting old for this.
Back to lurking mode, then.


--
There is an art, it says, or rather, a knack to flying.
The knack lies in learning how to throw yourself at the ground and miss.
Douglas Adams

Rikishi 42 wrote:
>On 2007-08-15, Floyd L. Davidson wrote:
>
>>>Ooops, my bad.
>>>You are right. RS-232 typically operates at signal voltages of 3V or 5V, but
>>
>> RS-232 does not typically operate at 3-5V. They
>> typically operates at 10 to 15 volts, and on virtually
>> all PC's they use +/- 12 VDC.
>
>You guys just caused a major flashback, here. Major. :-)
>
>I studied the CCITT norm on RS-232 signaling, some 20-odd years ago. At that
>time, I also looked into the hardware side of it. In those days, I used it
>to connect PC-XT's to a Intergraph Vax and later to some microcomputers,
>like the Synclair QL. Oh, and the pen plotters and first laser printers used
>RS-232, too.
>
>From memory (allways dangerous, but that's all I have here) an original
>RS-232 was expected to signal using +/- 12V. It had to be able to 'endure'
>24V at least, maybe 32V. And it had to be able to get a signal out of a
>tension as low as +/- 3V. That was for the old DB25 connector.

Your memory is fairly good!

Note that the DB25 connector has virtually nothing to do
with it though. No connector is specified by the RS-232
Standard, and a huge variety of connectors have been
used. The DB25 was very common in the computer industry
initially, to the point where most people think that it
is part of the standard. Then when things started to
get smaller and PC's began using a 9 pin connector (it
is in fact a DE9 connector), everyone called it a DB9!
The "DB" references the physical size though, and a DB9
would look just like a DB25 (with 16 pins removed).

The signal voltages are specified for the transmitter as
5 to 15 volts, while for the receiver it is 3 to 25
volts. That is with a load resistance of 3000 to 7000 Ohms.
The transmitter voltage with no load must be below 25 volts.
And the output impedance when power is off must be greater
than 300 Ohms.

The receiver input range is specified equal to the
driver specifications. Nominal input is 15 volts.
Impedance is 3k to 8k. It must survive 25 volts input,
and should "work" with as low as 3 volts. (See below for
the reason quotes are around the word work.)

The specification also states that the device must be able
to sustain a short between any two pins without damage, and
that the maximum current will be 0.5A.

>When the DB9 came into use, the typical voltage dropped to +/-5V, probably
>because PC's motherboards dropped in tension, too. But the high and low
>limits still had to be respected. This drop in tension might not have been
>part of any standard, but was common practise, especially on luggeables,
>portables and laptops.

The ability to work at +/- 5 volts was always part of
the standard, but actually using that voltage was not
and is not common or recommended. It is true that many
compact devices, battery operated devices, etc. etc.
will use lower voltages, but few will attempt to work a
5 volts with a production unit, though 9-10 volts is
often encountered.

The reason for that is because lower voltages have
several effects. The most obvious is that it reduces
the maximum bit rate which will work over any given
length of cable. Indeed, with some cables and some
devices that might mean that it won't work at all with
anything practical! A ten foot run of cable might be
restricted to bit rates lower than 300 b/s, for
example... and if your application requires a 56.7Kb/s
rate, it dies (even though the RS-232 implementation
technically does meet specs).

>Groan... I'm getting old for this.
>Back to lurking mode, then.

I've always found RS-232 interesting, mostly because so
many design engineers have no practical experience...
they read the specs and figure it means what it
says. :-)

They design to the specs; while at the same time very
few technicians have any theoretical experiences, and
have never read the specs. Techs are often totally at a
loss to understand what the design engineer intended
because it is some oddball offbeat never used by anyone
else implementation!

--
Floyd L. Davidson
Ukpeagvik (Barrow, Alaska) floyd@apaflo.com

Floyd L. Davidson wrote:

> Ton Nijkes wrote:
>>Floyd L. Davidson wrote:
>>> Ton Nijkes wrote:
>>>> ashamael@yahoo.com wrote:
>>>>
>>[snip]
>>>>
>>>>RS-232 serial ports are known to stop 'working' temporarily or even
>>>>permanently
>>>>(i.e. blow up!) when the attached device is (un)plugged or power-cycled.
>>>
>>> That is not true.
>>
>>Is is not true because you never experienced it yourself or because
>>it is...well...not true?
>
> Let me repeat that one more time: it is *not* true.
>
> The whole idea that power cycling a device will cause an
> RS-232 port to "blow up" is hilarious.

What about (un)plugging?
Are you saying that no person can possibly toast a serial port by
(un)plugging an RS-232 connector? Now that is hilarious.

>>>>Serial devices in general are not 'hot-pluggable' as opposed to USB
>>>>devices.
>>>
>>> Only in the sense that the OS does not support any kind
>>> of auto initialization for RS-232 devices. That is
>>> simply because there is no way to determine
>>> automatically what configuration it should be configured
>>> for.
>>
>>True.
>
> I hope you understood the use of "only" in that
> statement! :-)
>
>>> As for removing, power cycling, or reconnecting RS-232
>>> devices and then manually starting whatever process is
>>> required to initialize the device, that *is* *very*
>>> commonly done.
>>
>>I know, I do it myself _all_ the time :-) It works well _most_ of the
>>time :-(
>
> Exactly. The difference is that you don't know why it
> fails now and then... and those who actually work with
> this stuff day and and day out for decades do (or
> should).

I don't think you are in the position to decide what I know and what I
don't know. Or did we go to school together?

> In three words: Electro static discharge. It has
> nothing to do with an RS-232 port as such. It's a
> simply case of a person walking over and grabbing the
> connectors/cords/devices without taking appropriate
> steps to discharge any static build up. The same damn
> fool thing happens with circuit boards of any kind,
> RS-232 or not. (In fact, I've seen circuit cards blown
> by people just *walking* past them... and I don't
> mean on poorly designed sysytems either.)

In know what ESD is, thank you very much.
I think your approach is a purely technical one whereas I look at the
pragmatic side. Keep in mind that most people are not as knowledgeable
as you appear to be about this. OTOH, almost everyone is capable of
(un)plugging an RS-232 connector.
It is a fact that fooling around with RS-232 connectors on a powered-up
PC puts you at risk of damaging you hardware. For most people, it is
fairly irrelevant whether that has something to do with the RS-232 port
as such, or just with electronic principles in general.
I take precautions, that's why I can mostly get away with it. Other
people might not be so lucky...that's what the warning was about.

>>>>Some motherboards/chipsets are very forgiving to RS-232 'hot-plugging'
>>>>whereas
>>>>others are very sensitive.
>>>
>>> I've never seen even one that was "sensitive".
>>
>>I have. I have seen specific brand PCs (Wincor) lock up hard consistently
>>when
>>hot-plugging RS-232 devices.
>
> So a faulty mother board design is used to condemn
> RS-232 ports?

Again, pragmatic vs. technical. Who is condemning RS-232? I'm not...
Fact is, you have to be careful with RS-232. And with some brands you
have to be more careful than with other brands.
That's why I wouldn't call it 'hot-pluggable'.

> (To be honest though, I'm not willing to credit your
> ability to correctly observe and report, given the
> incredibility of the other statements in these two
> articles. But I have never seen a "Wincor" motherboard
> and cannot vouch for it either way, so I cannot claim
> you are necessarily wrong.)

It's not a "Wincor" motherboard, it's a Wincor POS (as in Point-Of-Sale,
not Piece-Of-... ;-) PC. I didn't bother to check the motherboard brand.

>>I have seen other brand PCs (NCR) have their serial ports toasted after
>>hot-plugging RS-232 devices. Not just one, but many.
>>Note that this all applies to POS (Point-Of-Sale) PCs and peripherals.
>>Bad hardware design on some component is entirely imaginable.
>
> Sounds more like poor maintenance practices that need
> improvement.

You're possibly right. That doesn't make it less true, though.

>>Never seen sparks coming from an RS-232 connector? I thought that's why
>>they call it 'hot' plugging ;-)
>
> Incredible. Sparks from an RS-232 connector???? Either
> you need to learn more about static discharge, or you
> need to learn more about what is and what is *not*
> RS-232. You cannot get an RS-232 lead to spark. The
> voltages and impedances make it virtually impossible.

It looks like RS-232, it smells like RS-232, it tastes like RS-232.
What is it?
Again, technical vs. pragmatic. I've seen sparks coming from that 9-pin
Sub-D connector that is commonly found on the back of most PCs. Whether
the actual source of the spark is what you call RS-232 or not, is not
very relevant in this context.
The whole idea if this thread was to warn the OP to be cautious with
RS-232 connectors.

> That is, assuming proper ESD precautions...

Of course, assuming...

>>> I've been using RS-232 devices for literally decades, on all
>>> sorts of equipment.
>>
>>Same here.
>>
>>>>Did you try to find out whether anyone unplugged or power-cycled the EFT
>>>>terminal prior to calling for your help?
>>>>Check the output of the 'dmesg' command for any reference to the
>>>>word 'UART'.
>>>>If the Linux device driver detects a shutdown of the serial port, it
>>>>will display a message (I forgot the exact message phrase).
>>>
>>> I don't think it does that...
>>>
>>> ...
>>
>>Well it doesn't really 'detect' a shutdown of the port. In fact, the
>>kernel complains that no UART chip is found at a given I/O-address,
>>although it was working perfectly fine just moments before... So probably
>>the UART decided to call it quits. The particular kernel message to look
>>for is:
>>
>> ttyS%d: LSR safety check engaged!
>
> Unplugging or power cycling of a connected RS-232 device
> does not cause the port to shutdown, the UART to
> disappear, or a kernel error.

Ok, then you tell me what does.

>>>>Be very careful about the proper cabling when connecting devices to a
>>>>serial port carrying 12V DC on one of its pins; you could blow up your
>>>>precious serial device...
>>>
>>> Not with 12 VDC you won't. Every pin on an RS-232
>>> interface can deal with significantly more voltage than
>>> that.
>>
>>Ooops, my bad.
>>You are right. RS-232 typically operates at signal voltages of 3V or 5V,
>>but
>
> RS-232 does not typically operate at 3-5V. They
> typically operates at 10 to 15 volts, and on virtually
> all PC's they use +/- 12 VDC.

True, my bad again :-(

>>it
>>is supposed to be able to handle voltages between 3V and 25V (+/-).
>>So 12V should be fine...i.e. not break anything.
>>
>>> But you advice that the cabling is critical is indeed
>>> correct. There are approximately 2 more ways to cable
>>> RS-232 than there are design engineers to decide which
>>> way it is, so not many are ever the same. :-)
>>
>>:-(
>>
>>> It isn't going to blow anything up, but it isn't going
>>> to work either.
>>
>>Rule #1 wrt RS-232: Never expect it to 'just work' :-(
>
> Rule #2, when it doesn't, it didn't blow up.

:-)


Regards,

Ton.
--
--------------------------------------------------------------------------------
//
// Ton Nijkes Email: ton@murphy.nl
//|\ || Murphy Software BV,
//||\\|| P.O. Box 285, Voice: +31 (0)53 4320055
|| \|| 7500 AG Enschede, The Netherlands Fax : +31 (0)53 5360448
--------------------------------------------------------------------------------

Floyd L. Davidson wrote:
> The whole idea that power cycling a device will cause an
> RS-232 port to "blow up" is hilarious.

Especially when everyone knows that it's a divide by zero that
causes computers to "blow up". ;-)

--
PLEASE post a SUMMARY of the answer(s) to your question(s)!
Show Windows & Gates to the exit door.
Unless otherwise noted, the statements herein reflect my personal
opinions and not those of any organization with which I may be affiliated.

Ton Nijkes wrote:
>Floyd L. Davidson wrote:
>> Ton Nijkes wrote:
>>>Floyd L. Davidson wrote:
>>>> Ton Nijkes wrote:
>>>>>RS-232 serial ports are known to stop 'working' temporarily or even
>>>>>permanently
>>>>>(i.e. blow up!) when the attached device is (un)plugged or power-cycled.
>>>>
>>>> That is not true.
>>>
>>>Is is not true because you never experienced it yourself or because
>>>it is...well...not true?
>>
>> Let me repeat that one more time: it is *not* true.
>>
>> The whole idea that power cycling a device will cause an
>> RS-232 port to "blow up" is hilarious.
>
>What about (un)plugging?
>Are you saying that no person can possibly toast a serial port by
>(un)plugging an RS-232 connector? Now that is hilarious.

I'm saying that when using standard techniques and
dealing with equipment of standard design, it is *not* a
problem.

How good the design is of course does affect how
sensitive it is to ESD. Not that you can not perhaps
cite some oddball equipment that isn't designed
properly, and not that poor technique with regard to ESD
won't cause problem...

But yes, I *am* saying that you are not correct, because
it is not generally RS-232 ports that are at fault.

>I don't think you are in the position to decide what I know and what I
>don't know. Or did we go to school together?

You've been fairly clear in the last couple of articles
you've posted. You probably do have a lot of experience
and know a lot of things. I can't tell about that.

You don't know all that much about RS-232 equipment in
general, and yes I *can* tell. (If nothing else, the
claim that an RS-232 port will "blow up" when a device
is power cycled is sufficient.)

It isn't so much that I've worked on equipment with a
lot of RS-232 ports, as it is on a few thousands of
_different_ types of RS-232 ports. A computer tech, for
example, might deal with 20 to 100 serial ports a
day... but they are almost all identical. In the
telecom industry it's not nearly as many, but every
darned one is *different*.

>> In three words: Electro static discharge. It has
>> nothing to do with an RS-232 port as such. It's a
>> simply case of a person walking over and grabbing the
>> connectors/cords/devices without taking appropriate
>> steps to discharge any static build up. The same damn
>> fool thing happens with circuit boards of any kind,
>> RS-232 or not. (In fact, I've seen circuit cards blown
>> by people just *walking* past them... and I don't
>> mean on poorly designed sysytems either.)
>
>In know what ESD is, thank you very much.
>I think your approach is a purely technical one whereas I look at the
>pragmatic side. Keep in mind that most people are not as knowledgeable
>as you appear to be about this. OTOH, almost everyone is capable of
>(un)plugging an RS-232 connector.

It is not uncommon that even good techs/engineers know
little about what happens when they plug and unplug
RS-232 cables. For average folks, it would be rare that
they do know.

But RS-232 is intended to be hot pluggable.

And only fools do *anything* to electronic equipment
without discharging themselves. That has nothing at all
to do with RS-232 as such, and people who manage to
damage things with ESD are going to do so whether they
are plugging in RS-232 cables or not. Just grabbing the
cable with a load of electrons on their fingers is
enough, and it has nothing to do with disconnecting the
cable.

And it certainly has nothing to do with power cycling
equipment, which you claimed would "blow up" an RS-232
port. It just is not true. (And any equipment that
does should *immediately* be removed from service as too
dangerous to be used.)

>It is a fact that fooling around with RS-232 connectors on a powered-up
>PC puts you at risk of damaging you hardware. For most people, it is

See above. It is a fact that fooling around with
equipment without being discharged does put it at risk.
That has nothing to do with the RS-232 port as such,
other than it has a cable sticking out that is a handy
place be discharged! Of course monitors and printers
and USB and Ethernet are all just as prone to being
grabbed as a point of discharge. (It does happen that a
monitor is more likely to use a shielded cable, and
RS-232 is the second most likely of those mentioned to
be shielded. And a grounded shield obviously does offer
at least some protection, though not much. Plus an
Ethernet cable might be less susceptible to lower levels
of discharge, though at higher levels there wouldn't be
much difference.)

>fairly irrelevant whether that has something to do with the RS-232 port
>as such, or just with electronic principles in general.
>I take precautions, that's why I can mostly get away with it. Other
>people might not be so lucky...that's what the warning was about.

Then you should tell people to use proper technique
rather than telling them that RS-232 will blow up. It
won''t, and powering down (other than that it might help
to discharge someone) doesn't make the equipment any
less sensitive to ESD.

>> So a faulty mother board design is used to condemn
>> RS-232 ports?
>
>Again, pragmatic vs. technical. Who is condemning RS-232? I'm not...

Oh, you said what? That RS-232 isn't sensitive and can
be unplugged and power cycled? NOT... ;-)

>Fact is, you have to be careful with RS-232. And with some brands you
>have to be more careful than with other brands.
>That's why I wouldn't call it 'hot-pluggable'.

You don't need to be any more careful with RS-232 than
any other part of the cabling coming out of a PC.

>>>Never seen sparks coming from an RS-232 connector? I thought that's why
>>>they call it 'hot' plugging ;-)
>>
>> Incredible. Sparks from an RS-232 connector???? Either
>> you need to learn more about static discharge, or you
>> need to learn more about what is and what is *not*
>> RS-232. You cannot get an RS-232 lead to spark. The
>> voltages and impedances make it virtually impossible.
>
>It looks like RS-232, it smells like RS-232, it tastes like RS-232.
>What is it?
>Again, technical vs. pragmatic. I've seen sparks coming from that 9-pin
>Sub-D connector that is commonly found on the back of most PCs. Whether
>the actual source of the spark is what you call RS-232 or not, is not
>very relevant in this context.

Yes it is. If you grab the cable to the keyboard with
that sort of charge on you, it is just as risky.

>The whole idea if this thread was to warn the OP to be cautious with
>RS-232 connectors.

Wrong caution.

Power cycling simply isn't a problem. *Any* touching of
the ungrounded portions of the equipment is risky, and
citing RS-232 as the cause is poor advice.

>> Unplugging or power cycling of a connected RS-232 device
>> does not cause the port to shutdown, the UART to
>> disappear, or a kernel error.
>
>Ok, then you tell me what does.

ESD, removing the chips, etc...

>>>Rule #1 wrt RS-232: Never expect it to 'just work' :-(
>>
>> Rule #2, when it doesn't, it didn't blow up.
>
>:-)

If the hardware is a pain... try learning to program
RS-232 ports. It takes some real experience to figure
out how POSIX termios works, and Linux (as do other
variations of Unix) has some unique quirks that are all
but impossible to know about.

--
Floyd L. Davidson
Ukpeagvik (Barrow, Alaska) floyd@apaflo.com

nobody@tek.com (Kevin the Drummer) wrote:
>Floyd L. Davidson wrote:
>> The whole idea that power cycling a device will cause an
>> RS-232 port to "blow up" is hilarious.
>
>Especially when everyone knows that it's a divide by zero that
>causes computers to "blow up". ;-)

Divide by zero is dynamite.

--
Floyd L. Davidson
Ukpeagvik (Barrow, Alaska) floyd@apaflo.com