[LinuxPPS] (OT) A question about DCD.

[Magnus Danielson]

Cirilo Bernardo skrev:
> On 1/28/09, clemens at dwf.com <clemens at dwf.com> wrote:
>> This is somewhat off topic.
>>  I have one machine, well two that linuxPPS and NTP run on just fine.
>>  However, on my 'main-machine' I have problems.
>>
>>  Back at the end of the summer I noted that I no longer had the PPS signal
>>  on the DCD line.  Neither ppstest nor ntpd reported a signal.
>>  I have two active GPS receivers, so I could trade them.  Same problem,
>>  and the 1st worked just fine on the 2nd machine.
>>
>>  OK, so its the machine, not the GPS receiver nor the cables.
>>
>>  After some thought and no obvious conclusions, I did the simple thing
>>  and bought a 4port tty card and plugged it into the machine.
>>
>>  I plugged the GPS card into it, and everything worked again, so
>>  it SEEMED that it had to be the actual tty port that was bad.
>>
>>  But given a month, that stopped working too.
>>  I tried a 2nd port on the 4port card, and it worked for a while,
>>  then stopped.  Mumph.
>>
>>  Since then, at one time or another, I have played with the problem,
>>  and the port has worked for a short period of time, but then stopped.
>>
>>  I have NOT pulled out a scope to see the height of the PPS signal, but the
>>  fact that it works effortlessly on the 2nd, or 3rd machine makes me think
>>  that that experiment isnt work the effort.
>>
>>  So, what could be going wrong?
>>  I can think of no way that the software (linuxPPS) could stop working,
>>  but neither can I come up with a reason why both the onboard and a 4port
>>  tty card should both stop working.  (this is an Oncore card, so there is
>>  plenty of tty chatter in addition to the PPS, so one cant just look
>>  at interrupts...)
>>
>>  On occasion it seemed that moving from one port to another of the 4port
>>  card was enough to get things working again,- just doesnt make any sense.
>>
>>  So, any hardware (or kernel) gurus out there that can make any sense of
>>  this???
>>  --
> 
> You may have a problem with the voltage levels on your interfaces.  If
> you are using +/-12VDC on your DCD pin and the serial port circuitry
> is meant to use 0to+5VDC, you may be damaging your circuitry.  The
> damage will not necessarily be instant (the first time you plug in the
> GPS + interface) because the pulse durations may be short, the maximum
> current is limited (as per electrical standards of RS232C), and there
> may be some internal protection circuitry which survives the abuse for
> a few hours/days/weeks.

I have never seen any PC side serial interface (markeded as RS232) not 
being able to cope with RS232 signal levels and only accept TTL/CMOS 
signals. I have seen that on the GPS side thought and only on the OEM 
board level, never on complete modules. So I am sceptic about this 
assumption.

> To check on the voltage issues:
> 1. read the datasheet for the specific UART you are using

It rarely is in the UART anyway... a nifty MAX-chip or something takes 
the power and switch it to a local dual supply in order to drive RS-232C 
levels.

> 2. check the voltage on the serial port's TxD pin after opening the
> serial port via minicom; for an interface which is truly RS232C
> compliant, you should have about -12VDC
> 3. instruct the serial port to set the output on the DTR pin to the
> high value.  check the voltages at 'low' and 'high' to see what the
> UART is using.  Unfortunately everyone claims that circuitry is to
> RS232C spec (which is an electrical specification) but MANY are lying.
> I've seen the following:
> 
> +/-12V (on spec)
> 0v,+5V (bad)
> +/-5V (bad, but within signal tolerance of spec)
> 0v,+3.3V (bad)
> 
> For example, the laptop I am typing on at the moment uses +/-5V.  The
> input in this case is claimed to be protected to about 12V (ideally it
> should be protected to at least 18V - no one bothers to read the
> electrical specification anymore!) so I can use it to receive from a
> device which *does* follow the RS232C.  I can also use it to transmit
> to a device which follows RS232C; I think the minimum voltage swing
> specified is +/-3V (but don't quote me - I haven't read the spec
> recently).

You should be happy you have a RS-232C on your laptop. They are not 
there anymore, which means to say that laptop producers are not 
interested in selling laptops to the telecom industry where serial ports 
still is important interface for management... and oh, telecom industry 
has way longer equipment lifecycles than laptop industry.

Cheers,
Magnus