[LinuxPPS] Using GPS for automotive applications

[Cirilo Bernardo]

On Sun, Apr 6, 2008 at 8:30 PM, Remco den Besten <besten at gmail.com> wrote:
> DCD is input and modern chipsets (16550's) are well
>  capable of detecting TLL levels.
>
>  I do not use an interface other than an inverter (LS00) as buffer
>  as precaution, but this is not necessary.
>
>
>
>  > Neal Probert wrote:
>  >> PPS Interface
>  >> -------------
>  >> The simplest way I can see interfacing the PPS is through a MAX232 chip
>  >> into the serial port's DCD input.  Right?
>  >
>  > In some cases (maybe even most modern cases) you can feed the (TTL?) PPS
>  > into the RS232 DCD pin without trouble. I am using this setup without
>  > problems.
>  >
>  > Depending on the board you maybe can jumper the RS232 port to 5V
>  > operation, as on some of my Epia boards.
>  >
>  > A max232 convertor costs a few ns of delay.
>

This is not quite true about the levels etc.  Many OEM GPS units
operate from 3.3v only; only some families of 5.0v logic chips are
input/output compatible with these 3.3v devices.  So you cannot simply
connect the GPS PPS to the DCD line.  To complicate things, devices
claiming to be RS232 compatible have very different characteristics:

1. DCD input is TTL (will be damaged by -V, or +12V)
2. DCD input is standards compliant and the line must transition from
<= -3.0v to >= +3.0v
3. DCD input is TTL levels but will tolerate the -12/+12 signals and
function properly
4. DCD uses only 0 .. +12V but has TTL input characteristics

And any combination of the above.  You really have to read the specs
for your GPS unit and your specific serial port controller.

If you have a TTL serial port and a 3.3v GPS, be extremely careful in
selecting logic gates to
ensure that states switch properly; otherwise you will have high power
consumption (things may actually be damaged by heating) and nothing
will work properly anyway.

In many cases, why use a chip like the MAX232 - the PPS is output
only, not in/out - a resistor and small-signal MOSFET should do
beautifully and the propagation delay is still in the low nanoseconds
(far too small to matter for DCD-based PPS).