[ntp:questions] Oddities in termination of cable from gps18.
unruh at invalid.ca
Thu Feb 23 17:18:12 UTC 2012
On 2012-02-23, David J Taylor <david-taylor at blueyonder.co.uk.invalid> wrote:
> "unruh" <unruh at invalid.ca> wrote in message
> news:4ut1r.6712$VR6.850 at newsfe14.iad...
>> The edge is much faster than 200ns. It is at the 20-50ns level which is
>> less than a wavelength away. From the
>> Sure, it is at the 5ns level.
> You might be better using the RS-232 output, not the TTL, because RS-232
> is intended for driving longer cables which TTL is not. I've not measured
> it here but, for example, the Texas Max232 datasheets suggests a rise-time
> around 400 ns, and that's without a heavily capacitive load. This helps
> avoid the wider-bandwidth problems which you may be seeing.
The Sure is only about a 2m cable, not 20m, and the TTL output on the
Sure seems to work fine into both the parallel port ACK and the Serial
port DCD interrupts. Ie, there seems to be no need to use the RS232
converter. And I would rather not introduce delays into the chain unless
I have to.
What I still find weird is that all that the terminated line seems to do
is to decrease the signal, while the unterminated line (open end) has
the full 5V level with no ringing.
But let me get my assumptions on the table-- if one has a cable with an
infinite resistance termination does the reflected signal have the same
polarity (in the E field) as the incoming or opposite?
If it is the same polarity, then what I am seeing makes sense-- Say the
output on the Garmin GPS has about a 400 ohm serial output impedance, so
driving the line (100 ohms) gives a drop of voltage by about 1/5, but as
the signal is reflected back with the same polarity, the effective
impedance of the line seems to rise, finally leading to the full voltage
developing on the line. If terminated by 100 ohm, the driver continues
to see the line as a 100 ohm load giving an output voltage of about 1V
rather than 5V. That also seems to fit with the structure I see on that
rising edge of the pulse measured on the terminal end (initial step of
about 1V on the unterminated line. I'll try to see if I can get a
>> The receiver is the ACK pin on a parallel port, not on a serial port.
>> And 1 V is just on the hairy edge seemingly without hysteresis.
> Using something which is intended for RS-232 reception would be a better
> choice.. If you /are/ seeing reflections on the signal, try adding a small
> resistor (say 47-82 ohms) in series with the /sending/ end, so that the
> cable is slightly better terminated and you don't get multiple
> reflections. Adjust with the 'scope for the cleanest edge. Even better,
> use the RS-232 signals.
More information about the questions