[ntp:questions] site with lots of GPS's and accessories

unruh unruh at invalid.ca
Thu Mar 22 03:08:58 UTC 2012


On 2012-03-22, Chris Albertson <albertson.chris at gmail.com> wrote:
> On Wed, Mar 21, 2012 at 10:48 AM, unruh <unruh at invalid.ca> wrote:
>
>> >>> The MAX232 datasheet gives maximum for output slew as 30V/us.
>> >>> The MC1488 circuit assumes a capacitor at the output and has
>> >>> a graph of slew rate vs capacitance. I've used values between
>> >>> 100pF and 330pF.
>> >>
>> >> No, it gives that as the maximum over all samples. Unless your
>> >> particular one was cherry picked, the typical is more like 5. 5V/us slew
>> >> rate means a transition time of 2us for -5V to 5V (even 30V/us is a
>> >> transition time of .3us) That then goes along a long (or short) wire,
>> >> which introduces noise) into an rs232 chip on the computer motherboard.
>
>
> This is easy to measure if you have a counter.  Put the TTL pulse on
> channel A and the RS232 on "B" and let the machine measure average of A-->B
> for a thousand seconds or so.   My older HP Universal counter has only

No it is not because what is important is not when the counter thinks a
transition has taken place, but when the RS232 chip on your computer
motherboard thinks it has taken place. And that may be very different.
And since there is absolutely no reason why an rs232 chip should do it
accurately, the designers are not really going to care if it is so noisy
that it produces a 1us fluctuation in that trigger time. 

Now if you carefully select your chips, or create your own converters, I
am certain you can do better. People regularly do ns timing with gps
receivers with special circuitry. And ethernet sends signals at 10Gbs
(100ps/bit)

> 100nS resolution and is not quite good enough to measure such a small
> delay but I'm setting up one that has a 250pS resolution.
>
> Pico seconds are well outside of the realm of NTP but the over all goal is
> to study radio propagation and oscillator stability.  So in addition to NTP
> I have an interest in higher precision timing so I've been collecting some
> esoteric equipment.
>
> I've measured a few things related to PPS and RS232.   First off yes the
> PPS to RS232 deal can be messed up so bad as to make it useless.  That is
> what happened on my first try.   Itried to use a MAC232 chip to drive 100
> feet of cat-5 cable.   The result was very poor.   Noise at the hand full
> of uSec level, unusable.
>
> Current system has a TTL level PPS going through about 4 feet of coax
> cable.  This controls the base on an RF NPN transistor that switches an 8V
> rail, then a short length of wire goes to the DCD pin on the motherboard.
> I can not yet measure the performance of this setup.



More information about the questions mailing list