[time-nuts] Clock Driver Design

Didier Juges shalimr9 at gmail.com
Sat Sep 28 08:24:39 EDT 2013


As always, be careful with the hard and fast rules. Most crystal oscillators generate sine waves first, but if the oscillator is part of a GPSDO, it will have to be converted to square to be processed by the logic within the GPSDO, so even if the device has a sine output, there will be a square wave version of the signal within the device. Eventually, the sine has to be converted into a square wave at the other end of the cable either to drive a mixer or a logic gate. 

Where you do that conversion (before the cable or after the cable), or whether you need to go to sine altogether is something that has to be considered as part of the overall system design. 

In many applications, there is absolutely no need to go to sine if you have a clean square wave to begin with, for instance if the cable is short or you connect with well matched devices.

For some info on what cable mismatch does to a square wave, look here: http://KO4BB.com/Test_Equipment/CoaxCableMatching.php

As Magnus pointed out, there are additional considerations such as ease of  troubleshooting that come into play also.

Finally, if your primary concern is frequency distribution to several instruments as opposed to precision timing, sine will probably be easier because multiple reflections and mismatches that would affect time of arrival don't matter if you are only interested in a stable and precise frequency for your radios.

Didier KO4BB


Magnus Danielson <magnus at rubidium.dyndns.org> wrote:
>On 09/27/2013 12:53 PM, Charles Steinmetz wrote:
>> Tom wrote:
>>
>>> One of my first applications is to use a 10MHz output to phaselock a
>>> VCXO master clock in a radio transceiver.  *   *   *   Next I went
>to
>>> IDT to find the best logic buffer I could find.  I am looking at the
>>> IDT 74FCT38072 2 channel clock driver for PPS.  It can drive about
>>> 50mA if needed with 1nS rise and fall times.  The one I am looking
>at
>>> for 10MHz is the ICS553 4 channel clock driver.  This one is good
>for
>>> 25mA drive and they actually give a typical output impedance spec of
>>> 20 Ohms.  With a 3.3V supply, it has 1nS rise and fall times and a
>>> little faster with a 5V supply, 0.7nS and 35mA drive.  To make a
>sine
>>> wave should I use one of the 4 ports on the 4 port driver to input
>to
>>> the filter or should I try to hook the filter input directly to the
>>> clock driver input?
>>> Are there tried and true 10MHz filter circuits or is that a non
>issue?
>>> After the filter would come the video amp set up for a 50 Ohm drive
>>> and into a splitter.  That sound simple enough.
>>
>> I strongly agree with Magnus that distributing square waves is asking
>> for trouble and that converting to sine is preferable unless there is
>> some very good reason not to.
>That comment came out of another thread. Yes, there are issues with
>square distribution. If you want to stick with it, then you need to
>take
>care of that square. Square can have a benefit over sine thought,
>smaller amounts of reflection can be ignored and not cause much shift
>as
>they will be voltage-shifted out from the comparator
>trigger-point/amplifier mid-point. This means that with some care, the
>next step can remove it. A square signal is easy to diagnose with a
>scope, just toss it in there and you have reflections showing up on the
>scope. Sine require a little more subtle methods to figure out what is
>wrong. So, it's a two-edged sword.
>> IIRC, you said the source is CMOS.  So you can do all of your fanout
>> digitally, then filter each output (I believe that is what Bob had in
>> mind).  Or, as you appear to be contemplating based on your comments
>> above, you could convert to sine immediately and then do the fanout
>in
>> the analog domain with a video DA or whatever.  One reasonable filter
>> type to hang on a CMOS output is an L-C-L "tee" filter (there is
>> really no reason not to add one more shunt C at the end, for
>> L-C-L-C).  This filter needs some termination at all times -- the
>open
>> circuit output voltage can be pretty high.  But you can usually get
>> away with an internal termination of ~1k or so.  If you need more
>> current to get the output level you want, parallel several CMOS
>> outputs (all on the same hex buffer chip, preferably).  There is no
>> need for very fast edges, particularly if you are filtering to sine
>> wave.  Nothing exotic is necessary.
>>
>> The same is true even if you decide to distribute square waves.  The
>> fewer higher harmonics you have, the better off you will be.
>Yes and no, depending on what plagues your installation. Sine is the
>perfect waveshape of converting additive noise (hum for instance) into
>jitter if it meets a comparator. The low slewrate will be the problem
>with the sine, so increasing the through-zero slew-rate reduces this
>problem. The jitter conversion follows this formula:
>
>t_rms = n_rms / S_signal
>
>Where n_rms is the Noise (including hum and other unwanted signal) RMS
>value (in V),
>S_signal is the signal slew-rate (in V/s) and
>t_rms is the jitter time (in s)
>
>This have bitten me many times.
>
>You can avoid having additive noise transform into jitter by not
>feeding
>it into a comparator but rather linearly gaining yourself out of the
>situation. Yes that gain-stage will saturate at the ends. See the
>TADD-2
>input stage, as adapted from the Wenzel page on the subject. Such input
>can also handle square signals.
>
>Not all inputs is made to handle sine signals properly, so you can have
>some additional noise on your reference as you lock up your instrument.
>This also means that some receivers is best served by square, even if
>square is tricker to handle to some degree. Others can be best served
>by
>sine.
>
>In the end, there isn't one way which is always right, rather you need
>to adapt to the receivers needs.
>
>It would be nice to have a little standard board with a good quality
>sine squarer such that it can be put inline at or in the instrument.
>
>Cheers,
>Magnus
>_______________________________________________
>time-nuts mailing list -- time-nuts at febo.com
>To unsubscribe, go to
>https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
>and follow the instructions there.

-- 
Sent from my Motorola Droid Razr 4G LTE wireless tracker while I do other things.


More information about the time-nuts mailing list