[time-nuts] Trigger circuit for TIC

[Bob kb8tq]

Hi

(wandering a long way away from the original post …..)

When you design a DMTD that runs into a counter, by far the biggest
part of the DMTD design is the limiter(s) that turn the mixer output into
a “useable” square wave. Even high end commercial counters are unable
to process the output of a DMTD directly. 

On the flip side, the counter “needs” of a DMTD generally are not all that
tight. The “best of the best” approach will still likely have a noise floor in
the 10’s of ns on that square wave output.

Some math:

If I go from 10 MHz to a beat note below 10 Hz, I “amplify” the error by
at least 1x10^6.

If I put that signal into a counter with a 1 us resolution (no not a typo, a 
1 MHz time base …. welcome to the 1950’s :) ) I get a LSD at 1x10^-6
per second. 

I now can do 1x10^-12 at 1 second ADEV (barely, but that seems to be
the criteria used by a lot of people). 

Since this is no longer the 1950’s, clock rates on the counter higher than 
1 MHz *are* possible. A $3  MCU can provide a 32 bit counter with 
an internal clock above 200 MHz.  Yes it’s got a wacky PLL and there is 
some noise, your signal has noise …..

That puts the MCU (if it works ….) at roughly the price of the two TDC’s 
that a DMTD would need. Since the MCU also provides computing and
com horsepower it (if it works …) is pretty much the slam dunk answer
for that end of the chain. 

======

One wrinkle you will run into with any mixer based DMTD is the “one 
supply or two?” question. With one supply, you will need to have a
virtual ground throughout everything. That might push you to a 6 layer 
board. With a bipolar supply setup, you have a point where you need
to get back to single supply for your 7200 or MCU. 

Toss in “only use parts that are in stock at the contract house” and it
gets a bit crazy.

Lots of fun !!!

Bob

> On Jun 3, 2020, at 5:49 AM, Tobias Pluess <tpluess at ieee.org> wrote:
> 
> Dear colleagues
> I am currently thinking about a concept for my own TIC. While I gathered
> some knowledge during my GPSDO project about using the TDC7200 as
> interpolator, I have not yet a good idea how I should design the input
> stage of a TIC. It should be somewhat universal, i.e. it should be usable
> with analog sine waves as well as with digital signals (TTL).
> 
> So the trigger input circuit shall accept a signal of less than approx. 5
> volts amplitude, and provide a digital output signal (3.3V logic level)
> which can then be used for further processing. It would be good if I could
> directly feed it with the output signals from a DMTD, but could also
> connect the 1PPS output signals of GPSDOs to it.
> I know that trigger jitter can be a problem, especially when the signals
> have small amplitudes and are sine instead of square. So what should I
> consider to minimise this effect?
> I have seen some schematics where people used ADCMP comparators, but I have
> forgotten which one exactly and I cannot remember where I saw it. On the
> other hand, on other pages I read that comparators are not recommended
> because they have a higher jitter, and it was recommended to use high-gain
> amplifiers instead.
> So what is the "proper" way to design such a trigger circuit which is
> usable for sine as well as square waves?
> (if the jitter was not critical, I would simply use some comparators.)
> 
> Background: I want to construct something like a timestamping counter which
> allows me to caracterise my GPSDO I built. For this, I would like to do
> different things: first, I want to compare the 1PPS signals of different
> GPSDOs with respect to mine (5V TTL signals), but on the other hand, I also
> want to use the DMTD method which provides signals with maybe some 10Hz
> frequency and several 100mV amplitude. It would be good if there was a
> trigger circuit which could be used for both.
> 
> 
> 
> BR
> Tobias
> HB9FSX
> _______________________________________________
> time-nuts mailing list -- time-nuts at lists.febo.com
> To unsubscribe, go to http://lists.febo.com/mailman/listinfo/time-nuts_lists.febo.com
> and follow the instructions there.