[time-nuts] New Timestamping / Time Interval Counter: the TICC

Tom Van Baak tvb at LeapSecond.com
Sun Nov 27 19:58:23 EST 2016


John,

The PD15 divider that you're using will sync to within 1 PIC instruction. That's sounds good, but the TICC reads down to picoseconds so the 400 ns PIC granularity will look pretty high.

There is another solution. And that is not to sync the h/w or s/w counters at all. You just let them free-run. But you allow a sync mode where the user places a known UTC 1PPS in chA or chB and then the Arduino takes a reading. That number is then stored and subtracted from all subsequent readings that the Arduino outputs. It's better than a h/w sync because it allows you to sync to the ps level. And it also takes into account any propagation delays on your board, connector and cables.

/tvb

----- Original Message ----- 
From: "John Ackermann N8UR" <jra at febo.com>
To: <time-nuts at febo.com>
Sent: Sunday, November 27, 2016 4:18 PM
Subject: Re: [time-nuts] New Timestamping / Time Interval Counter: the TICC


Hi Bob --

It's certainly possible to synchronize the TICC timescale epoch to an 
external pulse at startup.  The external sync pulse would have to 
simultaneously reset (a) the picDIV and (b) the coarse (100us) counter 
in the Arduino. The signals to do that are available, so it's a 
worthwhile experiment.

I have to think through whether there are any other gotchas but I'm 
pretty sure that you would not get closer than a few hundred nanoseconds 
given how closely the picDIV can sync -- I think Tom specifies that it's 
within 4 clocks, or 400 ns.

(Now, syncing two TICCs to each other is a different matter because 
there we one unit is master and provides the coarse clock directly to 
the slave; if both units are run from the same 10 MHz clock they should 
align within one 10 MHz tick.)

John
----

On 11/27/2016 12:26 PM, Bob Camp wrote:
> Hi
>
> Ok, so the guess was fairly close :)
>
> How about a connector to allow an external PPS to reset the internal 10 MHz divider? That way all
> the data is “in sync” with the house standard. If I want to know that my GPSDO is +32.751 ns off from
> the house standard, I just look at the data on a terminal program …
>
> Or am I missing something really obvious (again)?
>
> Bob
>
>> On Nov 27, 2016, at 10:04 AM, John Ackermann N8UR <jra at febo.com> wrote:
>>
>> Good guess.  The 10 MHz reference drives all the logic on the board, and particularly the counter that maintains a local timescale in 100us increments; the TDC7200 interpolates between the 100us ticks to stamp incoming events on channel A and/or B with picosecond precision.  The stamps on both channels are referenced to the same local timescale.
>>
>> Therefore, you can do a measurement of a PPS source against the 10 MHz reference and the resulting timestamp output can be processed by TimeLab or whatever into stability data (the requirement being that the software knows how to deal with timestamps that increment by the nominal measurement rate, e.g., 1 second per measurement for PPS data).
>>
>> So with PPS from GPSDO "A" on channel A the timestamp output after unwrapping will show the phase of A vs. 10 MHz.
>>
>> You can add PPS from GPSDO "B" on channel B and the TICC will also output timestamps of B vs. the 10 MHz source.
>>
>> If you want, you can subtract A from B to get the time interval between the two GPSDO, since both timestamp measurements are against a common timescale.  The TICC has a mode to output the (B-A) difference, so it can act as either a traditional time interval counter, or as a two-channel timestamping counter.
>>
>> And as noted in my other message to Luciano, the TICC can also output both timestamp and time interval data simultaneously to allow three-corner-hat measurements of (A-C, B-C, B-A) where C is the 10 MHz reference.
>>
>> John
>> ----
>>
>> On 11/27/2016 09:24 AM, Bob Camp wrote:
>>> Hi
>>>
>>> Without doing a bunch of actual *work* I’m not sure what is inside the guts of the board. Being
>>> lazy I’ll just guess ….
>>>
>>> There appears to be a 10 MHz time base input and a pair of measurement inputs. In a lot us will
>>> be comparing to a “house standard�. That standard has a pps output that is related directly to
>>> the 10 MHz reference. If I can uniquely identify one edge (out of 10 million edges) as the right
>>> edge, I can use the 10 MHz as my pps reference. Put another way, I don’t really need to measure
>>> a pps input from the house standard if I’m already locked up in phase to the 10 MHz. All I need to
>>> do is to tag an edge / reset a counter.
>>>
>>> The advantage of this is that I may not need another fancy TDC chip to set up the reference. I can
>>> use *both* inputs for DUT’s rather than using one as a reference.
>>>
>>> Part of the reason I’m guessing this would work is the claim that boards can be stacked for multiple
>>> input setups ….
>>>
>>> Bob
>>>
>>>> On Nov 27, 2016, at 7:36 AM, timeok <timeok at timeok.it> wrote:
>>>>
>>>>
>>>>    Hi John,
>>>> I have planned to buy two TICC.
>>>> An interesting feature would be to be able to do two simultaneous acquisitions,
>>>> and Timelab as real time display,using the two indipendent input channels and the 10Mhz clock as single reference.
>>>> Luciano
>>>> www.timeok.it
>>>>
>>>>
>>>>    From "time-nuts" time-nuts-bounces at febo.com
>>>>    To "Discussion of precise time and frequency measurement" time-nuts at febo.com
>>>>    Cc
>>>>    Date Wed, 23 Nov 2016 10:48:57 -0500
>>>>    Subject [time-nuts] New Timestamping / Time Interval Counter: the TICC
>>>>    Counters with resolution below 1 nanosecond are difficult. They require
>>>>    either outrageous clock speeds, or interpolators that are typically a
>>>>    bunch of analog components mixed with black magic and stirred by
>>>>    frequent calibration. The very best single-shot resolution that's been
>>>>    commercially available is 22 picoseconds in the HP 5370A/B, with jitter
>>>>    somewhat more than that. My 5370B has an one-second noise ADEV of about
>>>>    4x10e-11.
>>>>
>>>>    With the help of some very talented friends, I've been working on a new
>>>>    counter called the "TICC" with <60ps resolution and similar jitter,
>>>>    based the Texas Instruments TDC7200 time-to-data-converter chip. The
>>>>    noise ADEV is about 7x10e-11, not much worse than the 5370,
>>>>    but here's the trick: the TICC is an Arduino shield (mounting a Mega
>>>>    2560 controller) that weighs a couple of ounces, requires *no*
>>>>    calibration, and is powered from a USB cable!
>>>>
>>>>    The TICC is implemented as a two-channel timestamping counter. That
>>>>    means it can measure or two low-frequency (e.g., pulse-per-second)
>>>>    inputs against an external 10 MHz reference, or it can do a traditional
>>>>    time interval measurement of input against the other. It can also
>>>>    measure period, ratio, or any other function of two-channel timestamp
>>>>    data. (And by the way -- multiple TICCs can be connected to yield 4, 6,
>>>>    8, or more synchronized channels, though we haven't tested this
>>>>    capability yet.)
>>>>
>>>>    I've attached a picture of the TICC prototype as well as an ADEV plot of
>>>>    a 17+ day run of multiple measurements taken by two TICCs, and also
>>>>    showing the TICC noise floor. The good news behind that plot is that
>>>>    there are more than 6 million data points behind these results, and
>>>>    there was not a single glitch or significant outlier among them.
>>>>
>>>>    There's more information available at http://febo.com/pages/TICC
>>>>
>>>>    The software is open source (BSD license) and is available at
>>>>    https://github.com/TAPR/TICC -- the current version seems be reliable
>>>>    but there are still features to add and a *lot* of cleanup to do; it's
>>>>    currently ugly and very much a work in process.
>>>>
>>>>    As always, I'll be making the TICC available through TAPR. We're still
>>>>    finalizing details, but we expect the price to be less than $200 for a
>>>>    turn-key system: TICC mounted an Arduino with software loaded and
>>>>    tested for basic functionality. We hope to ship the TICC by February.
>>>>
>>>>    I'll post a note in a week or two with final price and ordering
>>>>    information. As a heads up, we will probably offer a small discount for
>>>>    pre-orders. TAPR is a shoestring non-profit group and the up-front cost
>>>>    to manufacture this unit will frankly be a challenge for us. Getting
>>>>    pre-orders will help our cash flow significantly, so we ask you to keep
>>>>    that in mind.
>>>>
>>>>    John
>>>> _______________________________________________
>>>> 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.
>>>
>>> _______________________________________________
>>> 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.
>>>
>> _______________________________________________
>> 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.
>
> _______________________________________________
> 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.
>



--------------------------------------------------------------------------------


> _______________________________________________
> 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.


More information about the time-nuts mailing list