[time-nuts] Thunderbolt Monitor (Didier Juges)

Bruce Griffiths bruce.griffiths at xtra.co.nz
Sun Sep 14 03:47:45 UTC 2008


WB6BNQ wrote:
>    Hi Dave,
>
>    Personally I question the value of such an item.  As hobbyist we all
>    start immediately planning such a project just because that is our
>    nature.  The Tbolt monitor program provided by Trimble, while lacking
>    in some regards, is adequate for determining the functional condition.
>    If it is working then most people are going to ignore it as its job is
>    just to sit there and do its job.  Very mundane stuff in the aggregate.
>
>    Probably the biggest thing missing in the Trimble software is a
>    pictorial view of satellite placement.  Unless an expensive graphics
>    LCD where planned, it would be out of the range of a simple PIC type
>    project.  As most all who use a Tbolt would have a PC laying around, it
>    reduces the need for such a project.
>
>    That said, were such directions taken seriously, I think it should be a
>    group effort.  This way those who seriously want to produce such a
>    project can collectively arrive at a single meaningful product.  This
>    project would be a good one to funnel through TAPR in a kit form.
>    While not a member of TAPR, I can see their usefulness as a focal point
>    for furthering interest in such projects and it provides a larger
>    audience besides us Timenuts.
>
>    It seems that projects like the TAC2 and such have died out at TAPR
>    though the GPS/time arena still is a high interest.  Probably because
>    the TAC2 idea was made into a commercial product.  Also because of
>    Motorola dropping their efforts at making GPS receivers at a time when
>    it was unclear if anything was going to be available to replace them.
>
>    Perhaps, from a view of the bigger picture, we can generate a renewed
>    interest to satisfy several needs at once.  Currently, only three
>    useful items I can think of provide an interface between the GPS signal
>    and one's local house standard.  The REFLOCK, Brooks Shera and a third
>    party's lesser idea of something similar to the Shera controller.  Of
>    the three, I think the REFLOCK was designed for a different purpose in
>    mind.  Of the remaining two, The Shera controller is the more serious
>    and complete design, particularly considering its design concepts went
>    into a commercial product.
>
>    The point of all this is a new person getting into the GPS locked house
>    standard game is faced with several issues.  Sure they buy a Tbolt (or
>    something similar), but then they have to figure out how to use it.
>    While the Tbolt provides a direct 10MHz output, there are a host of
>    reasons why you would want to buffer its existence.
>
>    Having a house standard sure looks pretty sitting there until you
>    realize the damn thing is not doing anything other then sucking power.
>    Except for an extremely small number of people (count like 3 or 4 {no
>    disrespect intended}) who have the means to stock a major museum or
>    have a fetish for mechanical clocks, the normal person is going to want
>    their house standard to do something useful for them.  So, we have the
>    problem of either distribution or comparison.
>
>    To distribute clean signals in different directions is not so easy,
>    although, currently, a product is available at TAPR for this function.
>    Distributing a signal has its limits and does not satisfy every
>    situation.
>
>    That leaves us with the need to compare lesser devices with the house
>    standard.  This is where the system breaks down.  The one thing that is
>    most plentiful is the oscilloscope, however, having your eyes glued to
>    scope for any length of time is quite tedious.  Such items as the
>    TRACOR 527 Frequency Difference Meter are available, but they are no
>    different then using the oscilloscope in that the measurement is
>    instantaneous, thus providing no history.  Besides things like the 527
>    are expensive and not plentiful.  A phase discriminator along with a
>    chart machine could be used for long term monitoring.  However, the
>    need for lots of hard to find (sometimes) chart paper is a serious
>    distraction.
>
>    To wrap it up, here is what I see as a useful project:
>
>      1.  It should contain at least two independent Shera type controller
>      circuits for locking two separate independent house standards
>      against the GPS signal.
>
>   
Can do much better than the Shera controller with even less parts.
>      2.  Additionally, it should contain at least a single independent
>      DVM display to select and watch the two [item 1] DAC values plus the
>      ability to feed information of those separate actions to a PC.
>
>   
If one sprinkles inexpensive high resolution sigma delta ADCs  as and 
where required then the difficulties associated with an analog 
multiplexer can be avoided.
The display CPU only need gather data from the ADC's measuring the 
voltage of interest.
>      3.  It should contain a third Shera type circuit to use as a
>      comparison channel between either of the two mentioned house
>      standards so that other lesser, uncontrolled devices can be
>      calibrated.
>   
Poor choice it really doesnt have enough range or resolution to be 
particularly useful as a phase comparator.
You can do much better with a classical linear phase comparator 
connected to a high resolution sigma delta ADC.
If the nonlinearity at the end of the range is a problem just use 2 such 
comparators driven driven 90 degrees out of phase.
If a programmable divider is used on each comparator input then the 
device becomes quite flexible.
If implemented using an FPGA or similar device internal crosstalk will 
limit the effective resolution to around 5E-11/Tau at best.
If external flipflops are used to resynchronise the divider outputs the 
effective resolution will be improved significantly.
The easiest way to build the linear phase comparator is to use an AD9901.
With some ingenuity the ADC reference can be made to track the AD9901 
gain significantly reducing thermal drift.
A temperature compensated linear output could also be provided.
>      4.  In Lieu of item 3 a similar arrangement to the TRACOR 527 as the
>      third comparison channel.
>
>      5.  In reference to items 3 and 4, the onboard controller should
>      provide some readout for the calibration function.  Preferably,
>      include an analog meter and the means for the onboard controller to
>      provide some limited means of giving an averaging over time readout,
>      plus the ability to provide this data to a PC.
>
>      6.  As we have lots of TIME (pun intended) running around this
>      board, consider the ability to provide time stamping.
>
>   
A timestamping resolution of around 100ps or so can be achieved  by 
using an FPGA using the vernier delay line technique.
Higher resolution can easily be achieved by either

A) Using a dual simultaneous sampling ADC to sample a quadrature pair of 
sinewaves and then using a processor to calculate the corresponding 
phase angle and combine this with a sampled count (counter clocked by 
the same source as the sinewave quadrature pair). A resolution of 10ps 
or better is easily achieved and if the timestamp rate is low enough 
almost any processor can do the calculations. An FPGA or CPLD is the 
simplest way of implementing the dual phase synchroniser and counter. 
However one ADC is required per input channel. Suitable ADCs are 
relatively inexpensive but are only available in SMT packages.

B) Low pass filter the signal to be timestamped and sample the low pass 
filtered signal with a high speed ADC clocked at 100MHz or so.
The relevant threshold crossings are then calculated from the ADC 
samples using windowed Whittaker Shannon interpolation.
However you will need to use a DSP or similar very fast processor to 
process the samples in real time.
Alternatively if an FPGA is used to collect a sufficient number of ADC 
samples straddling the transition of interest and the signal transition 
frequency is sufficiently low a much slower processor can be used, 
however 32 bit arithmetic may not suffice. 64 bit arithmetic may be 
required to avoid calculation roundoff noise degrading performance.

Other techniques are possible but tend to require a lot more parts and 
calibration can be much more difficult..
>      7.  Provide the ability to read several temperature devices.
>      Something as simple as the DALLAS One-wire devices as probes would
>      work.
>
>   
The capability of using higher accuracy, higher resolution temperature 
sensors such as RTDs and thermistors may also be useful.
>      8.  This project could be implemented with several CPUs,
>      particularly if using the Shera circuits, without too much trouble.
>
>   
Much easier in fact to use several CPUs dedicated to particular 
functions as the instrument can then be more modular and extendable.
>      9.  To make things easier as a kit, design it using thru-hole
>      devices, particularly the CPUs. (Partly said in jest)
>
>   
Difficult to do if CPLDs or FPGAs are used, although some lower 
complexity (64 macrocell) CPLDs are available on DIP compatible daughter 
boards.
>      10.  Do not forget to include an onboard negative supply system so a
>      single positive supply is all that is needed.
>
>      11.  Regarding the data to a PC, provide it as ASCII data that is
>      comma delimited so graphing could be accomplished in an EXCEL type
>      spreadsheet program.
>
>   
Some care should be taken to avoid tying the instrument to any 
particular operating system or software analysis package.
>      12.  Almost forgot.  The ability to communicate with the Tbolt and
>      perhaps the M12M type receiver.
>
>   
Preferably using an isolated RS232 port to avoid ground loops.

>      13.  Finally, PLEASE design for lighted LCDs or, better yet, using
>      bright Vacuum Fluorescent displays.
>
>    I think such a project would be worthwhile.  It could be done as a kit
>    in such a way as to be implemented in stages if one does not need or
>    want the full capability.  For that matter it could be done as a
>    motherboard with daughter cards to implement, for example, the three
>    Shera circuits.
>
>    This is my synopsis, right, wrong or in between.  I hope I have not
>    offended anyone with such a lengthy diatribe.
>
>    Bill....WB6BNQ
>
>   
Bill

If one is going to do all this in one instrument then care should be 
taken to avoid inadvertent coupling via low frequency ground loops.
RF transformers on all RF inputs can be very effective in minimising 
such problems.
Isolating the instrument from the PC by using either LAN or isolated 
serial link(RS232/RS485 etc) for communications would also be useful.
USB is less useful in that it can be somewhat more expensive to isolate 
satisfactorily.

If one requires even higher resolution when comparing standards then a 
dual (or N channel) mixer system is easily constructed using carefully 
designed modular parts.
Isolation amplifiers, mixer preamps, zero crossing detectors etc are 
required. However the zero crossing detectors can be replaced by a high 
resolution sound card.
For low frequency (< 100kHz) beat signals a capacitive mixer IF port 
termination is best.
With some substitutions the same modules can be rearranged to create a 
phase noise measurement system.

Bruce





More information about the Time-nuts_lists.febo.com mailing list