As an amateur radio guy, I can't help wondering whether I could use this as
a very low bit rate channel across the country.

Philip

On Sun, Feb 7, 2021 at 4:44 PM Andy Talbot andy.g4jnt@gmail.com wrote:

Not daft at all, Andy.  Closely related would be measurements of phase
change between
near the switched load and far away, over a stretch of a few system time
constants.
This would be more challenging, but perhaps doable.

Dana  K8YUM

On Sun, Feb 7, 2021 at 3:44 PM Andy Talbot andy.g4jnt@gmail.com wrote:

Am 08.02.21 um 00:11 schrieb Philip Gladstone:

I have done navigation for tv and phone sats, and the spread
navigation signals are just 20 dB under the MPEG data stream via
the analog transponder. If you back down another 10 or 20 dB, I'm
sure you could disseminate encryption keys or maybe phone without
anybody taking note. And everybody has a TV dish in the right direction.
Search the right one.
Really, I can't imagine that the agencies don't do that already.

Gerhard, DK4XP

On 2/7/21 3:28 PM, Dana Whitlow wrote:

If you happen to own something like a steel mill running electric
furnaces or an aluminum refinery, so you can manipulate the load...

Andy Talbot writes:

There are generally spaking two time constants, the physical and the human.

The physical time constants are probably best understood as water in a
bucket:  If you kick it, complex ripples happen but eventually die out.

The timeconstant is generally order of five or ten cycles, unless you
make a short, those are much faster.

Mechanisms: Huge lumps of rotating iron, lots of switch-mode equipment
(including solar, wind mills and HVDC converters), magnetic hysteresis
in really big transformers, Maxwells/Telegraphers equations on long cables.

Unless the grid is on the verge of collapsing, the human time constant
is around 15 minutes.  Mechanism: People making mostly financial decisions.

As for your experiment, unless you use a psedurandom pattern, you stand
no chance of detecting a pattern.

I expect you may be able to show a bigger effect than you hypothesize
because the grid is not that "stiff".

Unless your house is right next to a big plant, there are five to ten
transformers between you and the power producer, and all of those "delay"
the phase in response to increasing load.

Your primary singal will therefore be phase-shift rather than frequency-
shift, which means your switching will have to be slow enough that
you can measure the difference between them.

PS: If you get friendly with people at grid control, they will send
you a copy of a pile of laplace equations from hell, which they use
to model the grid, and you can start by plugging your experiment
into them.  It wont be pretty.

--
Poul-Henning Kamp      | UNIX since Zilog Zeus 3.20
phk@FreeBSD.ORG        | TCP/IP since RFC 956
FreeBSD committer      | BSD since 4.3-tahoe
Never attribute to malice what can adequately be explained by incompetence.

Lux, Jim writes:

More scary:  Several independent studies have shown that even relatively
moderate bot-nets in the hands of somebody who knows the math of grid
stability would mean no grid stability.

--
Poul-Henning Kamp      | UNIX since Zilog Zeus 3.20
phk@FreeBSD.ORG        | TCP/IP since RFC 956
FreeBSD committer      | BSD since 4.3-tahoe
Never attribute to malice what can adequately be explained by incompetence.

I'm not convinced they will be impressed by your scheme for ultra VLF
signalling Andy but it'd make a nice Radcom article

On Sun, 7 Feb 2021, 21:44 Andy Talbot, andy.g4jnt@gmail.com wrote:

On 2/8/21 12:02 AM, Poul-Henning Kamp wrote:

This is a generalized concern with distributed generation (i.e. rooftop
solar panels on houses), because a bug in the controllers for the line
connected inverter could really screw things up. I suppose there's also
concerns about  malicious activity, but fiascos like the NEST
thermostats stopping  in winter of 2015/2016 are more likely.

Years ago I recall reading accounts of efforts in the U.S. and perhaps Canada to track down the cause(s) of small but noticeable changes in the frequency of certain regional power grids that seemed to occur at more or less the same time each day.

Mark Spencer
mark@alignedsolutions.com
604 762 4099

Dear all,

it is quite an interesting premise. I think one of the challenges is to
even measure these small deviations in frequency. The time correlation
can be achieved with GNSS but the actual determination of the
zero-crossing might be the problem. I think the ZCD must be better than
100ns for this in each cycle. Local noise and environment conditions
render it quite difficult. When constructing a filter for the 50Hz, the
components must be very well designed to cope with the different
temperature, local noise and voltage amplitudes at two different places.

Regarding disruptions in the grid: there are various regulation
mechanisms. Generally the first 10-20 seconds physical effects stabilize
the grid. Afterwards the automatic control takes over.

• saved rotational energy: an initial frequency drop due to a unexpected
  change in load or production is smoothed out by the inertia of all the
  generators, motors, pumps, etc. For a 10% mismatch in power without
  additional measures, the initial frequency drop would be about 1Hz per
  30 seconds.

• self-regulation: a 1% drop in frequency leads to a ~2% reduction of
  load due to frequency dependent loads;

• primary, automatic load-frequency control takes about 30 seconds to
  activate

After this rather short timescales, as Poul-Henning mentioned the human
factor (trading) has the biggest impacts on grid stability. It is
clearly visible with a periodicity of 15 minutes and more pronounced in
the evenings.

A small note on the recent frequency drop in the European grid:
https://www.entsoe.eu/news/2021/01/26/system-separation-in-the-continental-europe-synchronous-area-on-8-january-2021-2nd-update/

Best
Joe

On 2/8/21 8:46 PM, Mark Spencer wrote:

Hi,

zero cross detection is not the best way to track grid frequency.
I built my grid frequency tracker with a uC sampling the grid at 160ks/s
and 12Bit. It is clocked by a TCXO. I fit a sine wave for every cycle of
the grid.

Fitting a sinewave is a linear problem, you do not need any approximations
and it is doable with a small uC. The samples of a sinewave are the
solution to the difference equation sig(n+1)=2*cos(w)*sig(n)-sig(n-1),
which is a linear equation for the unknown cos(w). That is straight and
simple linear algebra. I get a precision in the range of 4E-4 Hz.

Years ago I recall reading accounts of efforts in the U.S. and perhaps
Canada to track down the cause(s) of small but noticeable changes in the
frequency of certain regional power grids that seemed to occur at more or
less the same time each day.

In the european grid you have stable frequency changes in a 15min
timeframe, frequency being high in the morning and at night and low in the
afternoon. The 15 min timeframe is due to the 15min trade periode of the
European Energy Exchange trading platform located in Leipzig. The energy
brokers show a deterministic under/over-estimation of the european power
consumption which results in power shortage/surplus and thus frequency
rise or fall.

If I were a stock exchange geek I'd turn this information into money, much
money :))

Cheers
Detlef

"time-nuts" time-nuts-bounces@lists.febo.com schrieb am 08.02.2021
23:59:07:

relatively

grid

likely.