long term mains 60Hz precision

The mains will be more accurate. The power industry invests lots of money to ensure they're at exactly 60Hz.

I've tried to use the 4MHz xtal as a time reference on 2 occasions and found both times that the accurate wasn't acceptable. You could use a 32.768KHz watch crystal as an option, those crystals are laser trimmed to be exact.

I built a frequency counter (many, many) years ago and based the timing on the power frequency. I called the local power company to ask how accurate the 60hz was. The first person didn't know what a hertz was; the second person insisted that it was Exactly 60 hertz (just what kind of power company do you think we are, anyway???); then they switched me to some engineer there that looked at some reading and told me the 'variance' at that moment.

I believe that the short term variance is within a fraction of a percent, but the long term accuracy is very high.

Luhan

I built a frequency counter (many, many) years ago and based the timing on the power frequency. I called the local power company to ask how accurate the 60hz was. The first person didn't know what a hertz was; the second person insisted that it was Exactly 60 hertz (just what kind of power company do you think we are, anyway???); then they switched me to some engineer there that looked at some reading and told me the 'variance' at that moment.

I believe that the short term variance is within a fraction of a percent, but the long term accuracy is very high.

Luhan

It's a rental car company. Shoulda told him :-)

--
Regards, Joerg

http://www.analogconsultants.com

(Double posting courtesy of Google Groups).

Luhan

I also tried few weeks ago an external 32.768KHz reference from a small watch crystal as you said and it "seemed" to be more accurate even though I could not be absolutely sure. I takes many days to see a significative difference... and I don't have an atomic clock!!! Anyway, because I have a readily available 60Hz reference connected to the PIC, I will probably use it a the main reference. I will have my program to switch to the 4MHz reference only during power outage then back to 60Hz reference when AC power is back. That should do te trick!

Johnny

If you have a shortwave receiver then you've got one: Tune it to 10MHz or 15MHz WWV and they'll announce the exact time. For free, courtesy of the taxpayer :-)

Make sure there is a low pass filter so that spikes from the vacuum cleaner motor don't create "extra" clock cycles.

--
Regards, Joerg

http://www.analogconsultants.com

Good idea! I didn't know about that possibility.

There's been some discussion on this in the past. I believe the average frequency of the national grid is tied to the NIST atomic clocks so the power plants have a reference to sync up on.

The line frequency can be dragged low by large power drain, such as local dinnertime or heavy use of air conditioners. They try to make it up in each

24 hour period by increasing the frequency during light loads, such as after midnight. This means the instantaneous frequency can drift quite a bit, but the long-term average is very good.

If your local power plant is not tied to the national grid, such as Boulder Colorado was many years ago, you would see much larger variations in line frequency since there was less reserve to buffer load changes. A frequency counter would easily show the frequency change during a 24 hour interval.

The atomic clock services available on the web seem to be accurate to better than one second. My cmos clock used to drift slow by several seconds per day. After erasing some old software and cleaning up the registry, the time accuracy improved dramatically, and it hasn't needed adjusting for several months.

Regards,

Mike Monett

Antiviral, Antibacterial Silver Solution:

SPICE Analysis of Crystal Oscillators: Noise-Rejecting Wideband Sampler:

If you have a GPS receiver, you have an atomic clock. I was able to tell a

14 second difference between GPS time and my circuit's time after only 3 days.

During the recent power outages (on Nov 4th) in Europe, frequency dropped as far as 49 Hz (from the normal 50Hz) before automatic mechanisms started disconnecting loads.

But GPS is satellite time. Minus 38uS/day correction from relativistic error....

Just kidding :)

Johnny

On Sat, 09 Dec 2006 20:21:31 GMT, Joerg Gave us:

Then one must calculate arrival delay for an accurate time base reference. Better just to buy a radio clock that has the circuitry built in.

Far more accurate now is your GSP enhanced (or not) cell phone and any GPS receiver. The non enhanced phone gets its time from the cell node it is logged onto. The GPS enhanced version gets it from a GPS bird. The GPS receiver does as well. All three more accurate than the resolve a atomic clock radio receiver is.

60Hz mains fluctuate.

"Nevo"

** Watch crystals are less " exact " than the normal kind.

Initial accuracy is 20 ppm at 25C or 52 seconds per month.

Then there is the tempco....

The big advantage is low drive power - ie 1 uW v. 1 mW.

..... Phil

There's an interesting article in Scientific American titled "The Power Grid as Complex System". It shows power failures can be similar to a sand pile, where

one additional grain of sand will cause an avalanche that redistributes sand throughout the grid. This process keeps the slope of each small pile of sand just below the critical threshold.

Frequency and voltage errors can cause instabilities that can bring the entire grid down. According to this article, the critical frequency in the Eastern Grid is 18 millihertz:

An indicator of how well a grid is operating is the stability of the frequency at which it transmits energy - 60 cycles per second, or 60 hertz, is the norm. Since 1999 the average deviation from that frequency has been increasing in the East toward the danger threshold: 18 thousandths of a hertz above the norm. If this continues, the possibility of blackouts could increase as well.

Graph tracks monthly average deviation from ideal frequency in thousandths of a hertz since 1994.

The US has three separate power grids, Eastern, Western, plus one for Texas:

The United States does not have one national power grid, but rather three separate grids - a Western grid, an Eastern grid and one in Texas, managed by the Electric Reliability Council of Texas. The electric system in the western United States - called the Western Interconnection - operates independently from the electric system in the eastern United States. Western operates primarily in the Western grid and owns and maintains more than 10 percent of the transmission lines in the Western Electricity Coordinating Council area.

The grids are tied together with DC links:

There are six DC ties connecting the Western Interconnection and the Eastern Interconnection in the United States and one additional DC tie in Canada. Western is associated with four of them. Western owns and operates the Virginia Smith (Sidney, Neb.) DC tie; owns 60 percent of the Miles City DC (Mont.) tie and operates it; and operates the David A. Hamil DC Tie Stegall (Neb.) DC tie (owned by Tri-State Generation and Transmission Cooperative.) Western also operates the back-to-back DC Converter Station project in Rapid City, S.D., owned by Basin Electric Power Cooperative and Black Hills Power and Light. The station can transfer up to 200 MW or power between the Western Electricity Coordinating Council in the Western Interconnection and the Mid-Continent Area Power Pool in the Eastern Interconnection. The other two US ties are Public Service Company of New Mexico 's Blackwater N.M., DC tie and the El Paso Electric and Texas-New Mexico Power Company's Artesia, N.M., DC tie. Xcel Energy is building another DC tie in Lamar, Colo.

DC links use IGBT's in switching circuits to convert power:

Bridge Converters

A bridge converter (Figure 5) is a simplified form of converter that uses unipolar switches to transfer power from AC to DC (active rectifiers) or DC-AC (inverters). With high-speed switches, these converters can provide highly regulated and flexible DC and AC output. Multi-level converters (Ref. 5) are a special subset of bridge converters, and represent a complex and expensive method of circumventing the previous unavailability of high voltage switching devices.

The failure mode is known so extra IGBT's provide redundancy:

Redundancy is achieved because IGBTs always fail shorted, and as many as three devices (out of twenty in a medium voltage switch) can fail without adversely impairing the operation of the series stack.

"Dynamic Demand" could improve system reliability and reduce the number of power generation stations needed:

Canada has experimented with short-circuiting long transmission lines to break the accumulated ice. The short-circuit currents are huge, and the results are quite dramatic. Here's the cover photo:

De-icing EHV Overhead Transmission Lines by Short-circuit Currents.

Here's the paper:

That would be a fun job:)

Regards,

Mike Monett

Antiviral, Antibacterial Silver Solution:

SPICE Analysis of Crystal Oscillators: Noise-Rejecting Wideband Sampler:

The mains frequency is very accurate in the long term, but can vary quite a bit short term (maybe 0.1 Hz or so). The PIC crystal oscillator can be tuned slightly by changing the small capacitors to ground. For a PIC18F242 circuit I divide the the 14.7456 MHz clock down to exactly 1 or 2 Hz and measure this with a frequency counter in period mode. It might be a good idea to pick up the power line frequency and compare it to the internal clock, or at least have a fallback mode to use the crystal if the power fails.

Paul

I tried using the line 60Hz in my Nixie clock, as well as putting a 2ppm

32kHz TCXO on the board. I'm glad I made provisions for the TCXO, since I was shocked to find the clock could drift many seconds when using the line. Checking against my radio atomic clock.

I could also watch the line drift in one direction or another on the scope, when triggering off the TCXO (or vice/versa) for quite lengthy periods of time.

I coupled my TCXO into the Atmel mega8 real time crystal oscillator terminals. Against, the datasheet's recommendations which state you can't use an external clock, but it works. I think I use a 1Meg resistor or something. I wouldn't try it with a high speed XO, just 32kHz.

Good day!

--
_____________________
Christopher R. Carlen
crobc@bogus-remove-me.sbcglobal.net
SuSE 9.1 Linux 2.6.5

** That is really very funny.

Assuming that 2ppm figure is for real and maintained over a period of 12 months ( I'll bet $10 it is not ) - the expected drift in the crystal time base in a year amounts to 1/500,000 of the number of seconds in a year.

Know how many that is ??

It's more than a minute !!

While a 60 Hz clock would be at most a few seconds out in a year ( barring power outages) .

....... Phil

Did you monitor it for more than a day? I've read somewhere that the power companies take great pains to "catch up" with all of the clocks that are running synchronously off the line, and also to keep the various parts of the grid in sync.

If you try this, please let us know how it comes out. :-)

Good Luck! Rich