Depending upon where in the world you live, AC mains frequency is either 50Hz or 60Hz, and that frequency is maintained accurately enough over time that it can be used as a time reference for a clock. Oddly it’s rarely exactly that figure though, instead it varies slightly with load on the network and the operators will adjust it to keep a constant frequency over a longer period. These small variations in frequency can easily be measured, and [jp3141] has created a circuit that does exactly that.
It’s a surprisingly straightforward device, in which a Teensy takes its power supply from a very conventional if now a little old-school mains transformer, rectifier, and regulator. A sample of the AC from the transformer passes through a low-pass filer and a clamp, and thence to the Teensy where it is fed into one of the on-board comparators from which its period is measured using one of the timers. Even then the on-board crystal isn’t considered accurate enough, so it is in turn disciplined by a 1 pulse per second (PPS) signal from a GPS receiver.
The Teensy then reports its readings over a serial line every five seconds to a Raspberry Pi, which collates and graphs the data. In case you are wondering what the effect of mains frequency variations might be, we once covered the story of how an entire continent lost six minutes.
26 thoughts on “Accurately Track Your Mains Frequency”
Tom Scott video discussing extracting the mains frequency variations from video and, in combination with logged frequency data, establishing a forensic timestamp.
For those interested: the continental EU grid the current frequency live data is available on https://www.mainsfrequency.com/index.htm
Hi would be very interested to see the Allan deviation of the mains’ frequency
It’s linked in the github readme, but worth repeating here. The UTenn FNET grideye map shows the frequency and relative phases for the north american power grids. Interesting to see their examples of perturbations due to outside events: http://fnetpublic.utk.edu/sample_events.html
https://www.netzfrequenzmessung.de/ has some really nice information about the grid frequency, including the current frequency, load shedding and gridtime
After WWIII let’s rebuild the world on one standard. I propose 60Hz for smaller transformers and 220V for better electric kettles and better power tools in domestic garages.
So you need WW3 to change to a different output on your inverter! What does it take to get you to change your toothpaste?
400 Hz (used in many aircraft) is much more efficient.
Skin effect at 400 Hz is eight times that of 60 Hz, no?
The efficiency saved on 400 Hz transformers is probably lost on the larger mains conductors required.
MBTA uses 400 hz power for streetcars and subways precisely because it requires smaller conductors. They can pull a 6 car train from one thin overhead wire.
MBTA green line street cars use 600v DC overhead. The frequency doesn’t make a big difference in efficiency and DC is used for historical reasons. The subway (Red/Orange/Blue) also uses 600v DC on the third rail.
400Hz does not reduce wire size, it’s just physics. It reduces transformer size due to much lower V*s
MBTA operates on 400-800 VDC, hence the single wire.
MBTA electricity is generated and distributed at 400 Hz to substations where it is converted to DC for final delivery. Those substations are everywhere, make loud buzzing sounds and strange smells at the end of the platform. The power cables running in the tunnels are carrying AC.
I propose 50kHz to avoid audible noise and 20kV for cheaper, lower mass conductors.
Wikipedia has an interesting lemma about the synchronous grid of Continental Europe (also known as Continental Synchronous Area; formerly known as the UCTE grid) which is the largest synchronous electrical grid (by connected power) in the world. Apparently there was a serious decrease in frequency in 2018 over a longer period of time which caused clocks running of 50Hz mains to deviate 5 minutes from the actual time: https://www.entsoe.eu/news/2018/03/03/frequency-deviations-in-continental-europe-including-impact-on-electric-clocks-steered-by-frequency/ This was because of a South-European operator which struggled with delivery of energy. After that the frequency was increased for some time to get all the clocks back in sync.
LIVE view on the state of the European grid energy flows:
“ that frequency is maintained accurately enough over time that it can be used as a time reference for a clock”
Sadly not… a friend who worked at national grid said they were often quite significantly out. If they got far enough out they just “reset” it back to zero from an atomic clock.
I own a mains clock (built into a beer sign) and I can confirm it’s highly inaccurate. I have seen AC powered clocks that use a quartz movement; I’m assuming for this reason.
I thought I read here that there is a proposal to do away with this timebase correction? Much like Trump wanted to get rid of WWV the timebase on SW.
@echodelta said: “I thought I read here that there is a proposal to do away with this timebase correction? Much like Trump wanted to get rid of WWV the timebase on SW.”
(A) There is an effort to keep the mains frequency within set limits, +/-10mHz for example. But there is no effort made over the long term to correct the mains frequency so it is truly useful as a time reference.
(B) The Trump Administration saved the NIST time-standard radio stations in Colorado (WWV & WWVB) and Hawaii (WWVH), it did not try to get rid of them. The proposal to defund the stations came in the NIST FY 2019 Budget Request issued by the office of NIST’s Director. During the review of the draft F.Y. 2019 budget, it was pointed out that the plan to turn off the NIST time-standard radio stations was a bad idea, at many levels. The Trump Administration agreed; the funding for the stations was preserved.
1. Measurement of the mains frequency (EU)
“The primary control is the first step in the mechanism of bringing the frequency back to the 50.0 Hz. If the deviation from the nominal value exceeds ±10 mHz, then the primary control is activated. In a range of ±10 mHz the frequency can flow free, above or underneath this value the primary control is activated linear. This 10 mHz is the same as the allowed measurement error of 10 mHz, to prevent primary control from running with a false sign.”
2. NIST FY 2019 Budget Request
“Budget Request – NIST requests a total of $127.0 million to support core measurement science programs advancing the precision, accuracy and comparability of the measurements that underpin the U.S. economy and innovation ecosystem. The FY 2019 request is a net decrease of $49.0 million from FY 2018 levels. Illustrative program reductions in FY 2019 – -$6.3 million supporting fundamental measurement dissemination, including the shutdown of NIST radio stations in Colorado and Hawaii”
3. WWV (radio station) – Time signal transmissions
“WWV, along with WWVB and WWVH, was recommended for defunding and elimination in NIST’s Fiscal Year 2019 budget request. However, the final 2019 NIST budget preserved funding for the three stations.”
@ echodeltaIs that a politically motivated comment or a verifiable fact? As far as I am aware, Trump’a administration actually preserved the NIST time standard radio stations. And, for the avoidance of doubt I live in the UK and have no interest at all in who is the US president as long as they don’t want to involve me in yet another senseless war.
I’m not sure how precisely they want to measure, but is a GPS *really* necessary? This is more than $60 of hardware for 60hz
@Y said: “I’m not sure how precisely they want to measure, but is a GPS *really* necessary? This is more than $60 of hardware for 60hz”
(A) “…is a GPS *really* necessary?” In a mains power station or substation, GPS is not strictly necessary, but it is standard practice to use GPS-disciplined time-bases in the power grid. If the GPS reference fails, the time base will continue to operate in “holdover” mode using an oven stabilized crystal oscillator. Eventually though, the GPS reference should be restored.
(B) “This is more than $60 of hardware for 60hz” A GPS-disciplined time-base in a primary mains power station is quite sophisticated and costs a LOT more than $60! Think many thousands of dollars for a 1-for-1 auto-redundant GPS-disciplined time-base with VCOCXO-holdover and NTP plus SNMP network capability. And then there is the significant cost of installing the subsystem and integrating it with the power station’s Monitor and Control (M&C) system.
GPS-disciplined time-bases (clocks) often found in U.S. mains power stations are manufactured by Arbiter Systems, Inc. In a primary power station the Arbiter Model 1088B GPS Satellite Clock (40 ns) is often found. In substations the less expensive Model 1094B GPS Substation Clock (250 ns) is common.
1. Model 1088B GPS Satellite Clock (40 ns)
2. Model 1094B GPS Substation Clock (250 ns)
It’s about $50 — $10 for the R Pi, $20 Teensy, $10 GPS plus some smaller components. The GPS has zero long term drift, and is about 20 ppb in the short term. the Teensy’s internal XTAL is about 20 ppm. The project was interested in measuring cumulative time drift, and also possibly comparing 60 Hz phase from different locations (synchronized via GPS).
@jp3141 said: “It’s about $50 — $10 for the R Pi, $20 Teensy, $10 GPS plus some smaller components. The GPS has zero long term drift, and is about 20 ppb in the short term. the Teensy’s internal XTAL is about 20 ppm. The project was interested in measuring cumulative time drift, and also possibly comparing 60 Hz phase from different locations (synchronized via GPS).”
You should not attempt to strictly control the mains grid frequency using a $50 GPS and RPi dongle (or any GPS referenced system for that matter). There is far more going on than just obtaining a stable reference and brute-forcing the grid frequency through generator control. The frequency must be allowed to vary with the load, typically the higher the load the lower the frequency. Watching the grid frequency is one way the network operators gauge load magnitude and distribution. The GPS reference helps to establish a norm against which attempts are made to keep the line frequency within a reasonable range, +/-10mHz during normal operation. See my previous post in this thread for further explanation and a citation.
Please be kind and respectful to help make the comments section excellent. (Comment Policy)