help on monitoring power frequency

[netotse]

hello guys,
i need help with my project i want to monitor the frequency of the power system, basically the project is to create a device that monitors the frequency of the power system then when the frequency falls below a certain value(say 49.0 cos the frequency in my country is 50hz)it would disconnect some certain loads...(i've also thought about setting different frequencies for differnet loads i.e at say 49.0 disconnect one device at 48.5 disconnect another)...
the part i need help with is the monitoring frequency bit.

 

[Hero999]

Where do you live?

Is this running off a generator?

Where I live the mains frequency is pretty accurately controlled, so much that it's used as a timebase for some clocks.

 

[KMoffett]

Almost all clocks that plug into a wall outlet use the line frequency as their timing source (think syncronous motors)...even digital, ones like clock radios. Though the instantaneous frequency may vary slightly, power plants compare the generator's revolutions against "tracable" clocks, and adjust the speed to add or subtract revolutions for exact cycle counts over long periods.

Ken

 

[Hero999]

Also, most motors and transformers should work down to 45Hz or so with no problems.

 

[mvs sarma]

hello guys,
i need help with my project i want to monitor the frequency of the power system, basically the project is to create a device that monitors the frequency of the power system then when the frequency falls below a certain value(say 49.0 cos the frequency in my country is 50hz)it would disconnect some certain loads...(i've also thought about setting different frequencies for differnet loads i.e at say 49.0 disconnect one device at 48.5 disconnect another)...
the part i need help with is the monitoring frequency bit.

By disconnecting loads to maintain frequency, perhaps your country will work at 30% loading of your generators. You please try to appreciate the multiple generator grid controlled power distributions. The frequency is far more common across your company. The frequency may slightly vary +/- 2 cycles depending on peak load to slack-load times. As pointed out by Hero999, it should not be a problem for motors unless you are speed sensitive. In cases where speed is very important, you may go for controller based motor drives where the frequency is closely controlled to maintain speed.

 

[Rolf]

Theoretical question........

hello guys,
i need help with my project i want to monitor the frequency of the power system, basically the project is to create a device that monitors the frequency of the power system then when the frequency falls below a certain value(say 49.0 cos the frequency in my country is 50hz)it would disconnect some certain loads...(i've also thought about setting different frequencies for differnet loads i.e at say 49.0 disconnect one device at 48.5 disconnect another)...
the part i need help with is the monitoring frequency bit.

This is a theoretical question, correct?
And you want us to help you with your homework, correct?

 

[netotse]

clarification

This is a theoretical question, correct?
And you want us to help you with your homework, correct?

nope...sorry if i wasnt clear enough...

in my country the frequency varies a lot(we dont generate nearly enough electricity so we're in a state of constant load shedding! the frequency is anything but constant it can go as low as 48 and as high as 51.5 all in a matter of minutes)we have underfrequency relays at some points in the grid that disconnect loads at around 48.5hz but sometimes it leads to system collapses cos it might cause swings so i'm working on the premise that if i can build something that disconnect domestic loads it'd help reduce the load on the system seeing as frequency control is related to the real power output of a generator...i want to look into taking out some domestic loads(airconditioners,heaters etc) in a large number of houses as an alterative to disconnecting a section of the grid...

 

[crutschow]

The simplest way is probably to program a microcontroller to count the power line frequency for say 10 seconds which would give you a count of 500 for a resolution of 0.1Hz (or 100 seconds for a 0.01Hz resolution). You can then do a comparison of this count to see if it's within limits and output a signal accordingly.

You can also drive a display to give a digital readout of the frequency, if desired.

You will need to generate logic level pulses from the power line for the uC input. This can be done from a low voltage, isolated transformer output (perhaps the same one that powers the uC). Run the signal into a comparator (with some feedback hysteresis to minimize effects of power line noise) to generate the 50Hz pulses.

I did a similar thing with a Stamp controller to monitor drift in the power line frequency, which was easy to program in Basic. Basic is much easier than Assembly if you've never programmed before.

 

[Rolf]

nope...sorry if i wasnt clear enough...

in my country the frequency varies a lot(we dont generate nearly enough electricity so we're in a state of constant load shedding! the frequency is anything but constant it can go as low as 48 and as high as 51.5 all in a matter of minutes)we have underfrequency relays at some points in the grid that disconnect loads at around 48.5hz but sometimes it leads to system collapses cos it might cause swings so i'm working on the premise that if i can build something that disconnect domestic loads it'd help reduce the load on the system seeing as frequency control is related to the real power output of a generator...i want to look into taking out some domestic loads(airconditioners,heaters etc) in a large number of houses as an alterative to disconnecting a section of the grid...

Well, now it makes sense but why couldn't this been explained in the original posting it would have made your requirement much clearer.
And filling in your location wouldn't hurt either.

 

[Rolf]

The simplest way is probably to program a microcontroller to count the power line frequency for say 10 seconds which would give you a count of 500 for a resolution of 0.1Hz (or 100 seconds for a 0.01Hz resolution). You can then do a comparison of this count to see if it's within limits and output a signal accordingly.

You can also drive a display to give a digital readout of the frequency, if desired.

You will need to generate logic level pulses from the power line for the uC input. This can be done from a low voltage, isolated transformer output (perhaps the same one that powers the uC). Run the signal into a comparator (with some feedback hysteresis to minimize effects of power line noise) to generate the 50Hz pulses.

I did a similar thing with a Stamp controller to monitor drift in the power line frequency, which was easy to program in Basic. Basic is much easier than Assembly if you've never programmed before.

For a faster response if that is needed, just modify the frequency counter to period type of counter and you can have just about instentines feedback.

 

[ericgibbs]

For a faster response if that is needed, just modify the frequency counter to period type of counter and you can have just about instentines feedback.

Hi Roff,
I would also suggest this period measurement.

In the LM393 datasheet is 'fair' zero crossing comparator.

OT:
When I moved here, I also found an Irish penny in the garden, no wonder they were so poor,
its because they had holes in their pockets..

 

[netotse]

guys thanks for the replies i have a much clearer picture of what i want to do in my head now(it'll require quite a bit of reading up though but i'm grateful for the help i've recieved so far) i think assembly would be more of an advantage in my case cos then i'll get less flak from my lecturers

could someone please shed more light on the bit about programing a microcontroller to count the powerline frequency(or period as Rolf suggested) as i know very little about microcontrollers

@Rolf
am in Nigeria and here our electricity sector is something to be ashamed of so sorry if i seemd evasive!

 

[ericgibbs]

could someone please shed more light on the bit about programing a microcontroller to count the powerline frequency(or period as Rolf suggested) as i know very little about microcontrollers

hi,

Use the PIC's compare option with the internal Timers and an external zero crossing comparator
to measure the period of the mains half cycle.

 

[eblc1388]

This problem is purely academic.

Virtually every power generating system connecting to the grid is fitted with underfrequency protection system made up of protection relays built solely for that purpose.

One don't usually "build" a device to replace those standard protection relays, which had been tested vigorously under severe system disturbance conditions when they are depended upon to function correctly.

However, if one wants to see how an alternative detecting system performs than those existing relays, then that's another question.

 

[crutschow]

For a faster response if that is needed, just modify the frequency counter to period type of counter and you can have just about instentines feedback.

The period method can suffer inaccuracies from noise on the line, and power lines often have large amounts of noise from universal motors, SCR dimmers, etc. That's why I suggested counting the cycles for a least 10 seconds. Power line frequency doesn't change that fast anyway, so 10 seconds is plenty of time to respond to any change.

 

[ericgibbs]

The period method can suffer inaccuracies from noise on the line, and power lines often have large amounts of noise from universal motors, SCR dimmers, etc. That's why I suggested counting the cycles for a least 10 seconds. Power line frequency doesn't change that fast anyway, so 10 seconds is plenty of time to respond to any change.

hi,
Perhaps a combination of period and frequency measured over 10 sec, with a rolling average.
Most equipment is able to cope with a mains freq ranging from 40 Hz thru 65 Hz without any problems, especially items like lights, ovens/heaters etc.

What equipment is the OP trying to protect.?

 

[Ubergeek63]

a micro running off a 32KHz watch crystal is better than 0.5% accurate at measuring the period of half cycles if the micro compensates for temperature variation. It should be noted that this kind of accuracy is limited by the repeatability of the interrupt service latency, a problem which goes away with a capture and compare register.

Fancy comparators are not needed all you need is an extra diode between the bridge and the filter cap and a transistor "comparing" the bridge output to the micro's supply voltage.

An all logic version, if you do not need a readout, is to run a simple "up" counter and compare the output to fixed values in logic or a ROM and latch the ON/OFF signal every half cycle before clearing the counter for the next measurement. A 512 byte ROM is sufficient to have an address for each half cycle down to 30Hz.

 

[Dean Huster]

I made a frequency meter using the F/V converter chip common to the old 5.25" disc drives. Ended up with a 1ma meter displaying a few Hz either side of 60Hz. Good for instantaneous indications, cheap, accurate, etc. However, I can't find my original circuit! Still looking. It was designed for monitoring the output of a portable generator.

Dean

 

[netotse]

This problem is purely academic.

Virtually every power generating system connecting to the grid is fitted with underfrequency protection system made up of protection relays built solely for that purpose.

One don't usually "build" a device to replace those standard protection relays, which had been tested vigorously under severe system disturbance conditions when they are depended upon to function correctly.

However, if one wants to see how an alternative detecting system performs than those existing relays, then that's another question.

these relays take out sections of the grid to maintain stability (i.e. whole regions are disconnected) i've seen under frequency relays that operated 20times in a day,now imagine having your entire state lose supply 20times in a day(even though some pple in my country can go 20days without power supply)
wot my device is meant to do is to shed small doestic loads in many places(working on the principle of little drops making an ocean)

 

[netotse]

The period method can suffer inaccuracies from noise on the line, and power lines often have large amounts of noise from universal motors, SCR dimmers, etc. That's why I suggested counting the cycles for a least 10 seconds. Power line frequency doesn't change that fast anyway, so 10 seconds is plenty of time to respond to any change.

actually in my country does(...i didnt know this until i did my industrial training (internship)in a power station) i think it's cause the governors for the dams are not working so load regulation is done by mouth instead of by computers and stuff

the period method looks really attractive from this side. Could someone please shed more light on the 'interrupt service latency' ubergeek was talking about.