Frequency standard

S

Suzy

Guest
I am needing to generate about 1v p-to-p at 50Hz sine wave as accurately as
I can. Please don't suggest syncing to mains as that's the application -- to
calibrate an instrument to measure the actual variance of the mains from
time to time from 50 Hz precisely.

My current (ouch) thoughts are to get a 2 MHz crystal and divide down to
50Hz. Any comments on this idea or an alternative?
 
Suzy wrote:

I am needing to generate about 1v p-to-p at 50Hz sine wave as accurately as
I can. Please don't suggest syncing to mains as that's the application -- to
calibrate an instrument to measure the actual variance of the mains from
time to time from 50 Hz precisely.

My current (ouch) thoughts are to get a 2 MHz crystal and divide down to
50Hz. Any comments on this idea or an alternative?
Click to expand...
I wouldn't trust a 'plain' crystal on its own, it's not good enough to
compare against mains.

There are several options.

One "problem" that comes to mind is temperature stability.

To work around that, there are temperature compensated crystal oscillators,
that are pretty good (and cost more), and oven temperature regulated crystal
oscillators that are better (and cost lots more). I'm sure both are available
as modules, to make it easy to integrate into your design, but I can't think
of any off the top of my head.

I also vaguely recall hearing of "pre-aged" crystals too, so they're less
likely to drift as far due to ageing than new crystals. Basically, this means
it drifts less between each calibration interval (well, the first few anyway).


Would be best for you to come down to numbers, that is, find out what
stability you *DO* need, and go shopping from there.
You would need short term stability for this, for mains monitoring work,
this is vital. Long term stability would be important if you need to go long
periods of time between calibrations.
--
Linux Registered User # 302622
<http://counter.li.org>
 
"John Tserkezis" <jt@techniciansyndrome.org.invalid> wrote in message
news:47c10279$0$8437$afc38c87@news.optusnet.com.au...
Suzy wrote:

I am needing to generate about 1v p-to-p at 50Hz sine wave as accurately
as I can. Please don't suggest syncing to mains as that's the
application -- to calibrate an instrument to measure the actual variance
of the mains from time to time from 50 Hz precisely.

My current (ouch) thoughts are to get a 2 MHz crystal and divide down to
50Hz. Any comments on this idea or an alternative?

I wouldn't trust a 'plain' crystal on its own, it's not good enough to
compare against mains.

There are several options.

One "problem" that comes to mind is temperature stability.

To work around that, there are temperature compensated crystal
oscillators, that are pretty good (and cost more), and oven temperature
regulated crystal oscillators that are better (and cost lots more). I'm
sure both are available as modules, to make it easy to integrate into your
design, but I can't think of any off the top of my head.

I also vaguely recall hearing of "pre-aged" crystals too, so they're less
likely to drift as far due to ageing than new crystals. Basically, this
means it drifts less between each calibration interval (well, the first
few anyway).


Would be best for you to come down to numbers, that is, find out what
stability you *DO* need, and go shopping from there.
You would need short term stability for this, for mains monitoring work,
this is vital. Long term stability would be important if you need to go
long periods of time between calibrations.
--
Linux Registered User # 302622
http://counter.li.org
Click to expand...
Thanks for that. I've been looking at 2meg crystals with accuracy of
30herz. Wouldn't that give me plus/minus .06 Hz @ 50Hz?
 
"Suzy"

Thanks for that. I've been looking at 2meg crystals with accuracy of
30herz. Wouldn't that give me plus/minus .06 Hz @ 50Hz?
Click to expand...

** Huh ??

What sort of math is that ?

30 / 2exp6 times 50 = 0.00075



....... Phil
 
"Suzy" <not@valid> wrote in message news:47c103ad$1@dnews.tpgi.com.au...
Thanks for that. I've been looking at 2meg crystals with accuracy of
30herz. Wouldn't that give me plus/minus .06 Hz @ 50Hz?
Click to expand...
0.075Hz, at the specified temperature.

MrT.
 
"Mr.Turd"
"Suzy"

Thanks for that. I've been looking at 2meg crystals with accuracy of
30herz. Wouldn't that give me plus/minus .06 Hz @ 50Hz?

0.075Hz, at the specified temperature.
Click to expand...

** Get real.

You have quoted a percentage as a simple ratio.




........ Phil
 
"John Tserkezis" <jt@techniciansyndrome.org.invalid> wrote in message
news:47c10279$0$8437$afc38c87@news.optusnet.com.au...
Suzy wrote:

I am needing to generate about 1v p-to-p at 50Hz sine wave as accurately
as I can. Please don't suggest syncing to mains as that's the
application -- to calibrate an instrument to measure the actual variance
of the mains from time to time from 50 Hz precisely.

My current (ouch) thoughts are to get a 2 MHz crystal and divide down to
50Hz. Any comments on this idea or an alternative?

I wouldn't trust a 'plain' crystal on its own, it's not good enough to
compare against mains.

There are several options.

One "problem" that comes to mind is temperature stability.

To work around that, there are temperature compensated crystal
oscillators, that are pretty good (and cost more), and oven temperature
regulated crystal oscillators that are better (and cost lots more). I'm
sure both are available as modules, to make it easy to integrate into your
design, but I can't think of any off the top of my head.

I also vaguely recall hearing of "pre-aged" crystals too, so they're less
likely to drift as far due to ageing than new crystals. Basically, this
means it drifts less between each calibration interval (well, the first
few anyway).


Would be best for you to come down to numbers, that is, find out what
stability you *DO* need, and go shopping from there.
You would need short term stability for this, for mains monitoring work,
this is vital. Long term stability would be important if you need to go
long periods of time between calibrations.
--
Linux Registered User # 302622
http://counter.li.org
Click to expand...
2MHz divided down to 50Hz would give a possible inaccuracy drift of 0.0025%,
surely close enough???
What is the "real" purpose of this exercise in relationship to 50Hz? Prove
the electricity provider is 'off' frequency?
If so, what is the point? Sue them for being 0.000025Hz off? Your mains
synchronised alarm clock was out so you missed your bus or train by a few
microseconds? ROTFLMAO
 
"Alan Rutlidge" <don't_spam_me_rutlidge@iinet.net.au> wrote in message
news:47c113f9$0$27293$5a62ac22@per-qv1-newsreader-01.iinet.net.au...
2MHz divided down to 50Hz would give a possible inaccuracy drift of
0.0025%,
surely close enough???
What is the "real" purpose of this exercise in relationship to 50Hz?
Prove
the electricity provider is 'off' frequency?
If so, what is the point? Sue them for being 0.000025Hz off?
Click to expand...
No point in doing calculations when you just can pick figures out of the air
eh?

MrT.
 
Suzy wrote:
My current (ouch) thoughts are to get a 2 MHz crystal and divide down to
50Hz. Any comments on this idea or an alternative?
Click to expand...
Crystals has an inverse parabolic frequency variation with temperature.
If you know or can measure the parameters of that parabola, you can
measure the temperature and compensate the timings down to the order
of a couple of seconds per month.

That might be useful for a 0.000001Hz standard, though not as a 50Hz
frequency standard. Depending on exactly what you want, you might be
able to get away with counting 50Hz cycles over a longer (accurate)
time, to determine variance from the nominal frequency.

Clifford Heath.
 
Suzy wrote:
I am needing to generate about 1v p-to-p at 50Hz sine wave as accurately as
I can. Please don't suggest syncing to mains as that's the application -- to
calibrate an instrument to measure the actual variance of the mains from
time to time from 50 Hz precisely.

My current (ouch) thoughts are to get a 2 MHz crystal and divide down to
50Hz. Any comments on this idea or an alternative?


Hi Suzy,
Click to expand...

We have used these temperature compensated oscillators from Hy-Q.
Significantly better than a normal crystal.

http://www.hy-q.com.au/pdf/10120029.pdf


David
 
"Alan Peake"
Depends on how accurately you want to measure the mains frequency. Last
time I did this (in SA), the mains drifted by +/- 1.5 Hz with a period of
about 20 minutes.
Click to expand...

** Huh ???

That is way more than it actually varies.

You method must have been flawed.




....... Phil
 
"Alan Rutlidge" <don't_spam_me_rutlidge@iinet.net.au> wrote in message
news:47c113f9$0$27293$5a62ac22@per-qv1-newsreader-01.iinet.net.au...
"John Tserkezis" <jt@techniciansyndrome.org.invalid> wrote in message
news:47c10279$0$8437$afc38c87@news.optusnet.com.au...
Suzy wrote:

I am needing to generate about 1v p-to-p at 50Hz sine wave as accurately
as I can. Please don't suggest syncing to mains as that's the
application -- to calibrate an instrument to measure the actual variance
of the mains from time to time from 50 Hz precisely.

My current (ouch) thoughts are to get a 2 MHz crystal and divide down to
50Hz. Any comments on this idea or an alternative?

I wouldn't trust a 'plain' crystal on its own, it's not good enough to
compare against mains.

There are several options.

One "problem" that comes to mind is temperature stability.

To work around that, there are temperature compensated crystal
oscillators, that are pretty good (and cost more), and oven temperature
regulated crystal oscillators that are better (and cost lots more). I'm
sure both are available as modules, to make it easy to integrate into
your design, but I can't think of any off the top of my head.

I also vaguely recall hearing of "pre-aged" crystals too, so they're
less likely to drift as far due to ageing than new crystals. Basically,
this means it drifts less between each calibration interval (well, the
first few anyway).


Would be best for you to come down to numbers, that is, find out what
stability you *DO* need, and go shopping from there.
You would need short term stability for this, for mains monitoring work,
this is vital. Long term stability would be important if you need to go
long periods of time between calibrations.
--
Linux Registered User # 302622
http://counter.li.org

2MHz divided down to 50Hz would give a possible inaccuracy drift of
0.0025%, surely close enough???
What is the "real" purpose of this exercise in relationship to 50Hz?
Prove the electricity provider is 'off' frequency?
If so, what is the point? Sue them for being 0.000025Hz off? Your mains
synchronised alarm clock was out so you missed your bus or train by a few
microseconds? ROTFLMAO


I think you've missed the point Alan! The exact frequency of the mains gives
Click to expand...
some indication of the demand/supply balance. It is imposible to maintain
exactly 50 Hz unless supply and demand are perfectly balanced, which is a a
rare event. There are in fact legal limits and overall there is an attempt
by the managers of the grid to compensate so that clocks stay pretty well in
time over the long term. However, it is still informative to see
fluctuations as demand peak and more generators are brought onto line, and
vice versa.
 
"Phil Allison" <philallison@tpg.com.au> wrote in message
news:62d1gpF22l33rU1@mid.individual.net...
"Alan Peake"

Depends on how accurately you want to measure the mains frequency. Last
time I did this (in SA), the mains drifted by +/- 1.5 Hz with a period of
about 20 minutes.


** Huh ???

That is way more than it actually varies.

You method must have been flawed.




...... Phil
Click to expand...
Agreed Phil. And for all those interested, see this site for the UK grid.
http://www.dynamicdemand.co.uk/grid.htm#
Click to expand...
 
On 24/02/2008 22:08 Phil Allison wrote:
"Alan Peake"
Depends on how accurately you want to measure the mains frequency. Last
time I did this (in SA), the mains drifted by +/- 1.5 Hz with a period of
about 20 minutes.


** Huh ???

That is way more than it actually varies.

You method must have been flawed.




...... Phil
Click to expand...

Remember the old DATUM microprocessor training board which was in EA
around 1982?
I wrote some code for mine to watch the mains frequency and it never
deviated from 50Hz by more than a tiny part of 1Hz.
That was 25 years ago and I don't recall the exact figures but it
was far less than 1.5Hz.


Bob
 
On Feb 24, 3:39 pm, "Suzy" <not@valid> wrote:
I am needing to generate about 1v p-to-p at 50Hz sine wave as accurately as
I can. Please don't suggest syncing to mains as that's the application -- to
calibrate an instrument to measure the actual variance of the mains from
time to time from 50 Hz precisely.

My current (ouch) thoughts are to get a 2 MHz crystal and divide down to
50Hz. Any comments on this idea or an alternative?
Click to expand...
What is your real application here?
If it's just to get an occasional spot frequency check then your
crystal should do just fine.
But the easiest way to do that is to simply use a frequency counter
that has reciprocal measurement (to get a lot of resolution at low
frequencies like 50Hz).

If you are talking long term drift accuracy and/or logging then that
is a different ball game.

Dave.
 
Suzy wrote:
"Alan Rutlidge" <don't_spam_me_rutlidge@iinet.net.au> wrote in message
news:47c113f9$0$27293$5a62ac22@per-qv1-newsreader-01.iinet.net.au...
"John Tserkezis" <jt@techniciansyndrome.org.invalid> wrote in message
news:47c10279$0$8437$afc38c87@news.optusnet.com.au...
Suzy wrote:

I am needing to generate about 1v p-to-p at 50Hz sine wave as accurately
as I can. Please don't suggest syncing to mains as that's the
application -- to calibrate an instrument to measure the actual variance
of the mains from time to time from 50 Hz precisely.
My current (ouch) thoughts are to get a 2 MHz crystal and divide down to
50Hz. Any comments on this idea or an alternative?
I wouldn't trust a 'plain' crystal on its own, it's not good enough to
compare against mains.

There are several options.

One "problem" that comes to mind is temperature stability.

To work around that, there are temperature compensated crystal
oscillators, that are pretty good (and cost more), and oven temperature
regulated crystal oscillators that are better (and cost lots more). I'm
sure both are available as modules, to make it easy to integrate into
your design, but I can't think of any off the top of my head.

I also vaguely recall hearing of "pre-aged" crystals too, so they're
less likely to drift as far due to ageing than new crystals. Basically,
this means it drifts less between each calibration interval (well, the
first few anyway).


Would be best for you to come down to numbers, that is, find out what
stability you *DO* need, and go shopping from there.
You would need short term stability for this, for mains monitoring work,
this is vital. Long term stability would be important if you need to go
long periods of time between calibrations.
--
Linux Registered User # 302622
http://counter.li.org
2MHz divided down to 50Hz would give a possible inaccuracy drift of
0.0025%, surely close enough???
What is the "real" purpose of this exercise in relationship to 50Hz?
Prove the electricity provider is 'off' frequency?
If so, what is the point? Sue them for being 0.000025Hz off? Your mains
synchronised alarm clock was out so you missed your bus or train by a few
microseconds? ROTFLMAO


I think you've missed the point Alan! The exact frequency of the mains gives
some indication of the demand/supply balance. It is imposible to maintain
exactly 50 Hz unless supply and demand are perfectly balanced, which is a a
rare event. There are in fact legal limits and overall there is an attempt
by the managers of the grid to compensate so that clocks stay pretty well in
time over the long term. However, it is still informative to see
fluctuations as demand peak and more generators are brought onto line, and
vice versa.
Click to expand...
Hi Suzy

The grid may be slow by two or three seconds during the day and they
usually catch up at night. We've just upgraded to a Siemens T3000
control system and we use a GPS for time keeping. I think Silicon Chip
published a project recently using a GPS as a frequency standard.

Elmo
 
Depends on how accurately you want to measure the mains frequency. Last
time I did this (in SA), the mains drifted by +/- 1.5 Hz with a period
of about 20 minutes. Your 2 MHz crystal should be good enough for that.
Alan


Suzy wrote:
I am needing to generate about 1v p-to-p at 50Hz sine wave as accurately as
I can. Please don't suggest syncing to mains as that's the application -- to
calibrate an instrument to measure the actual variance of the mains from
time to time from 50 Hz precisely.

My current (ouch) thoughts are to get a 2 MHz crystal and divide down to
50Hz. Any comments on this idea or an alternative?
Click to expand...
 
"Elmo" <elmo@sesame.com> wrote in message
news:47c1677c$1_7@news.peopletelecom.com.au...
Suzy wrote:
"Alan Rutlidge" <don't_spam_me_rutlidge@iinet.net.au> wrote in message
news:47c113f9$0$27293$5a62ac22@per-qv1-newsreader-01.iinet.net.au...
"John Tserkezis" <jt@techniciansyndrome.org.invalid> wrote in message
news:47c10279$0$8437$afc38c87@news.optusnet.com.au...
Suzy wrote:

I am needing to generate about 1v p-to-p at 50Hz sine wave as
accurately as I can. Please don't suggest syncing to mains as that's
the application -- to calibrate an instrument to measure the actual
variance of the mains from time to time from 50 Hz precisely.
My current (ouch) thoughts are to get a 2 MHz crystal and divide down
to 50Hz. Any comments on this idea or an alternative?
I wouldn't trust a 'plain' crystal on its own, it's not good enough to
compare against mains.

There are several options.

One "problem" that comes to mind is temperature stability.

To work around that, there are temperature compensated crystal
oscillators, that are pretty good (and cost more), and oven temperature
regulated crystal oscillators that are better (and cost lots more).
I'm sure both are available as modules, to make it easy to integrate
into your design, but I can't think of any off the top of my head.

I also vaguely recall hearing of "pre-aged" crystals too, so they're
less likely to drift as far due to ageing than new crystals.
Basically, this means it drifts less between each calibration interval
(well, the first few anyway).


Would be best for you to come down to numbers, that is, find out what
stability you *DO* need, and go shopping from there.
You would need short term stability for this, for mains monitoring
work, this is vital. Long term stability would be important if you
need to go long periods of time between calibrations.
--
Linux Registered User # 302622
http://counter.li.org
2MHz divided down to 50Hz would give a possible inaccuracy drift of
0.0025%, surely close enough???
What is the "real" purpose of this exercise in relationship to 50Hz?
Prove the electricity provider is 'off' frequency?
If so, what is the point? Sue them for being 0.000025Hz off? Your
mains synchronised alarm clock was out so you missed your bus or train
by a few microseconds? ROTFLMAO


I think you've missed the point Alan! The exact frequency of the mains
gives some indication of the demand/supply balance. It is imposible to
maintain exactly 50 Hz unless supply and demand are perfectly balanced,
which is a a rare event. There are in fact legal limits and overall there
is an attempt by the managers of the grid to compensate so that clocks
stay pretty well in time over the long term. However, it is still
informative to see fluctuations as demand peak and more generators are
brought onto line, and vice versa.



Hi Suzy

The grid may be slow by two or three seconds during the day and they
usually catch up at night. We've just upgraded to a Siemens T3000 control
system and we use a GPS for time keeping. I think Silicon Chip published a
project recently using a GPS as a frequency standard.

Elmo
Click to expand...
Yes, Elmo, that's right. But it's not the overall slowness during a typical
high load day that interests me (which as you say is made up during the
night). It is the constant fluctuations that illustrate changes in the
supply/demand balance, which promise to be more dramatic as consumption
grows and various political (state, national and international) effects are
felt. As Phil intimates, such variations as exist are (currently anyway)
very small and fractions of a Herz, but they are there nonetheless. The link
I posted gives a most interesting monitoring of the UK grid. Significant
loads *do* drag down grid speed, and surplus supply *does* ramp it up,
albeit by very small amounts. In fact there are legal limits for the
variation (someone here will no doubt know them) but legal limits are not
brick walls... What I have in mind is somewhat of a "Grid stress" indicator.
Some time in the future, it may serve to give early warning of rolling
blackouts.

I'll look up that Silicon Chip project. Certainly to make an accurate unit
would require a very good calibration sources.
 
"Suzy"
"Phil Allison"
"Alan Peake"

Depends on how accurately you want to measure the mains frequency. Last
time I did this (in SA), the mains drifted by +/- 1.5 Hz with a period
of about 20 minutes.


** Huh ???

That is way more than it actually varies.

You method must have been flawed.


Agreed Phil. And for all those interested, see this site for the UK grid.

http://www.dynamicdemand.co.uk/grid.htm#
Click to expand...

** That web sites created by complete amateurs - greenie loopy ones too.

The supply frequency is not so simply related to demand as they naively
claim.

IOW - it's bollocks.


........ Phil
 
"Phil Allison" <philallison@tpg.com.au> wrote in message
news:62ebiiF22u018U1@mid.individual.net...
"Suzy"
"Phil Allison"

"Alan Peake"

Depends on how accurately you want to measure the mains frequency. Last
time I did this (in SA), the mains drifted by +/- 1.5 Hz with a period
of about 20 minutes.


** Huh ???

That is way more than it actually varies.

You method must have been flawed.


Agreed Phil. And for all those interested, see this site for the UK grid.

http://www.dynamicdemand.co.uk/grid.htm#


** That web sites created by complete amateurs - greenie loopy ones too.

The supply frequency is not so simply related to demand as they naively
claim.

IOW - it's bollocks.


....... Phil

But isn't it true, Phil, that more demand than supply will drag down
Click to expand...
alternator shaft speed and thus frequency, and vice-versa? Apart from the
clock "make-up" events in times of slow load, what other factors would you
offer that influence mains frequency?
 
You must log in or register to reply here.

Welcome to EDABoard.com

Sponsor

Back
Top