frequency standard ideas ?

[Flump]

i use a precision 5v regulator and 0.01 % resistor to check my multimeters with
but i would like something to check the frequency accuracy of them (those that have freq testing)
and also to check the accuracy of my scope.

So I would like to make a very basic and simple frequency standard
not sure of the frequency yet, i guess it depends if there is cheap pre made
solution but anywhere from 100hz to 1mhz i think.

anyone got any idea for a simple diy solution ?

[dfmischler]

It depends a lot on how accurate you feel you need to be.  Will you be happy with 10 ppm (i.e. off by up to 10 Hz at 1 MHz)?  Or do you feel you need to be 1000 times that accurate?

Maybe this article will give you an idea.

[jpb]

If you want to avoid calibration, then you need to compare to a GPS which at the 1Hz (1pps) are cheap (£25 or so new) or possibly to a calibrated Rubidium off e-bay which are now rather expensive.

The U-blox LEA-6T is a timing GPS module that can be programmed to produce a frequency locked to the GPS from 0.25Hz to 10MHz though some frequencies have more jitter than others:
http://www.u-blox.com/images/downloads/Product_Docs/LEA-6_DataSheet_(GPS.G6-HW-09004).pdf
A module with the LEA-6T is around £100 off e-bay or a more sophisticated one is 181 Euro from sysmocom
http://shop.sysmocom.de/products/osmo-lea6t-gps
Note these modules have quite a lot of jitter, though for testing scopes you won't notice it. (My cheap GPS module's 1pps has less than 10nsecs of jitter so is good to 8 decimal places - I tested the frequency counter on my scope and it read 999.998 mHz which I was quite pleased about as the scope's spec is quite a bit wider than that.)

A cheaper option is to buy a OCXO from e-bay for around £10 to £20 but these will typically be 10MHz or 5MHz so you need to then build a divider circuit. Also the accuracy is not guaranteed so you may need to calibrate.

I have a TTi 930 counter so what I did was get a 10MHz OCXO off e-bay and use it as the reference to the counter and then adjust it until it read the 1pps from my GPS module as having a period of exactly 1 second to 10 digits.

[han]

just buy standard oscillator and divide it with flip flop or logic counter.
http://www.digikey.com/product-detail/en/MXO45HS-3C-4M0000/CTX742-ND/1801857

i can find one in old AT Motherboard(20Mhz one)

[olsenn]

A ovenized crystal is probably all you need (if you even need that). Atomic (Rubidium) oscillators have poor short-term stability and are probably not what you're looking for (unless you need like 12-digits of resolution).

[VK5RC]

+1 for the ovenised crystal, 10MHz is quite useful as later on you can use it as a frequency reference for other Lab equipment, some go quite cheaply on eBay if not an HP 8011 etc.
http://stores.ebay.com.au/Rubidium-and-GPS-DO-From-China/High-reliability-OCXO-/_i.html?_fsub=202159619.
I have bought from this guy before. If you have a friend with a good counter he can get its frequency very accurately for you.

[linux-works]

Atomic (Rubidium) oscillators have poor short-term stability and are probably not what you're looking for (unless you need like 12-digits of resolution).

I'm not sure I understand; why do you think Rb is bad short-term? 

[Flump]

thanks for all your input so far guys.

i stumbled across this on ebay
could i use this ?
http://www.ebay.co.uk/itm/1-4-8-10-12-15-16-20-25-27MHz-Crystal-Oscillator-Osc-5V-H-DIP-/181176601343?pt=UK_BOI_Electrical_Components_Supplies_ET&var=&hash=item2a2ef78aff

is it as simple as supplying it 5v and it outputs an accurate 1Mhz ?

[linux-works]

the canned oscillators are easy; power, ground and signal-out.

they may drift with voltage and temperature and you can buy various degrees of accuracy and precision.

if you are looking for a 'standard' though, an oven or tcxo would be better than a simple can.  plus, be careful buying 'precision' parts from ebay.  ebay is all about fakes, these days.  buy from a real distributor if you want to trust the device and be sure about it.

[jpb]

the canned oscillators are easy; power, ground and signal-out.

they may drift with voltage and temperature and you can buy various degrees of accuracy and precision.

if you are looking for a 'standard' though, an oven or tcxo would be better than a simple can.  plus, be careful buying 'precision' parts from ebay.  ebay is all about fakes, these days.  buy from a real distributor if you want to trust the device and be sure about it.

With OCXOs the ones on ebay are pretty old and one issue with them may be that the Voltage tuning range has reached its limit (the oscillator has drifted beyond the range over a period of say 20 years). But the ones that I've bought have been ok - the main one I have tunes to 10MHz at around 2.7V in a 0 to 5V tuning range (the crystal itself provides 5V from its internal regulator so you can tune it with just a pot.)

I think it is worth the gamble of getting OCXOx from ebay because they are generally about one tenth the cost of buying a new one. I recently bought a pair of 5MHz OCXOs on boards with tuning pots and rather nice minicircuit power splitters for £15 for the pair + postage bringing it up to £20 for the pair. I was the only person to bid but they work fine (I had to dig around to find what the components on the boards were and the specs and pin outs of the OCXOs themselves).

If you do a search on ebay on OCXO several come up. The 5MHz are cheaper than the 10MHz because most modern equipment wants 10MHz as a reference so they are more popular.

If you hang about a bit and are not in a rush then cheap and good condition local ones should appear from time to time if you don't want to risk importing from China.

EDIT : perhaps I should qualify my remarks a bit. There are drawbacks with buying low cost OCXOs from ebay. One is they might be physically broken (the oven component) or more likely the device can't be tuned to the exact frequency (perhaps not a problem if your just using it as a test rather than a reference for a PLL in an instrument). Secondly these older OCXOs tend to be larger and more power hungry. They also often require 12V or even 15V or more and draw close to an amp when the oven is first warming up, dropping later to 200-300mA. This isn't a problem if you have a bench power supply  and only want to power it up occasionally. It is more of a problem if you're trying to build a GPSDO you want to leave on for long periods (as I am) as the current draw is a bit non-green, though they are much less power hungry than a Rubidium.

[ejeffrey]

Atomic (Rubidium) oscillators have poor short-term stability and are probably not what you're looking for (unless you need like 12-digits of resolution).

I'm not sure I understand; why do you think Rb is bad short-term?

Most cheap rubidium clocks use the Rb output to discipline a standard quartz crystal oscillator.  The long term stability is great, but their phase noise in the 10 Hz to 1 MHz region is relatively poor and exactly what you would get from the same cheap crystal without the rubidium source.  This doesn't matter for making the time base to a frequency counter but if you use it as an external clock for e.g., a spectrum analyzer or a frequency synthesizer, that noise will affect the performance.  An OCXO has much lower phase noise, and is a better choice when spectral purity is more important than drift or absolute frequency accuracy.  Higher end Rb oscillators like the stanford research units use an OCXO to get the best of both worlds.

[linux-works]

while the Rb boxes are not technically oven controlled (most of the cheaper ones), they do have an oven effect in that they want to maintain a temperature and will fight to keep it.

I understand the phase noise isn't as good as possible, but for many home labs, its the best freq standard we'll likely see.

that said, some people want their home lab standards to run 7x24 and in that case, rb is entirely the wrong thing.  too power hungry, runs too hot and the bulb wears out sooner that way.

[G0HZU]

i use a precision 5v regulator and 0.01 % resistor to check my multimeters with
but i would like something to check the frequency accuracy of them (those that have freq testing)
and also to check the accuracy of my scope.

So I would like to make a very basic and simple frequency standard
not sure of the frequency yet, i guess it depends if there is cheap pre made
solution but anywhere from 100hz to 1mhz i think.

anyone got any idea for a simple diy solution ?

The following should be more than good enough for nearly everybody:

Find a surplus 10MHz OCXO on ebay or on an old PCB at an electronics fair. Try and get one with a digital output rather than a low level sine wave.

Power it from a decent regulated and stable supply and buffer it and then use it to clock a series of /2 and /5 counters in a few 74HC390 based divider chips. These chips cost about 50p each and can get you various division ratios and you can tap out the output of each divider.

So you could easily get 10MHz, 5MHz, 2MHz, 1MHz etc etc down to 100kHz from 1 74HC390 chip. Then go down another /100 in a second chip to reach 1kHz.

The downside of the OCXO is that it could take anything up to 45minutes to stabilise after you first power it. If this is not ideal for you then try to find a 10MHz TCXO. Obviously, you won't get the same stability or accuracy over time with the TCXO but it is still going to be good enough for most people.

There are 'fast' OCXOs available, eg I have one here that takes about 90 seconds to warm up but it obviously compromises the accuracy and stability in order to achieve this fast warmup time.

I wouldn't really recommend the cheapy £3 oscillator on ebay because it will have 50ppm temperature stability but this could be on top of a 100ppm basic accuracy. For maybe £10 you could get a fairly decent 10MHz TCXO or OCXO.

[linux-works]

save the dividers.

google 'picdiv' for an 8pin dip to do the dividing for you.

[dfmischler]

is it as simple as supplying it 5v and it outputs an accurate 1Mhz ?

Yes, it is very simple.  It will probably be fine if 50 ppm is accurate enough.

[gdewitte]

Perhaps this? http://www.aade.com/newfreq/new%20freq.htm

[dannyf]

I would like to make a very basic and simple frequency standard

The answer depends on what you meant by "frequency standard". Most (all?) of the multimeters utilized a crystal to clock its frequency measurement. That's typically 100ppm or better. So for measuring a typical LC/RC oscillator, it is more than sufficiently accurate, and nothing needs to be done there.

If you want more, you can utilize pre-made oscillators. Most of the crystal oscillators are spec'd to 50 or 100ppm, and they usually are good to 10-20ppm. They are dirt cheap and requires a simple power / ground connections to get going.

However, that approach really doesn't offer much beyond what your multi-meter can do. For better results, you may consider TCXO. Two flavors, analog-tuned or digitally-tuned. Buying an analog TCXO 2nd hand can be risky, if you don't have a high(er) accuracy frequency meter / counter to know exactly at what frequency it is oscillating. A digitally tuned TCXO needs no calibration and I would go with that. You can typically get within 10ppm spec-wise, and typically within 2-5ppm performance-wise - there are TCXOs that offer sub 1ppm spec but they are quite expensive.

Note: Some of today's TCXOs are VC-TCXO - they are not that useful for what you do.

OCXOs offer better performance but they consume a lot more current and are more expensive - I wouldn't buy them 2nd hand.

[Tac Eht Xilef]

A digitally tuned TCXO needs no calibration and I would go with that. You can typically get within 10ppm spec-wise, and typically within 2-5ppm performance-wise

Any devices you'd suggest / recommend in the 1-100MHz range? I've seen the Maxim DS3231 etc., but 32.768kHz is a bit low for anything other than RTC uses.

I'm not a time/frequency nut, but it'd be handy to have something like that in the parts box if the price is right.

[rfdave#gmail.com]

Atomic (Rubidium) oscillators have poor short-term stability and are probably not what you're looking for (unless you need like 12-digits of resolution).

I'm not sure I understand; why do you think Rb is bad short-term?

Most cheap rubidium clocks use the Rb output to discipline a standard quartz crystal oscillator.  The long term stability is great, but their phase noise in the 10 Hz to 1 MHz region is relatively poor and exactly what you would get from the same cheap crystal without the rubidium source. 

The actual Rb frequency measurement is at ~6.9 GHz, so all Rb clocks have an internal Quartz Oscillator that's locked to the Rb Oscillation. Expensive Rb clocks have better Quartz oscillators built in.

[Flump]

just looked ppm up on google and 50ppm = 0.005 i think ?
http://www.rapidtables.com/convert/number/PPM_to_Percent.htm

so 1Mhz could be
999.95Hz to 1000.05Hz
is that right ?


I Ordered this just to see what its like, its very cheap so
nothing lost if it doesnt work out right
http://www.ebay.co.uk/itm/181176601343?var=480257110155&ssPageName=STRK:MEWNX:IT&_trksid=p3984.m1439.l2649

[David Hess]

Atomic (Rubidium) oscillators have poor short-term stability and are probably not what you're looking for (unless you need like 12-digits of resolution).

I'm not sure I understand; why do you think Rb is bad short-term?

Most cheap rubidium clocks use the Rb output to discipline a standard quartz crystal oscillator.  The long term stability is great, but their phase noise in the 10 Hz to 1 MHz region is relatively poor and exactly what you would get from the same cheap crystal without the rubidium source.  This doesn't matter for making the time base to a frequency counter but if you use it as an external clock for e.g., a spectrum analyzer or a frequency synthesizer, that noise will affect the performance.  An OCXO has much lower phase noise, and is a better choice when spectral purity is more important than drift or absolute frequency accuracy.  Higher end Rb oscillators like the stanford research units use an OCXO to get the best of both worlds.

Their phase noise is poor because they generate a synthesized, usually DDS, reference output.  This is great if you want an adjustable output frequency but bad for phase noise compared to using a dedicated phase locked crystal oscillator which is unnecessary for most applications.

[Flump]

ok my oscilator came today
next day delivvery cant grumble with that
http://www.ebay.co.uk/itm/181176601343?var=480257110155&ssPageName=STRK:MEWNX:IT&_trksid=p3984.m1439.l2649

It works with my extech 330 dvm, that reports 999.6Hz
but its too output strength to read on my frequency counter.

could i use a 2n3904 to amplify it
and if so could anyone suggest a circuit please

[dannyf]

could i use a 2n3904 to amplify it

You probably want to make sure that it is an amplitude problem. Otherwise, use a transistor or a gate.

[jpb]

ok my oscilator came today
next day delivvery cant grumble with that
http://www.ebay.co.uk/itm/181176601343?var=480257110155&ssPageName=STRK:MEWNX:IT&_trksid=p3984.m1439.l2649

It works with my extech 330 dvm, that reports 999.6Hz
but its too output strength to read on my frequency counter.

could i use a 2n3904 to amplify it
and if so could anyone suggest a circuit please

According to the specs in the link you provided it should be producing a square wave between 0.4V and 90% of 5V = 4.5V at least, this should be plenty large enough for a counter to read - it might be too high.

Perhaps you need to ac couple it to the counter and even attenuate it a bit not amplify it.

[David Hess]

It works with my extech 330 dvm, that reports 999.6Hz
but its too output strength to read on my frequency counter.

What make and model of frequency counter is this?