Checking 10MHz OCXO frequency?

[mole42uk]

I have a piece of test equipment which uses a 10MHz oven-controlled crystal oscillator to provide all the critical timing. The oscilator is working, the test eqipment is working.

Is there an easy way to check the actual frequency of the oscillator? It needs to be correct to several decimal points!

[Terry_VK5TM]

Easy or accurate, take your pick of one.

If it really needs to be that accurate, then unless you have suitable calibrated test equipment, a calibration lab is going to be your best choice.

Even the old standby of zero beating the oscillator with WWV or some other time station will not get you that accurate (some claim to be able to get zero beat down to 1Hz or less, but I've never seen it done).

[Radio Wrangler]

Not easily.

You need a frequency source of known greater accuracy than what you're trying to set.

'several decimal points' doesn't mean much. The number of decimal places... but not saying whether that is MHz, kHz or Hz leaves the field rather open.

The easiest thing would be a GPS disciplined frequency standard, or something using the 1 pulse per second output from some GPS receivers.

There are also standard frequency transmissions (5 and 10MHz) which you could check for zero beat against using a communications receiver. However variation in propagation path length can make the frequency wander.

There are also standard freq transmissions at VLF which propagate more stably EG 60kHz 'MSF' from Anthorn in Cumbria. There are off-air standards receivers for these.

Lots of ways, but none directly easy unless you have the hardware.

David

[Refugee]

I once checked a counter by connecting it to a signal generator set to 198khz and listening for the beats on Radio 4.

[mole42uk]

I don't have a specification for overall frequency accuracy, but the user manual provides this:

Normal Frequency: 10MHz
Temperature coefficient: <±5 parts in 10⁸ from 5 to 55°C
Ageing rate: <±1 part in 10⁷/month
Short-term stability: <±1 part in 10⁹, RMS frequency error over a 1s period
Retrace error: <±2 parts in 10⁷ over 24 hours at constant temperature

I have seen 'typical' accuracy figures quoted of ±2·10-8 ±0.2Hz in a free-running OCXO

I do have an HP 5302A universal counter but I don't know if that will be any use.

I am seriously out of my depth here.

[Andrew2]

Originally Posted by Refugee

I once checked a counter by connecting it to a signal generator set to 198khz and listening for the beats on Radio 4.

Wouldn't the teleswitch phase mod cause confusion? Just listening to R4 with the BFO on reveals a rumble.

[cmjones01]

To check an OCXO with that sort of performance you need a really good frequency reference. Using an 'ordinary' frequency counter won't tell you whether the OCXO really meets its specification. The OCXO is likely to be more accurate than the crystal inside most frequency counters.

These days, such a reference is most likely to be a GPS-disciplined oscillator, like the popular one from Leo Bodnar: http://www.leobodnar.com/shop/index....roducts_id=234
Or, if you have a standards lab handy, a Rubidium or Caesium (!) based frequency reference.

None of these things are necessarily cheap, so probably best borrowed.

Chris

[Pamphonica]

You do need a much better standard than another OCXO to check DFM/counter accuracy.
After lockdown, perhaps try contacting Chriswood1900. He's local to you and definitely has various frequency standards (Rb, GPSDO etc). Sadly I am too far away to help easily.
-Jeremy

[Nymrod121]

Hi Richard - give me a tinkle on the blower when you have a mo ...
Best regards
Guy

[Vintage_RC]

Zero beating with 198KHz is never going to be better than tens of Hz due to the low frequency roll off of the AF amplifier and the speaker. The best you can do is to find the points either side of where the sound disappears and go for half way between the two points. I use a GPS disciplined oscillator.

[G0HZU_JMR]

One question to consider is "how important is it that it is in spec"? In other words, "does it really matter?"

If it is because you are going to be using it for radio repair/servicing then you could probably find someone nearby with an offair standard or GPS reference to help you calibrate it
However, over the course of a few months or so you probably won't benefit much from this because the ageing specs for your OCXO are not that great anyway. 1e-7 ageing over a month is 1Hz ageing a month at 10MHz.

This means that even if you paid for the best calibration service on the planet the reference in the counter could be 'below average' within a month in terms of error due to ageing. So you could have an error of 2Hz or more after a year.

If you want consistently better accuracy than this over time then I think you should consider buying or making some hardware (GPS std or offair std) that lets you check/calibrate it yourself.

[Radio Wrangler]

Audio cutoff is always a limitation with simple zero-beating, but you can sometimes do better by watching a S meter follow the beats at lower frequencies.

One source of 10MHz into an oscilloscope Y, the other into X and you can time the rotation of the Lissajous pattern. This can even handle other frequencies on simple ratios.

Whatever method of comparison, you still need a better standard than the one you are trying to calibrate.

David

[G0HZU_JMR]

I'd definitely recommend you try and find out what the annual ageing rate is as it might affect your decision. Often the annual ageing is only 1.5 - 2.0 times worse than the monthly ageing.

In your case this might mean the annual ageing might be <2Hz at 10MHz. A lot depends on what spec you really need and how often you will need to check/calibrate the OCXO to stay inside those limits.

I've got various 10MHz OCXOs in my RF test gear and I haven't checked them for many years. The one in my VNA hasn't been checked for over 5 years and some of my sig gens haven't been checked for 2 or 3 years. They probably are still in spec but I'm not that fussed really.

[G6Tanuki]

IMHO the thing here is to authenticate the calibration-reliability of what you're calibrating your OCXO against.

If the accuracy of thing-you-are-calibrating-against can't be traced back to a trustworthy source, then your calibration is meaningless.

and note that 'calibration' does not mean 'adjusting your thing to be precisely the same as the standard' - it means 'documenting the fact that your thing is 2Hz off-frequency'.

Errors are cumulative back to the master reference, of course.

[Radio Wrangler]

There is a calibration (and paper!) trail stretching back through various standards labs to a trio of hydrogen masers at NIST in Colorado. Once upon a time, Caesium was considered definitive, but now the worlds national standards labs compare standards and largely defer to the masers.

Norman Ramsey got a Nobel gong for inventing the coupled oscillatory mode hydrogen maser frequency standard. He also founded Fermilab.

David

[G0HZU_JMR]

IMHO the thing here is to authenticate the calibration-reliability of what you're calibrating your OCXO against.
If the accuracy of thing-you-are-calibrating-against can't be traced back to a trustworthy source, then your calibration is meaningless.

Agreed. I went through all this stuff nearly 30 years ago when I was still doing the odd radio repair for people. I would often get asked how accurate the frequency counter was. In those days I only had a Marconi TF2430 and a fairly scruffy Philips PM6669 counter.

I knew from experience (often checked it at work) that the Marconi counter was usually about 10Hz off frequency on startup and would drift to be within a couple of Hz after 30-45 minutes or so. I'd owned the counter for nearly 10 years already by then and it was always a consistent performer in this respect.

However, saying all this wasn't enough for some customers so I designed and built my own 198kHz offair standard so I could check the counter any time I pleased. The Marconi counter only had a standard crystal reference (not an OCXO) but in those days it was still good enough for a lot of amateur/CB stuff after a 30 minute warmup. The offair standard was overkill for the stuff I was doing back then but it did satisfy my customers.

These days I'd typically expect a decent 10MHz OCXO to stay within 0.2Hz of 10MHz over a year even if the ageing spec says 1Hz over a year. An example of this kind of OCXO would be the little 1E5 OCXO model used in mid range HP/Agilent gear.

[G0HZU_JMR]

I haven't watched this video in many years but W2AEW shows how to zero beat to WWV in the US using an old HF amateur radio. I'm not sure WWV can be received that well in the UK but the video does show the basic principle.

Note that he uses two counters in the video and the second counter is a decent Tek model. The second method he describes uses a counter and a sig gen and the s meter on the receiver.

https://www.youtube.com/watch?v=QCJ4cQGOQLI

[Radio Wrangler]

These ways of formulating specs miss the fact that crystal ageing isn't linear. Most often you get an asymptotic approach to a final value so the first day gives you so much ageing, the next ten days does roughly an equal amount, then the next hundred.

If you want a low ageing crystal, one of the things the manufacturer does is sit on it for a couple of months (well, it's usually in a temperature cycling oven).

David

[mole42uk]

Originally Posted by G0HZU_JMR

I'd definitely recommend you try and find out what the annual ageing rate is as it might affect your decision. Often the annual ageing is only 1.5 - 2.0 times worse than the monthly ageing.

This one specifies: <±1 part in 10⁶/year after 1 month's continuous use.

[Refugee]

The beat frequency method does work with a domestic radio without an S-meter.
You just need a little bit of bare lead between the signal generator and the counter under test.
The bare bit of signal cable can be moved about until there is just enough of the locally generated signal present to drive the AGC up and down modulating the audio as it goes.
I did it with a Marconi boat anchor sine wave generator that goes up to 200Khz and a germanium transistor internet radio.
The counter under test has an OCXO that is about a quarter the size of a shoe box.