frequency standard ideas ?

[Flump]

jpb you were right at 5v supply the output signal was too strong
5v in = 2.53v out , freq registers on dvm but not on counter
1.6v in = .83v out freq registers on counter as 1mhz exactly does not register on dvm
(v measurements were taken on normal dvm on dc)

can i use a vr to attenuate the signal ?

david the counter is an optoelectronics 2810 handi counter
sorry cant find any specs on it to link

[jpb]

jpb you were right at 5v supply the output signal was too strong
5v in = 2.53v out , freq registers on dvm but not on counter
1.6v in = .83v out freq registers on counter as 1mhz exactly does not register on dvm
(v measurements were taken on normal dvm on dc)

can i use a vr to attenuate the signal ?

david the counter is an optoelectronics 2810 handi counter
sorry cant find any specs on it to link

I'd just use a resistor divider to attenuate as 1MHz is comparatively low. It depends on the impedance of your counter, mine has a choice between 1M and 50ohms. But assuming that your counter is high impedance you could just use something like 100 ohms in series with 50 ohms and with the counter across the 50 ohms so that the voltage is around one third.

This will be much cheaper than say using a bnc attenuator.

Of course, if your counter is 50 ohms then just using 100 ohms in series might do the trick.

[David Hess]

jpb you were right at 5v supply the output signal was too strong
5v in = 2.53v out , freq registers on dvm but not on counter
1.6v in = .83v out freq registers on counter as 1mhz exactly does not register on dvm
(v measurements were taken on normal dvm on dc)

can i use a vr to attenuate the signal ?

david the counter is an optoelectronics 2810 handi counter
sorry cant find any specs on it to link

I'd just use a resistor divider to attenuate as 1MHz is comparatively low. It depends on the impedance of your counter, mine has a choice between 1M and 50ohms. But assuming that your counter is high impedance you could just use something like 100 ohms in series with 50 ohms and with the counter across the 50 ohms so that the voltage is around one third.

This will be much cheaper than say using a bnc attenuator.

Of course, if your counter is 50 ohms then just using 100 ohms in series might do the trick.

How fortuitous that I have its big brother, the Optoelectronics 3000A.  I like these counters but they are not substitutes for real test equipment.  If you want a nice, basic, and inexpensive bench counter, I recommend a HP5314 (100 MHz 1 MOhm), HP5315 (100 MHz 1 MOhm), or HP5316 (100 MHz 1 MOhm, optional 1 GHz 50 Ohms, and GPIB).  The HP5315 and HP5316 are also reciprocal counters which is nice for high resolution low frequency measurements.

I ran some tests on the 50 ohm inputs on my 3000A and these counters are picky about signal level at the lower end of their frequency range so I am not surprised that you had problems.  I think your 2810 lacks it but the high impedance mode on the 3000A works up to 50 MHz with fewer problems.

[Flump]

Thank you all for your input so far
this is all new to me and a steep learning curve
but i am taking in everything you have said
and i have learned quite a lot over the last
few days about crystal oscillators and their accuracy.

so the canned 1mhz oscillator i bought is +/-50ppm
which over 1mhz means it could be +/- 50Khz which is
quite a lot (only just learned what ppm is)

so I need something more accurate and searched for what some of you sugested earlier
an OCXO, 10Mhz was suggested so I found this
http://www.ebay.co.uk/itm/251355497721?_trksid=p2055120.m1438.l2649&ssPageName=STRK%3AMEBIDX%3AIT

and I wonder what you guys thought about it as i know nothing about them
what does Output Level :   HC-MOS Compatible   mean ? does that equate to a voltage range ?
Frequency Tolerance :   +/- 0.15 ppm max i think is +/- 1Hz is that right ?

maybe i should have put this in the beginner section as i know so little about this


Dave and jpb
my counter does have 50R and 1MR inputs, and a VR works to attenuate the signal on the 1MR input
strangely my counter reads 001.000.018 but I dont know how accurate that is because of the +/- tolerance of the oscillator

[ve7xen]

Output Level :   HC-MOS Compatible   mean ? does that equate to a voltage range ?
Frequency Tolerance :   +/- 0.15 ppm max i think is +/- 1Hz is that right ?

HCMOS is a standard logic family specification, the basis for 74HCxxx logic chips, and the specification means that it's compatible with this type of logic. It will have a square wave output with V(high) ~5V and V(low) ~0V when powered from 5V. Your counter probably (but by no means certainly) expects a sine wave input with 50R source impedance, which this is not. It might work anyway, read your counter manual if you plan to use it as a 10MHz reference and find out what the requirements are.

PPM works just like % does, but instead of per-cent (hundred) it's per-million. So to work out 0.15% of 1MHz you would divide the % by 100 to get 0.0015 then multiply by 1MHz to get 1.5KHz. Likewise, for 0.15ppm you divide by 1e6 to get 0.15e-6 then multiply by 1MHz to reach an error of 0.15Hz. Or to put it more simply, if the error is max 50 parts per million, and 1MHz is a million parts (Hz), then your error can be 50 parts or 50Hz (not 50KHz). Makes sense?

I am skeptical of the 0.15ppm tolerance claimed. The datasheet on the manufacturer's site doesn't mention absolute tolerance at all, and since this is a voltage controlled part with a pulling range of 4ppm +/- and aging of 0.5ppm in the first year, this tolerance seems to be very tight. Generally you would go to an OCXO for increased stability. Absolute accuracy requires a transfer standard, and a bare OCXO doesn't make a great one.

[David Hess]

Dave and jpb
my counter does have 50R and 1MR inputs, and a VR works to attenuate the signal on the 1MR input
strangely my counter reads 001.000.018 but I dont know how accurate that is because of the +/- tolerance of the oscillator

Your 2810 counter may not be any more accurate than the oscillator; if it is like mine, then it is rated at +/- 1pmm over temperature and 1 ppm per year.  There was a high stability TCXO option with double the accuracy but it was never clear to me how that was done since the frequency reference in mine looks like a normal XO.

If you have a shortwave receiver, then you can beat your reference oscillator against one of the WWB signals to get into the 0.1 ppm range:

[babysitter]

You can add a (diy) comb generator behind your reference source that gives you a lot of signals which relate as n times reference frequency. You want one.

Broadcast stations usually hold their frequency very well, too, not just WW* and DCF77, and of course the mobile phone base stations that donate the Rubidiums which are all over the place

To compare, a SDR dongle or so might help - there is a rtl calibration tool around.

[David Hess]

The local broadcast solutions I have checked have been an order of magnitude worse than WWV even without accounting for propagation issues.  Some analog TV stations used to phase lock their carriers to facilitate network transmission but that was a long time ago now.

I suggested WWV because it will be more than accurate for calibrating a simple crystal based reference (0.1 ppm is feasible) and it is about the easiest and least expensive source to use.

[jpb]

The local broadcast solutions I have checked have been an order of magnitude worse than WWV even without accounting for propagation issues.  Some analog TV stations used to phase lock their carriers to facilitate network transmission but that was a long time ago now.

I suggested WWV because it will be more than accurate for calibrating a simple crystal based reference (0.1 ppm is feasible) and it is about the easiest and least expensive source to use.

WWV works well in the US but I think it isn't available in the UK. (I'm not an expert but that is my understanding.) Most radio synchronized sources here are based on either the 60kHz time standard broadcasts or the 198kHz BBC Radio 4 long wave broadcast but I don't think there are any neat broadcasts at 1MHz or 10MHz.

[David Hess]

I missed that you are in the UK.  It should still work when propagation is good.

There occasionally are inexpensive GPSDO kits which show up.  Even the simple designs will be accurate enough for what you want to do.

[SeanB]

I can receive WWV or WWVH quite well here in South Africa at night, even with high levels of noise. Only way to tell them apart is to wait for the voice announcements.

[babysitter]

Have a look into the cheap SDR dongles again, they can calibrate against mobile phone base stations with a tool iirc called kalib-rtl.
Multiplied output frequencies have the advantage that the error gets multiplied too.

At least in germany the broadcasters take care of their carrier frequency, usually by disciplining to well-maintained rubidium.
153 kHz Deutschlandfunk Donebach, which I visited once, gives you another radio station to compare against in the LF area that has the level of "frecuency standard dissemination". But dont trust it 5 minutes around 20:00 pm - they disappear to switch the radiation pattern then ! :-)

[Flump]

Output Level :   HC-MOS Compatible   mean ? does that equate to a voltage range ?
Frequency Tolerance :   +/- 0.15 ppm max i think is +/- 1Hz is that right ?

HCMOS is a standard logic family specification, the basis for 74HCxxx logic chips, and the specification means that it's compatible with this type of logic. It will have a square wave output with V(high) ~5V and V(low) ~0V when powered from 5V. Your counter probably (but by no means certainly) expects a sine wave input with 50R source impedance, which this is not. It might work anyway, read your counter manual if you plan to use it as a 10MHz reference and find out what the requirements are.

PPM works just like % does, but instead of per-cent (hundred) it's per-million. So to work out 0.15% of 1MHz you would divide the % by 100 to get 0.0015 then multiply by 1MHz to get 1.5KHz. Likewise, for 0.15ppm you divide by 1e6 to get 0.15e-6 then multiply by 1MHz to reach an error of 0.15Hz. Or to put it more simply, if the error is max 50 parts per million, and 1MHz is a million parts (Hz), then your error can be 50 parts or 50Hz (not 50KHz). Makes sense?

I am skeptical of the 0.15ppm tolerance claimed. The datasheet on the manufacturer's site doesn't mention absolute tolerance at all, and since this is a voltage controlled part with a pulling range of 4ppm +/- and aging of 0.5ppm in the first year, this tolerance seems to be very tight. Generally you would go to an OCXO for increased stability. Absolute accuracy requires a transfer standard, and a bare OCXO doesn't make a great one.

I totally screwed up the math's on that one , I blame my calculator, the decimal point button is intermittent

I cant get wwv from my qth at any time of the day

I have been looking at the data sheet for an OCXO but I cant workout what the accuracy (unpulled just standard) is,
theres no ppm in the data.

Iv'e attched the pdf incase anyone fancies a look and can decipher it lol

[jpb]

Does your counter go down to 1Hz (or measure period up to 1 second)?

If so you can check your counter using a cheap GPS module with 1pps output.

I found this one which I think originates from Adafruit though I got it from proto-pic :

http://proto-pic.co.uk/ultimate-gps-breakout-66-channel-w-10-hz-updates-mtk3339-chipset/

It is a good module to have because it allows an external antenna and I've found that the jitter on the 1pps is less than 10nsecs which is pretty good.

[Flump]

hi jpb my counter has 2 inputs but neither of them go down that low,

Input A - 50R , 1Mhz-3Ghz
Input B - 1MR, 10Hz-50Mhz

that looks like a nice module but I think it would be best if i get a 10Mhz OCXO
so i can use it for other things as someone said earlier.

thanks for the idea though

[jpb]

hi jpb my counter has 2 inputs but neither of them go down that low,

Input A - 50R , 1Mhz-3Ghz
Input B - 1MR, 10Hz-50Mhz

that looks like a nice module but I think it would be best if i get a 10Mhz OCXO
so i can use it for other things as someone said earlier.

thanks for the idea though

I wasn't thinking of a replacement for the 10MHz OCXO but as a means of setting its frequency.

The big advantage of GPS is it is an absolute standard (jitter aside) so that there are no worries about drift.

There are modules that do higher than 1Hz but these tend to be snapped up by people building themselves GPSDO (at least the ones on e-bay) and ones like the
Ublox LEA-6T are expensive (just under £100 is the cheapest I can find).

My counter (TTi 930) does 1 Hz (this was one of the reasons I chose it as I knew I wanted to design a GPSDO using 1pps) and by using my OCXO as a reference to the counter I can fiddle with the OCXO frequency until it reads exactly 1Hz for the GPS 1pps input (actually it works a bit better using the period measurement rather than frequency).

Most counters seem to start at 10Hz or so - I found this when looking for mine where I was specifically looking to go down to 1Hz (of course the expensive Agilent ones go down to mHz but they are very expensive).

It might be worth looking out for an old Motorola GPS module on e-bay with 100Hz output BUT be warned, these older modules like 5V antennas whilst most modern antennas are 3.3V and if you then start buying an antenna as well prices can mount (modern 3V puck antennas are only about £5).

[G0HZU]

If you ask for a frequency standard recommendation on this forum you will get plenty of people telling you to go down the Rubidium or GPS route! Both options are a bit silly for hobby use and both options come with baggage because the GPS option means you ideally need to use an outdoor aerial to get it to work reliably, plus it can take 30 minutes to lock fully and the Rubidium standard is going to be expensive and power hungry and have a warmup time of many minutes as well. Plus you won't know if it is working reliably or how much useful life it has left as Rubidium oscillators have a finite lifetime.

The thing to be wary of is that this forum has plenty of members where a significant part of their hobby activity revolves around playing with time standards and dc voltage standards where they try to achieve silly levels of accuracy. I haven't seen any evidence that they actually 'need' this accuracy in anything they are working on at home. I can only guess that they just like to have something 'accurate' to satisfy some inner need.

The little MicroCrystal DIL14 OCXO is a fairly sensible choice as a fast warmup reference for a general purpose frequency counter for hobby use. I'd expect it will warmup in 30-60 seconds. You can expect it to age maybe a few Hz per year (away from 10MHz) if it is new. It will probably get better with age.

The second OCXO you linked to (the 8663.pdf file) is a very high spec OCXO. This would be typical of an OCXO offered as a $$ factory option to fit to a lab grade signal generator or a decent frequency counter from one of the top test equipment manufacturers. i.e. it is going to be way better than the reference oscillator they fit as standard.
It looks like it can achieve about +/-0.03ppm ageing over 1 year. So that would be about 0.3Hz ageing away from 10MHz (per year)  after you initially calibrate it with its trimmer.

I would expect that the 8663 OCXO would be quite expensive! It would also take quite a while to warm up and stabilise. Do you really need this level of performance? Maybe the faster warmup DIL14 OCXO would be the best compromise for hobby use even if it can only achieve 1ppm over a year.

You still have to adjust it to initially calibrate it but this can be done by various means.

[dannyf]

You can use a low frequency source (like the gps) to gate your oscillator to measure its frequency accuracy.

[jpb]

You still have to adjust it to initially calibrate it but this can be done by various means.

I don't disagree with your comments, though I think anyone interested in science and electronics will tend to have Voltnut and Timenut tendencies. I'm building a GPSDO for fun and a learning experience but I certainly don't need the accuracy it should provide. But then your average 100m runner doesn't need to run as fast as possible over 100m, but it is fun to try and see how well you can do.

My comments on this particular post were to try and answer your last point above rather than give super accuracy. If your equipment measures mHz (both my counter and the trigger frequency counter on my scope do) then the 1pps on a cheap GPS module does provide an accurate to at least 1ppb easy source that you can check in 30 seconds or so without the need of any calibration.
This is very different from using a full blown GPSDO or Rubidium reference.


The problem here is that the counter doesn't go below 10Hz (like many counters) so the approach won't work without more expense and complication.

Dannyf's suggestion is a good one though it adds the complication of an and gate and flip-flop/latch perhaps.

I think a cheap OCXO from e-bay is a good route but access to some means of measuring it is required as it may be out. But then if you're only requiring 1ppm accuracy it will probably be within that without adjustment.

[dannyf]

Dannyf's suggestion is a good one though it adds the complication of an and gate and flip-flop/latch perhaps.

It probably isn't that difficult to put together an instrument that does just that, with binary read-out.

I think a cheap OCXO from e-bay is a good route but access to some means of measuring it is required as it may be out.

I agree. Without it coming from a reputable manufacturer / distributor, you have no way of knowing how inaccurate it is.

A tcxo is a better way to go in this case.

[G0HZU]

In the original post the requirement was to calibrate the frequency function in a DMM and the timebase of a scope. Most DMMs will display 4 or 5 digits in this mode and I'd expect that the frequency reference inside the DMM will be pretty dire anyway. I'd be surprised if it was spec'd better than +/-100ppm.

So even the little DIL14 OCXO is going to be more than adequate.This Microcrystal DIL14 OCXO is quite 'old tech' in that we used to fit a custom version of this little OCXO in some of our receiver products back in the 1990s. We had our own custom XV5 spec and I think we paid extra to get +/-0.1ppm over a wide temperature range and I think the warmup time ended up being longer than the usual 30seconds.

I've got a few of them here salvaged from old dev boards and they do come in useful as a simple drop in upgrade to a std 10MHz DIL 14 oscillator or to any kit with a 10MHz reference oscillator. eg I've got one fitted to one of my Marconi 2019 signal generators after the std 10MHz oscillator failed on it. I've got another I use with an old Philips counter.

If a few cheapo 74HC390 dividers were fitted then it would provide tap off points for 10MHz, 5MHz, 2MHz, 1MHz etc etc down to below 1kHz. So there would be outputs that would fall within the frequency range of the DMM and also the timebase ranges of the scope.

[G0HZU]

I dug out a couple of Microcrystal DIL 14 10MHz OCXOs and tested them today. The versions I have here run from 12V and are to our XV5 custom spec and they warmup in about 20seconds to be within 1Hz. They take a couple of minutes to get within about 0.2Hz of the steady state condition. (takes 25mA from 12V in steady state)

The total adjustment range is +/- 50Hz so even if you just set the trimmer at the mid point you would get within about 1 or 2ppm. So assuming the ones on ebay are genuine Microcrystal OCXOs then they are reasonably good value at £19.

The only slight worry (if you wanted to get the best from it) is the sensitivity to the PSU voltage. I could get 1Hz drift for a 100mV change in supply voltage. The 10k trimmer resistor just acts as a pulldown resistor on one of the OCXO pins so this is isolated from the 12V supply.

[yramgu]

An OCXO is indeed the best choice for this particular application.

Then if you want porn-grade precision, and you're in the UK, you can try to catch a station like DCF77 (as previously proposed by Babysitter)^^
DCF77 used an atomic clock producing a frequency of precisely 9 192 631 770 Hz, which is used to lock a 10MHz reference. From that reference, they produce a 77.5kHz carrier wave which is broadcast.

You can use that 77.5kHz to lock your own 10MHz oscillator to produce a very stable reference. ON4JLV built one with a precision of 10^-3ppm

[babysitter]

Tell us how life is with a 10 MHz minus 10 Hz Oscillator when it has arrived ! O0

[G0HZU]

An OCXO is indeed the best choice for this particular application.

I agree but on this forum 'best choice' always gets trumped by 'coolest most accurate' choice especially if if involves using test gear with lots of digits