Frequency Counter Aging Rate

[digitaladdictions]

I bought a Fluke 1912A Multi-counter on eBay for $115.  The device is rather old (the manual I found was last revised in 1979). It has option 04 installed which is a 10MHz 0.5ppm TCXO. So new in box 30 years ago that would mean this would have been accurate to 0.00005%.

The temp dependance is +/-5x10^-7 (0.5ppm) The aging rate is +/-3x10^-7 / Mo. Based on the manual I will assume the device was manufactured in 1979. If I take 0.3ppm and multiply it by 34 that means I should add a potential error of 122.4 If I add the 0.5ppm for the temp dependency and the 0.5ppm inherent accuracy that means the device might only have an 123.4ppm or 0.01% accuracy which is worse than my Fluke 87V (granted it covers a much wider range of frequencies) providing my logic has not gone wrong anywhere. 

My first question is does aging rate reset to 0 when the device is calibrated or is that some physical characteristic of the oscillator? 

It has a sticker a calibration from tektronix on the back covering one of the screws and another tektronix sticker on the top giving a ID, S/N, Date, Due, and Tech.  The date is 08Sep2012 and its Due 08Sep2013.  Im assuming this means it was calibrated on September 8th 2012. This is quite surprising to me as I assumed this was probably sitting in storage for the last 20 years.

My second question; would hooking a rubidium frequency standard make it as accurate as the frequency standard or would the counter have other inherent limitations?   

Finally do you think this was a ok buy at $115? I received it today and all the functions seem to work although I did not get it up into the MHz range.

Thanks.   

[alm]

The temp dependance is +/-5x10^-7 (0.5ppm) The aging rate is +/-3x10^-7 / Mo. Based on the manual I will assume the device was manufactured in 1979. If I take 0.3ppm and multiply it by 34 that means I should add a potential error of 122.4 If I add the 0.5ppm for the temp dependency and the 0.5ppm inherent accuracy that means the device might only have an 123.4ppm or 0.01% accuracy which is worse than my Fluke 87V (granted it covers a much wider range of frequencies) providing my logic has not gone wrong anywhere.

Drift is usually not linear with time. If the drift is assumed to be purely random, expect it to scale with the square root of time (so 0.3 ppm * sqrt(age in months)). If it is systematic (like carbon composition resistors absorbing moisture and going up in value), then it will be linear with time. The reality usually lies somewhere in between.

My first question is does aging rate reset to 0 when the device is calibrated or is that some physical characteristic of the oscillator? 

If out of spec, the oscillator will usually be adjusted to be in spec.

My second question; would hooking a rubidium frequency standard make it as accurate as the frequency standard or would the counter have other inherent limitations?

A common accuracy spec is +/- 1 least significant digit + timebase uncertainty. If the external frequency standard is assumed to be perfect, the +/- 1 LSD spec should apply. Note that Rubidium standards are not primary standards: they require calibration and will drift over time, although at a lower rate.

[mazurov]

If you want to have a frequency standard, buy yourself a GPSDO. Trimble Thunderbolt is popular and not too expensive but gives you pretty accurate clock after a couple of days of running. You can run your instruments from this standard but this involves hassle like erecting the antenna and running the cable from the roof into the lab. I usually just tune the OCXO in my counter once a year and then run all other instruments off of it.

Don't trust calibration stickers on equipment that came from eBay. Get a crystal oscillator and check that your instrument is at least in the ballpark - old Ethernet cards were often equipped with 10ppm oscillators.

[digitaladdictions]

Thanks for the responses.

I already bought a FE-5650A Rubidium standard off eBay for $70 (shipped from CA (most of them where from china)) before writing my first post. Now you have me really wanting a GPSDO already and I do not even have the Rubidium one yet.

I will be buying my first house this summer and am going to make sure I get a place with no HOA or city ordinances against antennas so I can put up a tower for ham radio purposes. If I am going through that trouble one more run for a GPS antenna wont be a problem. There are benefits of being a single young professional with a little money to spare and no wife to tell me no.

The frequency counter came from a US seller who had two of the devices and did not even mention the calibration so I don't know why it would be faked. For just a hacker basement lab I think its reasonable to assume it was calibrated when it claims to have been and that I can assume it is within spec and I do not need to add the aging drift for 34 years.

I'm not doing anything life or death here I'm just a geek and think an extremely accurate time/frequency source is something cool to have. 

[mazurov]

With GPSDO you can have your own Stratum 1 NTP server which is also quite cool. Looking at the NTP client logs could be depressing though since they are rarely able to stay closer than within 1ms to the NTP server, even when on a same LAN.

[digitaladdictions]

With GPSDO you can have your own Stratum 1 NTP server which is also quite cool. Looking at the NTP client logs could be depressing though since they are rarely able to stay closer than within 1ms to the NTP server, even when on a same LAN.

My day job is a linux administrator and I was having that same thought.

[saturation]

See this post on characterizing crystal aging:

https://www.eevblog.com/forum/reviews/racal-dana-1996-frequency-counter/msg181049/#msg181049

[JackOfVA]

Crystal oscillators tend to age down in frequency due to outgassing. The outgassing contaminates the quartz plate surface and add mass, thereby reducing frequency. Outgassing increases with temperature so all else being equal, two identical crystals, one in an oven and one at room temperature, the oven crystal will age downward faster than one at  room temperature.

However, crystal oscillator manufacturers are aware of this problem and a high precision oven oscillator will have a crystal in a hermetically sealed enclosure, possibly vacuum or at least dry nitrogen. But, they still tend to age down in frequency as it's impossible to have the crystal holder perfectly free of contaminating particles and sealed.

On the other hand taking a run of the mill el cheapo microcontroller crystal and running it at 85 deg C in an oven will quickly drive it down in frequency.

A good oven oscillator will age slower with operating time, with a good percentage of long term drift occurring in the first few thousand hours of operation.