What device is this about?

Last year I got a used HP 5316B universal counter. This frequency counter is part of the HP 531x counter series which had its first design with the portable HP 5315A at around 1978. The 5316A followed in 1981 and the 5316B in 1987.

The HP 5316B is an 8-digit reciprocal counter, measuring signals up to 100 MHz It measures Frequency, Period, Time interval, Time interval average, time interval hold off (delay) and Ratio.

For time interval measurements two independent input channels are used.

Below 10 MHz, the counter measures the input signal period, then computes and displays the reciprocal frequency with 7 digits of resolution in one second for signals as low as 0.1 Hz. Above 10 MHz, the counter automatically performs as a conventional counter.

To extend the capabilities of the counter there were three options available:

  • Option 001 TXCO – Temperature compensated Oscillator
  • Option 004 OCXO – Oven Oscillator
  • Option 003 Channel C – additional input channel to measure frequencies from 50 MHz to 1 GHz

Why this page?

The HP 5316B I bought was still equipped with the standard time base oscillator. As I wanted to make some accurate measurements the idea was born to equip the device with an OCXO Oven Oscillator, which is most stable (in terms of frequency constancy over time) type of a crystal oscillator.

I know there are many HP 5316A/B still in use, most of them are only equipped with the standard time base option.

Oscillator comparison

For the typical use of an crystal oscillator as a frequency counter time base this are the most important requirements an oscillator should have:

  • high accuracy and precision over time
  • high accuracy and precision within changing environment (temperature constancy)

Every crystal in a crystal oscillator is subject to mechanical aging. This means a frequency shift over time. The aging rate will be the highest for brand new crystals, but it will never stop completely. The aging rate is determined by several crystal parameters, i.e. the cut of the crystal, external mechanical stress applied to the crystal and contamination of the crystal during production. Even the mounting of the crystal in its case produces mechanical stress.

Crystal frequency is subject to change within changing temperature. Therefore, a standard oscillator will change its frequency with the warming of the device or just by opening the window of your lab.
One way to handle this is to add some other temperature sensitive parts to the oscillator circuit that will change the frequency to the opposite direction of the crystal change. This type of oscillator is called a temperature compensated oscillator (TCXO).
Another method to stabilize the crystal frequency is to put the crystal in a small oven and heat it up to a constant temperature. Typically this temperature is 75°C.

Some typical values listed in Wikipedia are:

oscillator typeaccuracyaging / 10 years
crystal oscillator10−5 to 10−410 × 10−6 to 20 × 10−6
10 MHz Crystal Oven (OCXO)2 × 10−82 × 10−8 to 2 × 10−7
Rubidium Oscillator10−910−12 bis 10−11
GPSDO (gps referenced oscillator)4 × 10−8 to 10−1110−13
Caesium atomic clock10−11 to 10−1210−12 to 10−11

 

Advantages and disadvantages of the time base options?

Every of the three time base options that were available for the HP 5316A/B has its advantages and disadvantages:

TypeCostStability over Temperature
(change per °C)
Accuracytime to stabilize after powering upaging rate (per month)
OCXOhigh2 x 10−8highslow
(>20 minutes)
< 5 x 10−8
< 3 x 10−7 per year after 180 days of continous operation
TCXOmedium1 x 10−6mediumfast< 1 x 10−7
standard crystal oscillator HP 05316-60008low5 x 10−6lowmedium< 3 x 10−7

What do these aging rates mean?

An aging rate of 1 x 10−7 / month mean that the frequency of the Oscillator may change from 10.000 000 MHz after calibration to a value between 9.999 990 and 10.000 010 MHz within one month. Depending on the frequency measured, this is up to the last two digits of the display, which you will not be able to trust any more after just one month.

However, this are worst-case aging values. In practice, the aging will decrease over time.

Temperature drift

Another important aspect is the change of the frequency over temperature. Even on the same day after adjustment, the frequency of the standard oscillator will only be reliable when there is no temperature change. The listed value of 5 x 10−6 for the HP standard oscillator

To get a real view of the temperature dependency i did a small measurement of the (non-OXCO) time base:

HP 05316-60008 time base frequency over temperature
23,8°C24,8°C25,8°C26,8°C27,8°C28,8°C
9,999 999 2 MHz9,999 998 0 MHz9,999 996 4 MHz9,999 995 2 MHz9,999 993 8 MHz9,999 992 5 MHz

The measured value of drift per degree Celsius is around 1,2 x 10−6. This is much better than spec, but not really a good base for reliable measurements.

Why to use an OCXO

The cheap price is the main reason why the default ordering option of the frequency counters is to use an unstabilized crystal oscillator even today.
So many of the used HP 5316 on the market today are still equipped with that cheap old standard oscillator and buyers on the second hand market are questioning if they should buy a device that has an unreliable oscillator in it.

On the other hand, the ovenized oscillator option has one negative aspect: it takes it’s time to heat up and get a stable (regulated) inside temperature. Especially the older original devices need up to 20 minutes to get a stable frequency. To accomplish an immediate correct measurement result after switching the power switch of the 5316 HP the oven heating is always on, even when the device is in standby mode. This also reduces the aging because cooling down and heating up the crystal produces additional mechanical stress resulting in a faster aging value. During my measurements I could see that sometimes an OCXO device had been heated up for many days changed its frequency up to 5 x 10−10 immediately after a cool down and reheat (a change that this devices normally had only after several days of being heated up due to normal aging). However, this aging produced by temperature change stress is susceptible for every crystal type oscillator, not even the ovenized ones.

Modern integrated OCXO devices are better in almost every parameter than the original HP OCXO from the 1980’s. Here is a comparison of the old spec sheet data and the spec sheet data of the OCXOs my design is built up:

OCXO TypeStability over Temperature
(change per °C)
time to stabilize after powering upaging rate
original "Option 004" HP OCXO2 x 10−8<20 minutes
(accuracy of better than 5 x 10−8)
< 5 x 10−8 / month
< 3 x 10−7 / year
Morion MV85 C20F
used parts
2 x 10−8<2 min
(accuracy of better than 1 x 10−7)
< 5 x 10−10 / day
< 5 x 10−8 / year
AXTAL AXIOM 75-275 x 10−8<5 min
(accuracy of better than 5 x 10−8)
< 2 x 10−9 / day
< 5 x 10−8 / year

With this values and a maximum aging of 5 x 10−8 / year (and especially the temperature independency that will is increased by an factor of 250 compared to the standard oscillator) the device will be reliable in “home use” measurements for maybe the next decade. The reduced heating up time will allow to power off the whole device without any doubt if needed.