13-Nov-2019

Oscilloquartz 8600-series OCXO warm-up

High-performance DOCXO (Double Ovenized Quartz Crystal Oscillator) tend to warm up very slowly. This is ok since once powered up they are typically left running for years, even decades.

Many models of Oscilloquartz oscillators exist. The high-end models include the 8600 / 8601 (superb) and 8607 (unobtainium). For each model there are sometimes additional suffix numbers which further define the factory measured performance and specification of the oscillator. When verified working a normal grade 8600 might be worth $1k - $2k and a premium grade 8607 might be an order of magnitude more. In other words, the rarity and price is directly related to performance: short-term stability ranges from 1×10-12 (good) or 5×10-13 (very good) all the way to 5×10-14 (unbelievably good).

The oscillator tested today

The model being tested here is an Oscilloquartz 8601.02.

Measurement system

Multiple hp instruments were used to watch the warm-up process carefully. A hp 53132A frequency counter was used to measure the output frequency. Two hp 34401A DMM were used to measure DC input voltage and current. Another 34401A was used to monitor the resistance of the internal thermistor as the oscillator warmed up.

None of this is required to operate an oscillator; you can just apply 24 VDC and enjoy a stable frequency a few hours later. The purpose of this experiment is to watch the warm-up process itself. Measurements were made continuously in RS232 talk-only mode and data was logged by a PC. The plots below show the first hour after power-on. The points are 10 seconds apart.

Power supply current

The double oven oscillator draws 450 mA for 25 minutes and then drops gradually to 180 mA.

Power supply voltage

A 24 VDC power supply was used but due to cable length and shunts there was some current related drop.

Thermistor resistance

I don't know what kind of thermistor is used. It's about 12 kΩ at room temperature and 1168.5 Ω at operating temperature. The glitches at 7 and 9 minutes are interesting. Next time I power-up this oscillator I will check if it's real.

Output frequency, full scale

The AT-cut crystal oscillator starts out 180 Hz above 5 MHz when "cold" (room temperature). It takes about 45 minutes to get within 1 Hz.

Output frequency, zoom

It's a very smooth and slow process.

Output frequency, fine zoom

Just before one hour a low point is reached and the frequency even more slowly climbs again.

Ambient temperature

For the record, the lab room temperature during the run was 70.8 ± 0.2 °F.

Comments

The first hour of warm-up is always interesting to watch because 99% of the changes occur during this time. The next 24 hours is also useful as the oscillator settles down to be a stable frequency source. From that point on it's a matter of monitoring the very gradual decline in frequency drift; a process that goes on for weeks and years.

Power connections (DE9)

The DE9 connector is conveniently designed. Think columns, not rows. The left two pins (1, 6) are for thermistor monitoring. The next two pins (2, 7) are ground and power. The right 3 pins (4, 5, 9) are for 10k multi-turn trimpot fine frequency adjustment.

In the simplest case connect pin 2 (top row) to ground, and pin 7 (bottom row) to +24 VDC. The thermistor pins and EFC trimpot pins are optional.

 

Here a BVA oscillator is being tested while still in its factory plastic bag.

 

This example has both pin 2 and pin 3 connected to ground.

 

Older style (pre-DE9) connections. Without and with EFC trimpot.

 

Lastly, the connector I use with convenient Anderson PowerPole for DC power and optional leads for thermistor resistance measurement.

Notes

dir:
  pc161-lab\bva\run2-sn270
files:
  bva\log1-mjd58800.txt
  bva\log2-mjd58800.txt
  bva\log3-mjd58800.txt
  bva\log4-mjd58800.txt
  bva\log5-mjd58800.txt
  sfe\log1-mjd58800.txt
first hour, 360 x 10 s samples, from MJD 58800.7478 to 58800.7895

Do you have comments/questions?