By Brian Rickerby, former Technician in Charge of the NZPO Crystal Lab
Nearly all the crystals produce during my time in charge of the Crystal Lab, which was from when I replaced Brian Fanning, were AT and BT cut crystals. These were face shear vibration crystals.
There were not many BT cut crystals produced because their temp/frequency characteristics were not as good as an AT, but they did allow higher frequency fundamental crystals to be produced because they were thicker than an AT at a given frequency.
For lower frequencies other cuts of crystals were used but these were difficult to manufacture and had poor frequency stability. We started using small AT cut crystals with a CMOS oscillator and CMOS divider in a larger metal can (the size was just a bit bigger than the “F”).
Brian Fanning had done a lot of development work just before the changeover so I was able to introduce many of the things he had developed. The biggest change was probably introducing round crystals.
The crystal cutting machine was in the same room as the grinding machines, in front of the window. It had a diamond band saw. It was never used during my time. Brian Fanning said it didn’t work very well and was no longer required because we were buying blanks which were the required thickness already.
The grinders (we had two) were built in the development workshop (4th floor of the Wellington East Post Office), as was the crystal cutting machine. I think Len Svenson may have been involved with some it. They consisted of a circular doughnut shaped grinding wheel mounted on the end of the main shaft. Any blanks that needed to be thinned before lapping were held on to the cross slide by vacuum. A pin in the side of the holder was used to change the holes for different sizes of crystal. The cross slide was hydraulically moved back and forth, and stepped in 1/1000 inch automatically after each stroke past the grinding wheel.
The picture below shows the temperature/frequency/angle of cut characteristics of AT and BT cut crystals. We used the x-ray machine to check the angle of cut of blanks purchased from Japan, to make sure they were as ordered. Also, to select blanks for use in ovens and various temperatures to get the minimum frequency change for a required temperature range. For crystals in an oven at 70 degrees C, the cut needed to be an angle about +6 degrees from the reference angle. For best stability over a temperature range -10 degrees C to +50 degrees C, the cut needed to be at about the reference angle for minimum frequency change over the temperature range. From the curve below you can see that BT cut crystals were not too good over a wide temperature range. Both AT and BT cut crystals centred on about 25 degrees C.
We converted over to using round blanks for nearly all crystals. The blanks were 7mm and 12mm diameter. We purchased special lapping machines. One was similar to the one shown at right.
We had fewer holes to hold the round blanks. The disk was thin mylar film, clamped in the middle by a hard, thicker disc with a hole in the middle. Underneath was a steel doughnut-shaped lower lapping wheel that was driven from below. A similar wheel was held on top, concentric with the lower wheel. The disc holding the crystals was offset so that when the machine was running the crystals went from side to side, as well as around. This was done so that the lapping wheels remained flat and did not get circular or other types of trenches. Higher frequency crystals (particularly overtone) have to be extremely flat and parallel in order to get good activity. The grinding solution was continuously applied. Various types of grinding fluids were used (silicon carbide, aluminium oxide, etc).
Lower frequency crystals (below about 1.5 MHz) had to be curved on one side to prevent other modes of oscillation. Previously the square blanks were bevelled at the edges. The machine for doing the convexing consisted of a dished type wheel with the calculated radius, driven from underneath. A pen-shaped device was mounted on the base of the machine in such a way that it could be adjust to be above the curved lapping wheel. Three round crystals were held on to three steel discs with a ball on top. A triangular piece with a ball on top and 3 dish-type holes (not going right through) fitted on to the pen-type device. The pieces holding the crystals fitted under the triangular piece. The angle of the pen device was set to be perpendicular to the surface of the curved driven disc at the centre of where the crystals were. When the machine was started, the curved driven disc rotated and because the crystals covered a band of diameters it made the crystals and the triangular piece spin.
The round lapped crystals were mounted in the spring clips of the holder, and glued in with silver loaded epoxy resin which was a good conductor.
We had various plug-in oscillators attached to the plating machines: some ordinary fundamental oscillators with various input capacities plus a number of overtone oscillators for 3rd and 5th overtone with different types of loading. The frequency could be measured while the process was on.
The machines had a mechanical rotary pump (like a Wankel engine) to get the pressure down very low. Then an oil diffusion pump to get it much lower.
The oil diffusion pump operated with an oil of low vapor pressure. The high speed jet was generated by boiling the fluid and directing the vapor through a jet assembly. The oil was gaseous when entering the nozzles. Within the nozzles, the flow changed from laminar to supersonic and molecular. Often, several jets were used in series to enhance the pumping action. The outside of the diffusion pump was cooled using either air flow or a water line. As the vapor jet hit the outer cooled shell of the diffusion pump, the working fluid condensed and was recovered and directed back to the boiler. The pumped gases continued flowing to the base of the pump at increased pressure, flowing out through the diffusion pump outlet, where they were compressed to ambient pressure by the secondary mechanical forepump and exhausted.
Unlike turbomolecular pumps and cryopumps, diffusion pumps have no moving parts and as a result are quite durable and reliable. They can function over pressure ranges of 10−10 to 10−2 mbar. They are driven only by convection and thus have very low energy efficiency.
One major disadvantage of diffusion pumps is the tendency to backstream oil into the vacuum chamber. This oil can contaminate surfaces inside the chamber, or upon contact with hot filaments or electrical discharges may result in carbonaceous or siliceous deposits. Due to backstreaming, oil diffusion pumps are not suitable for use with highly sensitive analytical equipment or other applications which require an extremely clean vacuum environment, but mercury diffusion pumps may be in the case of ultra high vacuum chambers used for metal deposition. Often cold traps and baffles are used to minimise back streaming, although this results in some loss of pumping ability. The oil of a diffusion pump cannot be exposed to the atmosphere when hot. If this occurs, the oil will burn and will have to be replaced.
When finished the crystals were canned then engraved. This was necessary to meet certain specifications, as labels could come off fairly easily. We normally never used the motor to rotate the cutter, but used a diamond point inset to do the engraving. The crystal type and frequency were normally engraved on just one side.
Brian Rickerby was the second-last Technician in Charge of the NZPO Crystal Lab.