The use of quartz as a frequency controlling device has developed continuously over the last 50 years. The improved processing of the quartz together with techniques to shrink the quartz blank size has resulted in smaller and technically better crystal based devices. However, the basic method of using the inherent piezo-electric effect of quartz and combining it with an amplifier & feedback-loop to provide a quartz controlled oscillator has remained relatively unchanged.
The raw quartz used for crystal production must be of very good quality and high purity. Many years ago, the quartz used was mined and small amounts were of good enough quality to be used in the crystal industry. However, the need for high volumes, high quality and low cost drove the development of artificial quartz production.
Modern synthetic quartz is now grown artificially in an autoclave. This is a high pressure, high temperature vessel that holds “seed” quartz onto which the quartz grows. This method controls the purity of the quartz produced making it suitable for modern day applications.
*See Figure 1.
Having produced the synthetic quartz bar or stone, it is then cut into wafers. As the ‘angle’ at which the cuts are made affects the stability over temperature, it is critical to cut the quartz precisely. This is achieved using an X-ray machine to ensure the cuts are correct according to the crystallographic axes. As 90% of quartz crystals are based on what is termed the ‘AT-cut’, the quartz is cut at an angle of 35deg 15mins from the Z axis of the original bar.
*See Figure 2.
Specialist multi-blade saws are used to cut the raw quartz bars and can include laser refraction measurement equipment within the cutting saw, combined with a mounting and gluing system. This enables stones to be bonded together with the crystallographic angles aligned. The saw can then cut through the bars and give finished products with deviations of around 10 arch secs (a 360th of a degree!).
AT cut crystals have a temperature curve similar to y = x3 and deviations in the angle of cut used will affect the shape of the curve and frequency stability of the finished product.
*See Figure 3.
Modern electronic applications demand high stability crystals that operate over wide temperature ranges. As can be seen from the graphs of AT cut quartz angles, there is a basic physical law that can’t be over come and so the use of electronic temperature compensation circuitry can be used as in a TCXO device to compensate for frequency movement over temperature.
*See Figure 4.
As with any production process, the yield of the quartz produced must be controlled. To manufacture crystals that meet tight angle requirements requires a larger quantity of crystals to be produced which can then be separated into groups. The rest are used to meet other wider specification applications, hence crystal factories often have a library of crystal blanks to hand.
After producing the crystal blank, the frequency at which the crystal will resonate needs to be set. The frequency is inversely proportional to the mass of the quartz crystal. For AT-cut blanks the resonant frequency of the parts will be approximately 1680 divided by the thickness in mm. For example a 10MHz product will need to be ground, lapped, etched and polished until the thickness is around 0.168mm. At each process stage there is a risk of chipping, cracking, scratching or loss of parallelism. Any of the problems can cause the final product to function incorrectly, give spurious operation or perhaps sudden changes in frequency under particular circumstances.
With the demand for higher and higher frequencies, a point is reached whereby it is impractical to process the blank as it has become too thin. For older HC49 style blanks, the highest fundamental frequency that can be practically produced is around 40MHz. However, to achieve higher frequencies there are a number of techniques that can be used.
Firstly, the crystal can be operated at an ‘overtone’ of its fundamental frequency. As with all resonating systems there are harmonic resonances at odd multiples of the fundamental mode. By adding an appropriate filter to the oscillator circuit that can suppress the fundamental oscillation, a higher frequency mode of operation can be produced. Each successive mode means the gain is less, making the circuit design more sensitive and combined with the complexity of the filter this adds cost to both the design and component count. IQD’s HC49s are available up to 9th overtone and frequencies up to 270MHz.
A ‘multiplier’ circuit can also be used to create a high frequency circuit from a low frequency crystal. The problem is they can require more power, produce longer start-up times and are likely to have detrimental effects on the circuit noise.
Most standard crystal oscillator products employ optimized circuits using both the above techniques to achieve outputs of up to 800MHz. However, for high frequencies where noise is critical, we can use a quartz blank that is termed an inverted MESA. This design is also sometimes termed a High Frequency Fundamental blank (HFF). This technique uses a quartz blank that has its centre portion etched away to give the thinness needed for high frequency oscillation, while the outer ring is left thicker as a support for the crystal.
*See Figure 5.
Although the MESA crystal requires only a small area of the surface to be etched, this is a complex process and can only be done one at a time. The inverted MESA is an excellent solution to the problem of high frequency quality crystals as long as cost is not a critical factor.
For example, IQD’s IQXO-660 series uses a HFF crystal to generate the high frequencies required.
As the electronics industry constantly demands smaller & smaller components, this has driven the automation of the production process. Since the crystal blanks are now smaller this has meant that they also become thinner while still maintaining the appropriate dimensional proportions.
This means that higher frequency fundamental modes can now be achieved. For instance IQD’s IQXC-26 (housed in 1.6x1.2mm package) is available up to 80MHz in fundamental mode.
Having produced a crystal blank the next stage is to plate both sides of the crystal to create the electrodes that provide the external electrical connections and then mount the crystal inside an appropriate package.
Plating is done by mounting the crystal behind a mask, placing it in a vacuum chamber and vaporizing either silver or gold to coat the exposed surface. The choice of metal is determined by the cost and ageing requirements. This can be illustrated by IQD’s CFPX-181 which uses gold plating to achieve guaranteed ageing within +/-1ppm in the first year compared with the CFPX-218 which offers guaranteed ageing within +/-3ppm in the first year.
The plated crystal then needs to be mounted & bonded inside a holder. This process traditionally used a silver epoxy which provides mechanical anchorage & electrical conductivity. However with the introduction of ceramic SMD packages, this process has become automated and the epoxy changed to a silicon based adhesive. This gives an advantage as the bonding is slightly less rigid which increases the ability to withstand shock and vibration. It also reduces the possibility of causing mounting stress within the blank which could cause unwanted frequency changes and increased ageing.
*See Figure 6.
Having mounted the crystal in the holder it can be fine tuned to the final frequency. Until recently this was done by measuring the resonant frequency while depositing a spot of metal on the crystal electrode. The process was similar to the plating process using a vacuum and mask.
The demand for smaller components means the limit of this technique has been reached. The mask size has become so small for package sizes below 5x3.2mm the deposition can’t be controlled to a high enough standard.
The latest process makes use of an ion beam, this is fired through a mask and used to deplete the metal surface reducing the mass and raising the frequency of the crystal. This method has made it possible to achieve crystals with small sizes such as IQD’s IQXO-26 at 1.6x1.2mm.
With the crystal complete the unit needs to be sealed. For ceramic SMD packages there are two main technologies, seam and glass sealing. Both types give a hermetic seal which is necessary to create an inert internal atmosphere and avoid excessive ageing. Glass sealing is cheaper but it involves passing the component through a reflow process with a temperature above 350degC which can affect the internal structure of the device. Seam sealing is a more controlled process as it can be performed in a sealed chamber within a pure nitrogen atmosphere or for the new smaller components this equipment has been modified to leave the finished part containing a vacuum.
In the drive for smaller components, many obstacles have been overcome such as the use of ceramic for the base of the device. These base units are constructed from ceramic layers which are bonded together, the walls of the box being the most delicate part. In a 7x5mm package the wall thickness is 1.4mm but this method is not sufficient for use on a 2.5x2.0mm device.
To overcome this issue, manufacturers of the base units have had to develop new ceramic compounds which can be manufactured and handled at even thinner dimensions without breaking. The new components have flexural strengths 1.5 times more than earlier components (620MPa as compared to 400MPa) to allow the next generation of small size crystals and oscillators to be manufactured at 2x1.5mm and below. This same advanced method is used to achieve thinner packages suitable for use in for example PCMCIA cards.
We have come a long way in the 50 years that quartz crystal timing products have been in large scale production and at IQD we are constantly looking forward to the next innovation that will best meet the demands of modern electronic designs.
Orion Technologies Inc. is the manufacturer's representative for IQD in Canada.