I’m very interested in time and frequency. One, for technical reasons and two, because when I was a kid it just grabbed me. Quartz crystals are at the heart of most electronic equipment that needs precise time and frequency. It is the “quartz” in a quartz watch. That is changing some with the advent of MEMS (Microelectromechanical systems) timing devices. Not much use today is made of mineral quartz with its impurities. Now a days the quartz used for timing and frequency generation is grown in autoclaves. A video by Bell Labs from 1962 explains how quartz is grown in labs and factories.
The military crystals were made to .001% tolerance. That is not too hard to get these days. On some of the boards I use for frequency measurement I have oscillators good to .0002% which can be easily adjusted with a voltage change to less than .00001%. That is to better than 100 parts per billion. That is not the best you can do these days but doing better costs a lot more. That is not exactly what I do (I adjust finer and expect to hold .00001% over temperature) but it illustrates what has been accomplished over the last 70 years.
I thought this WW2 video on quartz crystals might be of interest since the technology is fading. Not very fast for now. But it will accelerate as time goes on. I got the link off a list which prefers to remain nameless.
The FT243 crystals that became available as war surplus after WW2 were a great boon to amateur radio of the 50s and 60s. I used to have a bunch of them myself. The pinning was designed to match the hole pattern of octal tube sockets. Back then you had to use tubes to get up to the 7 or 8 MHz that the crystals could produce. That was back in the day when the best common transistors – the CK722 – could at best oscillate only up to about 1 million cycles per second. Now a days we say 1 MHz. I have microprocessors on my bench that run at 50 MHz and higher. We have come a fair distance since I was a kid.
Nice video on grinding FT243 crystals to get the “exact” frequency you want. But it really is not quite exact as it will depend on the circuit used. Varying circuit values allows crystals to be “pulled”.
Comments
3 responses to “Crystals Go To War”
Amazing video. I have a friend whose mother told me that she worked in the “crystal room” at Motorola in Schaumburg, IL in the 50s and 60s. I always wondered what that might have entailed…now I know.
Crystal filters need several very similar, but not identical crystals. I read an article in an old amateur publication which recommended buying some cheap surplus crystals at a convenient frequency and carefully grinding down some to raise the resonance a little bit. If you overdid it you could scribble on the face of the crystal with a soft pencil to lower the f back down.
Somewhere in storage I have an app note from Bliley on how to design crystal oscillators stable enough for broadcast use. The important bit is to have minimal loading on the crystal. So, the typical CMOS oscillator circuit is out of the question (the overdrive from the crystal makes the input protection conduct, lowering the input Z). A Franklin oscillator is a good choice. That’s a high-gain non-inverting amplifier with very small capacitors coupling into the crystal from the input and output.
MMM,
I’m very familiar with the Franklin Oscillator. I built one to test Ls and Cs.
Here is another idea. Tuning fork crystals have a parabolic curve. If To is the turnover temp then (T-To)^2 *.04 ppm will be the change in osc freq. Now if you could hold it to .1 deg C…. or better. Small thermoelectric coolers might work.
===============
http://www.qsl.net/va3iul/High_Frequency_VCO_Design_and_Schematics/High_Frequency_VCO_Design_and_Schematics.htm