Ok well today’s task went to shit. I was going to start measuring all these crystals. Managed to get nowhere. Went for an amble, did some shopping and generally got sod all done.
Reminder: Discord this UTC evening
I also need to measure the motional parameters of 2-300 crystals, mostly form Golledge and Russia, but I've hit the following snag, the bloody Colpitts oscillator (the one with switchable series cap for determining ser/par resonance) it overloads the crystal (almost 9Vpp ) and creates a huge short term frequency instability, and I want to get them into 10-20Hz tolerance pools for my filters, combined with the fact that they're freaking ultra sensitive to temperature (if you touch them with your hand while inserting in the socket, they drift, I needed to use a plastic SMD tweezers) makes life miserable .
Does anyone have an idea of an oscillator that loads the crystal a bit less ?
Thanks,
DC1MC
Why dont you measure them in an IEC444 test fixture on a VNA ?
If it needs to be measured in an oscillator, the following caveats apply:
- never overrun maximum crystal power (thats a few ten microwatts, normally). Reasons: Nonlinear effects (pulling), heating,
in worst case electrode damage, permanent frequency change or death.
- Your oscillator needs limiting. Simple way: Schottky diode from AC base to ground.
Better: power regulator. Pick off Colpitts output, use AD8307 log detector and op amp regulator to drive base bias.
Advantage: power level can be accurately chosen and stabilized.
Regards, 73
Wolfgang DL1DWG
Well, I don't use a VNA and a IEC444 test fixture, mostly because I don't have them
, and also I have a bit of doubt that NanoVNA clone of a clone of a clone... that I can find around will have the necessary precision for sorting in a 10-20Hz range or better, I may be of course wrong, but this is it, I still don't have any of this.
I new about the the maximum power and overdriving issues, but I've thought that the people publishing these papers already designed these schematics with these considerations in mind, it seems that I was wrong as my mouth fell open when using 12V power supply for the whole test setup I've seen almost 9Vpp on the crystal terminals, I was expecting something like 0.5Vpp or such, no wonder that the drift was crazy.
Anyways reducing the supply voltage didn't change too much and around 5-6V it just stop oscillating with the current components, but just cut it abruptly instead of falling linearly.
I didn't want to put more capacity in parallel with the crystal and add more thermal instability with the diodes, but the variable biasing idea sounds cool, as the crystals are theoretically identical, even from the same manufacturing lot, it may be possible to get away with a trimmer for a specific lot instead of AD8307, or ?
Any acctual schematic, paper and such will be most appreciated.
73 de DC1MC