I did some breadboard tests a while back. Mostly I used an AS3340A although some of it I also checked with an AS3340. And I wasn’t specifically following the 1222 schematic although I doubt if it matters for most of these.
That has nothing to do with CEM vs AS, it’s the current limiting resistor for the voltage reference. Value isn’t critical but should be above about 1k so that 1k to 4.7 k range is good.
In my breadboard tests 1M worked, larger values didn’t suppress ringing at the lowest frequencies.
I found a value of 4.7M, not 20M, is what I needed to suppress the frequency shift and @stmllr needed 6.6M. But @ChristianBloch apparently finds 20M good. I’m not too surprised at needing different values, since what you’re trying to do is cancel out one behavior with another and I’d expect it to vary depending on the chip and other variables, but then why did Christian find 20M worked for all the modules he made? It’s weird. If building a 3340 oscillator from scratch I’d say it’s better to use a -5 V regulator on pin 3, but that’s not practical for a 1222, so try 20M but you might need something different.
Yes. Polypropylene is good.