I got a pretty nice arbitrary function generator for free and it seems to not have any issues. It only has a standard crystal oscillator, not the optional crystal oven, but I'd still like to get an idea of the reference accuracy.
Years ago I remember an article in Radio Electronics, or maybe Popular Electronics, that would pick up the horizonal scan frequency of an NTSC TV and that could be used as a very accurate reference. Of course NTSC TV transmissions are gone (good riddance), but it got me thinking, could I use an AM radio station to calibrate the FG?
There's a Youtube video on this sort of thing, but it's a specific station in the UK and ignoring if I could even get it in the middle of the US I don't have a receiver that can pick this up.
What about a well established AM radio station? How accurate are they typically? This isn't important enough to me to buy a reference, but it would be cool if I could do this for free.
In the 1970ties, I built my first UK based, Droitwich 200kHz receiver as a very stable and precise reference for adjustment of my DIY - TTL based counter.
Nowadays, the transmitter still exists, based on a Rb, but emits on 198kHz.
The signal of our PAL color TV system was quite inconvenient to use, and I also experimented with the 19kHz stereo pilot tone of our FM radio, which was also quite stable.
Later I built a DCF77 receiver, from a Dutch / German electronics magazine, 'elektor', which converted the 77.5kHz to a precise 10MHz reference signal, provided one averaged over several minutes or hours. This device still works very well (after 35 years or so).
For the last 25 years, I worked near the Mainflingen DCF77 transmitter, and our favorite lunch restaurant is only a few hundred meters away from the antennas.
So even the fish on Fridays is AM-modulated.
Anyhow, I'd recommend to simply get a GPS time receiver, although all these 'historic' sources are quite exciting to explore.
Frank