I use two types of flux.
Kester rosin rma type for all soldering but the critical stuff (bga, csp, qfn). This is semi thick and does not run all over the board when heat is applied like the the no clean types do. Its the darker brown colored stuff over the thinner lighter yellow colored stuff, it can be found just about anywhere, a $10 bottle lasts longer than the shelf life of the stuff. Most times I just add solder as this has the flux.
67/33 63/37 rosin core is my favorite solder as it has the lowest melting point, then 60/40 would be my next choice.
I also use chemworks tacky flux for the critical stuff. This has a tacky function. Apply flux, align part under microscope, hot air for a few seconds while pressing the part down lightly, this makes the part stick in place. This is excellent for bga, csp, qfn parts and critical for FIB parts. Fyi, fib is a part that the die has been reworked, most times, they use boiling acid to remove part of the plastic package to access the die, then rework one of the metal layers (connections) to fix a problem on prototype parts, the cost of these fib's are usually 1,000 to $3,000 dollars. So screwing up the soldering is not an option. The other option is making a new wafer, 3 months delay and 75~$150K
Tacky Flux is ~$15 a small tube, I don't recommend it for hobbyist (cost wise).
https://www.chemtronics.com/circuitwork ... tacky-flux
It also works great for high speed signals and does not needs to be cleaned off for 10GHz and below, regular rosin flux should be cleaned off for 50MHz and above (especially 2.4GHz antenna soldering). We use Rosenberg mini-smp 60GHz connectors, modified for 90GHz bandwidth.
Here is our new real time scope, check the price (yep, half million dollars)
https://www.keysight.com/en/pdx-x202060 ... =US&lc=eng
63 GHz of real-time bandwidth
80 GSa/s sample rate on 4 channels and 160 GSa/s sample rate on 2 channels
A single probe is $10K+. Rental of this unit is $10K a month.