How to know if a specific flux (e.g. F-SW-33) is no-clean or require cleaning.
Usually it will tell you in the datasheet, or the manufacture will mention it on the product page, if it doesn't have a product page or manufacturer information, then don't buy it. F-SW-33 flux ( or Fluxi FL 110) does not have a datasheet that I could locate, and does not have specs on how it preforms in a manufacturing setting (probably geared for prototyping). It is listed as a no-clean flux on farnells website.
It is also important to note that if you are doing prototyping, you may want to also use the same flux as your assembler if the design might be sensitive to flux degradation. (same thing also applies to solder) I've been bitten once by that, and this is just one variable to eliminate
How to know if a specific flux conduct electricity?
There isn't a resistivity spec given to most fluxes, if you want to know that you will have to test it yourself. There are two applications that flux residue could be a problem:
1) leakage current in high resistivity applications like ph meters or electrometers, where you are counting electrons.
2) Changing the capacitance of RF applications, a residue might affect the few pF's that need to be tightly controlled in the GHz range
In these situations it would be best to use a water soluble process or flux with solvent cleaning and avoid the issue of no-clean flux residues.
There are a few other problems with fluxes, they become conductive when placed in high humidity environments, with some having resistivities in the kΩ range (disturbing eh?)
The industry is aware of this and does preform testing, if your product is going to be in a high humidity environment, make sure it has been tested with the appropriate
Surface Insulation Resistance (SIR) testing is a methodology used to
characterize the PCB manufacturing and electronics assembly process
residues and their impact on reliability. It is usually performed on
industry standard test board coupons containing patterns, typically
interlocking comb test patterns designed for process testing purposes.
The patterns are exposed to a high humidity environment which
mobilizes any surface contaminates and reduces the insulation
resistance of the test pattern.
What is the impact of Halides?
Halides can wick/travel up the pins of an IC and into the wires and onto the die of an IC via electromigration. If the component is particularly sensitive to halides (like chlorine, or fluorine) then it can disable or destroy the functionality of that component. Halides typically affect components like optical components, that use different processes and chemistries than those of silicon IC's. The manufacture will generally set the halide limit, and mention it in the components datasheet.
Avago has a nice write up on this.
But, how to clean below SMD small resistors or IC? That seem to me not
an easy task. My conclusion is: Better use a no-clean flux, or at lest
a non-conductive flux.
I use a water soluable flux and clean my parts with DI water (this process is not acceptable for high MSL (moisture sensitivity level) parts. But I know I'm getting all of the flux off. The problem is with assembly, most manufacturers are moving to no-clean due to be environmentally friendly (no CFC solvents). Another thing to do is use techspray flux remover which gets most everything off your board. If your worried about residue for any reason, then don't use it. IF your design can take some abuse (all digital) then don't worry about it.
This answer suggest that for electronic I should only use Rosin flux.
Is that generally correct? or could an F-SW33 fill the task?
A rosin flux with a good flux remover solvent is another good way to make sure you have no residue, if you spray it down enough and use enough solvent.