The build up of ice in the intake system works on exactly the same rules as on the airframe outside. Solid ice does not stick to aeroplanes. To get ice to adhere you have to have supercooled water present. Supercooled water can be present from just below zero to about minus 45deg, but is only present in effective quantities from zero down to about minus 20deg. Therefore you will get your most serious carb icing at intake throat temperatures of zero down to about minus 20deg. Your average carb, throttle closed, will reduce the air temp by about 25deg, and throttle open – less Bernoulli – by about 10deg.
Thus, with high humidity air at about 20deg throttle closed is your danger zone, the final intake temp coming down to minus 5deg, right in the serious icing zone. Likewise, in plus 5deg air throttle open is your danger area.
Carb heat should take the intake temp back up to above zero, out of the icing zone. Let’s guess that it raises intake temp by 20deg.
If the outside air temp is minus 20deg then you would not normally get carb icing. First, the temperature in the intake will be too low. Second, even if the air at minus 20deg is saturated it will still hold very little water vapour so the rate of build up of any ice will be small. If you now heat the air you will reduce its relative humidity, and no free water of any kind will form, not liquid, not ice and not supercooled water. You will get relatively warm dry air in the intake. The total water content is still the low value that it was for saturated air at minus 20deg. These conditions do not give icing.
You all know there are maximum ambient temperature operating limits. If the outside air is 20deg and you add 20deg of carb heat and open up to full throttle you will almost certainly exceed the temperature in the intake that the engine will take without detonation and loss of power. Believe me! I tried to open up a Pratt and Whitney with full carb heat still on and an OAT of 35deg and it hardly worked at all.
Best answer is if you have a cylinder head temp gauge. Then, if you suspect carb icing you select carb heat and wait for a rise in cylinder head temp and switch off the carb heat for 5 minutes and check again.
Dick W