Most of available formulas need prevalence or known odds ratio, which are not applicable for my current research. The study of population characteristics; prevalence of the studied phenomenon are unknown for us.
In complete absence of information about the phenomena in question, one approach is to carry out a pilot study sampling study using a pre-determined protocol to obtain estimates for the parameters needed to design the full sampling plan. Even in that case, you will still have relatively little information so perhaps it would be best to think of the full sampling plan based on an adaptive approach or an inverse sampling approach. These ideas may not be relevant for certain types of phenomena, of course.
This is to add to Neil's suggestion. As Neil said, you may try to do a pilot study to understand the characteristics of sampling universe to determine sample size. Alternately, you may try to do quick survey of selected study area covering only 2-3 key characteristics of entire sampling universe based on which you could determine sample size fro case control study. The sample size also depends on your outcome indicators. More important: in experimental studies, you need to randomize sample selection from the homogenous sampling universe for both case and control group sample.
Look up Sequential Sampling Plans. Despite the confusing name, SSPs are methods of calculating when to stop sampling. This is useful with crop pests where density and dispersion can change over time making pilot studies are of limited use. Of course, if you have no idea where your study organism lives, a preliminary presence/absence survey would be advisable and same time and money.
Nyrop JP & Binns MR. 1992. Algorithms for computing operating characteristic and average sample number functions for sequential sampling plans based on binomial count models and revised plans for European red mite (Acari: Tetranychidae) on apple. Journal of Economic Entomology 85, 1253-1273