It is usually considered that ELISA CV below 10% are normal. The exact numbers can vary depending on the particular assay and mainly of  the manipulation, of course. Handling is the most important factor influencing the outcome. I have used many different ELISA plates, Greiner, Nunc, Costar (either made of polystyrene or polyvinyl chloride) with consistent results. 

The percent CV between duplicate samples is acceptable if less than 20%. Most people prefer Nunc maxisorp plates for sandwich ELISA. The steps during the ELISA really matter in deciding  variation in the samples. It is worth to follow the datasheet strictly if commercially available kit is used or standardize the steps (majorly the concentration of standards and washing of plate ) if you are developing the ELISA indigenous. 

Ensure that you lay plates flat on a surface as temperature can be a cause of abnormal CV%. Don't stack plates if you can help it.