I've used MRF successfully for years (+/-5% results for a 1.5M cells is quite okay) except with the last prototype I missed for like 50%. I'm now wrestling with transient to try to make MRF responsible