Combining several different affymetrix expression microarray experiments for further MDS analysis

Assume that I have three different Affymetrix microarray experiments, 2x U133 (from 2015 and 2017) and 1xU133 Plus 2.0 (2019). Each array contains several different samples, some of them are common for 2 or 3 arrays. I’d like to combine these thee results to perform further MDS analysis using whole sets.
I’m aware about some batch effect that I’d like to avoid.
When I do MDS using separately normalized data from three experiments, I can see that expression pattern of sample present in two of 3 matrices is the same but the normalized signal values are raised by a certain value. It visibly affects further MDS analysis.

My questions are:

1. Is there some bioconductor package or strategy to do such analysis avoiding bias and batch effect?
2. Can I just normalize all 3 experiments together using RMA? And then MDS analysis will be unbiased.

Thanks in advance!