look at this method:

https://www.researchgate.net/publication/235624054_projection/file/32bfe511f8d60eed1a.ppt

then if you find it suitable to your needs download the code a reference data set and the paper

https://www.researchgate.net/publication/235623992_Three-Dimensional_Representation_of_Geochemical_Data_from_a_Multidimensional_Compositional_Space?ev=prf_pub

Data projection

Article Three-Dimensional Representation of Geochemical Data from a ...

Weka can also do. http://www.cs.waikato.ac.nz/ml/weka/

caret R package is a good option.

if you have good command on MATLAB, you can use PCA in MATLAB

The ade4 R package does a good job at both PCA and PCoA and is quite fast.

http://cran.r-project.org/web/packages/ade4/index.html

RapidMiner is easy, quick, and free, but I wouldn't say that the graphical outputs are a plus.

However, the package RATTLE in R, I found to be the most direct way. If you have R installed, just install rattle - type in: install.packages("rattle") - and then call the package - type in library(rattle). Then, just type in rattle(), to get a very friendly user interface! Navigate through the tabs. I will be happy to help guiding you through the interface if you decide to give it a try.

KNIME also has PCA included. KNIME is easy, quick, free and powerful. The cool thing about KNIME is that with the node based interface, the boring parts of the data exploration are clicky-clicky while the more interesting parts can be tweaked to one's hearts content. There's even R and Weka integration when a text editor is the only way.

http://www.knime.org/

PCA with statistica seems cool! great graphics.

I finally ended up using the MFA functions of FactomineR. An excellent R package.

You should try out JMP. Seems easy to use.