There are are several sources which are helpful in creating curves. I'm partial to Ron Reeder's book on digital negatives from Lulu.com. Ron offers the clearest and most reliable account of making adjustment curves and a most interesting website.
Tim Gray had two excellent articles in the Luminous Landscape, the second confirming Ron's curve-writing for the QTR program for the Epson. In it he described a program called Chartthrob which runs as a script in Photoshop.
I've since tried quite clever program which enables you 1) to generate a grayscale step tablet of which you make a palladium print and then 2) to "analyze" the result. This feature actually generates a compensatory curve which appears as a layer mask adjustment. Like the author, Tim Gray, however, I was unable to come up with consistent results.
I have since worked along the lines that Ron laid out. Not all of the curve creation features are active in QTR for the 4900, but Roy Harrington, the author of QTR, pointed out in his release notes that it is possible to manually create a curve in a textfile, something I was able to do utilizing one of the "new" colors (orange) of the 4900. I still need to tweak that one a bit.
The curve creation tool in QTR works fine if you don't use the "new" colors orange or green. I have two curves that are almost there. I've learned to "tweak" the Photoshop-produced *.acv file that you include as the gray curve in QTR.
I've gotten one QTR curve and embedded gray curve that is spot on until you get to the 80% patch of a positive palladium print of Ron's stepwedge. From 80-100% the actual positive print is a little lighter than it should be. For example, the patch that should be 85% is only 79% in terms of "K" (or darkness) units. 90% is only 81%, etc. I think I have corrected this at just this end of the spectrum. The challenge is to correct these values while not "uncorrecting" everything else. Will keep you posted.