I went into the rig and using my DDS sig generator and my RIGOL 'scope, measured actual performance. It looked worse than the prediction (part of the worsening is a difference in vertical scale):
LADBUILD lets you play around with the values of the components in the filters. I know that ripple is usually related to an impedance mismatch. So in LADBUILD I experimented (virtually) with different impedance values at the end. I noticed that at about 1000 ohms, the ripple and insertion loss got better:
So I went and built two broadband toroidal transformers. 4 turns primary with 12 turns secondary (1:9 Z). I'm assuming that the BITX has around 150 ohms at either end of the filter. That would put about 1350 ohms at the ends of the filter.
Here are the results:
Our book: "SolderSmoke -- Global Adventures in Wireless Electronics" http://soldersmoke.com/book.htm Our coffee mugs, T-Shirts, bumper stickers: http://www.cafepress.com/SolderSmoke Our Book Store: http://astore.amazon.com/contracross-20