Thanks for your post. I prefer little to no centre channel so I'll try your suggestions when I get back from my holiday.
I've done some more work on getting good Quad upmixes and think I have cracked the workflow - but it is a bit more involved than my first few 5.1 upmixes. I have used the following workflow on two albums so far (2nd Law and Drones by Muse) and the results are excellent IMHO!! Essentially I am doing a two pass conversion - then taking the front channels from the first output and merging with the rears from the second.
1) Up-sample the source files to 176.4/192 using sox and merge together any tracks that segue into one another first
2) Start SpecWeb Play with the following command line switches - "SpecWeb.exe -P1 -M1 -c0 -f0 -a0 -l0 -r0 -o0 -40 -50 -C0 -z1 -L-0.1". This essentially presets most settings to zero, turns of ZAG, and sets to Artcan mode.
3) Set gain for Centre and LFE to -110 and tick the 'Norm only' box
4) Isolate the front channels and gradually raise the Front width until it is not distorted.
5) Switch to isolate the rear channels and check that you have some good isolation in the rears and can not not hear too much vocals or other 'centre oriented' content - This generally seems to be somewhere between 70 and 100.
6) Increase the centre width to somewhere between 20 and 30 to take out the remainder of the vocals.
7) Set Adjacent speaker to 0.04
8) Save an ini file in a subfolder called sLsR named as per the source file.
9) Set the centre width back to zero and save another ini file in a sub folder called LR
10) Repeat steps 2-9 for each source file
11) Put a copy of all source files in both the 'LR' and 'sLsR' subfolders and run the conversion process through the QQ Helper
12) Once complete, extract the Front channels from the LR output and the Rear channels from the sLsR output and put the two sets of newly created 2 channel files in separate folders. I have written a couple of batch files to do this in SoX.
13) Run both sets of files created in step 12 through DRMeter in Foobar to get peak and RMS levels. Also get DRMeter levels for the stereo source files.
14) The next process will adjust the levels of the output files to match the relative RMS levels of the source file, with the rear channels set to 3dB lower RMS than the fronts
15) I have written a spreadsheet into which you can input the peak and RMS levels calculated in step 13 and create a custom batch file to adjust the levels of the LR and sLsR sets and merge to form your final 4.0 Flac files. It is reasonably self explanatory - If the figure in cell I:50 goes red, manually put the same number in the adjacent cell and it should change to 0.00
16) Copy paste the content of cells A:55 to A:112 to notepad and save the file as gain.bat in a folder that has the output from step 12 in two subfolders named LR and sLsR
17) Double click gain.bat and you should get your final files a few minutes later, ready for final tagging etc
On the two conversions I have done using this process, I have found there to be much less distortion and artifacts in the rear channels - and of course, without a centre channel there can be no distortion/artifacts there either.
The attached zip file contains the batch files used to extract out the LR and sLsR files, the spreadsheet and an example gain.bat