Thanks for all the positive comments
I took the wav, sysex, and mix files MK sent me and did some tests. I ran the mix files through my IR mixer here and converted the mixes to sysex. The resulting files were bit for bit identical to the files MK sent me. I can't vouch for the sysex conversion because I can't test it, but it should fail every time if there was something wrong with it and it seems like a pretty straightforward process based on what AlbertA told me. So, I'm 90% sure that the conversion is not the problem. The details of the test are below, but my rather rambling conclusion is that the problem could be one or a combination of a few things:
1. The IR upload fails. I talked to Cliff about this one and he assures me that if the sysex checksum fails (meaning the data did not make it to the Axe-Fx exactly as it was supposed to) then the IR won't load at all. So, it is possible the upload is crapping out because of some MIDI transmission snafu, the Axe-Fx displays an error, but the user doesn't catch the message and assumes the IR is screwed up when it really hasn't been loaded at all. Maybe there is a subsequent change in volume when the cab block is on (not bypassed) even when no IR is loaded, so it sounds different than bypass even when no IR loaded. The whole Fletcher-Munson curve thing can play tricks on you when comparing the same sound at different volumes. Maybe someone with an Axe-Fx can check this one out. Is there a volume setting or EQ on the cab block that works even when an IR is not loaded?
2. There may be corrupted IRs out there. The "Lenny" patch seems like a prime candidate for this one since a few people have complained about it. I don't even know where to get it and I couldn't test it anyway, so I have no way of knowing if it's corrupted. Is the Lenny patch zipped up or just a naked sysex file out there. I would recommend that all patches and sysex files get placed in a zip container. The zip format has a CRC built in so it will tell you if the data is screwed up.
3. An Axe-Fx setting is set incorrectly. Maybe the mic sim is on. Maybe EQ applied. Maybe you uploaded to cabinet 2 (correct me if I'm wrong, I'm just quoting what I've heard). I can't help much with this one as I spent all the money on new cabinets and mics and don't have enough left over to buy an Axe-Fx
4. One mans thick is another man's thin. I would suggest that when people post patches and IR mix recipes that they include a short MP3 of what the patch is supposed to sound like. We all have different ways of hearing things and having a reference may help sort that out.
There may be some other issue still lurking out there, but #1 seems like the most likely candidate. It would be pretty easy for someone with an IR problem to test by recording the output of the Axe-Fx with the cab block bypassed and then with the IR "loaded" (at least attempted) and the cab block on, taking note of all the settings on the cab block.
Details of the IR mix testing:
---------------------------------
I ran this rhythm guitar clip through the resulting mixed IRs using SIR2 in my DAW:
http://www.redwirez.com/ir/NoCab-Guitar.mp3
As you can hear it's pretty fizzy without the IR applied. The clip is from a 10 watt London Power Studio 10 with 6V6 power tubes, so obviously it's not dialed in for these particular IRs, but it serves well enough to verify that the IRs themselves are not screwed up.
Scott's Bogner:
http://www.redwirez.com/ir/scott-bogner.mk-Guitar.mp3
SP Marshall1:
http://www.redwirez.com/ir/SPMarshall1.mk-Guitar.mp3
Vox1:
http://www.redwirez.com/ir/Vox1.mk-Guitar.mp3
FenderClean2:
http://www.redwirez.com/ir/FenderClean2.mk-Guitar.mp3
And remember these IRs are not particularly suited to this clip, so don't judge them on the basis of tone, just by the fact that they do what speaker IRs should do and filter out all the nasty fizz (among other things).
I did not use the 1024 sample truncated IRs, so these have quite a bit of room sound in them. Just for fun. Here's the same mixes using the 1024 sample truncated IRs:
Scott's Bogner:
http://www.redwirez.com/ir/scott-bogner.mk-Guitar_trunc.mp3
SP Marshall1:
http://www.redwirez.com/ir/SPMarshall1.mk-Guitar_trunc.mp3
Vox1:
http://www.redwirez.com/ir/Vox1.mk-Guitar_trunc.mp3
FenderClean2:
http://www.redwirez.com/ir/FenderClean2.mk-Guitar_trunc.mp3
You can download the 48KHz/24bit wav files of the test here:
http://www.redwirez.com/ir/MK-test-wav.zip
So, the IR mixes themselves are not the problem.