Just for discussions sake, if I understand things correctly, you are putting 20 ug on column (1500/15 X 1mL injected at 5:1.) Slam. You should have a huge, ugly peak. At that concentration you could probably run a split of 100:1 since the loading then would be about 1 ug which is plenty for an FID detector although probably still above the capacity of a 0.32 column.

As an aside, water elutes under the methanol on a DB-624. You don't see it (very well, see other posts) until you try to do this with an HID and then it is pretty obvious. So, in essence, you phase is buried under water all the while the methanol is coming off. Therefore, Rodney's advice about a short piece of PEG at the front is what I would try next, after changing concentrations or split ratios (both of which are easier to do than a PEG pre-column.)

Best regards.