In the particular circumstances of the problem there is perhaps something that you can do that might help.
This is because it seems that any English letters will do as long as they are legible. Is that correct?
It also depends upon how the font is intended to be used.
For example, if you are wanting to use the font on the Windows platform at 12 point or maybe 18 point or 24 point or 36 point or 48 point or 72 point and not at 10 point or 14 point then the following could help.
If each of the glyphs is a whole number multiple of 128 font units wide and both sides of the vertical lines are located horizontally on multiples of 128 font units wide, then better results might well be obtained at point sizes that are multiples of 12 point.
The Microsoft paltform uses 16 pixels for 12 point, 24 pixels for 18 point and so on.
This is a technique that I called mathematical pseudohinting, which does not use hinting at all, yet the glyphs are designed to make the display process easier for the rasteriser so that decisions come out well.
I searched in the forum for pseudohinting earlier this morning and found some links. viewtopic.php?f=3&t=3204 viewtopic.php?f=10&t=1279 viewtopic.php?f=3&t=1198 viewtopic.php?f=3&t=1120
I seem to remember that there was also another thread where I advised the same solution yet did not put the word pseudohinting in it and the person who had had the problem wrote back to say that it had worked and the font was now much improved. I cannot now find the thread in the archive.
I would mention that I almost always use 2048 as the value for Typo Ascender, Win Ascent and Ascender in my fonts. This has the effect that mathematical pseudohinting also works for horizontal lines. I do not know what would happen if an English alphabet from one of my fonts were copied into a font where Typo Ascender, Win Ascent and Ascender have a lower value.
I hope that this helps.
15 June 2011