Quite curious this one. If you are sent direct LAM from a long way out, you would probably do the following. Execute direct LAM, with the 'abeam points' active. You end up with straight line to LAM with logan, braso, sabre abeams. Enter the restrictions as appropriate and descend with the FMC. What happens is you end up about 600 to (max) 1000' high at sabre (if you take the abeam to be an equivalent DME range) This is due to the way the FMC computes the abeam point. So I end up re-constructing it with range arcs to make the restriction.
Q. Have you noticed and does it matter?