C/G noise in direct sequencing results - (Oct/04/2011 )

My project consists of analyzing the methylation status of CpG's in the promoter region of the single copy genes on the Y chromosome.
When analyzing my direct sequencing results, I encounter some C background noise (when using a forward primer (G when using reverse primer)) throughout my DNA sequence. It is quite unpleasant since I am looking for the methylation status of the CpG's. This problem also persists when sequencing cloning products. Some of the genes however, didn't show this noise at all. Did anybody encounter this problem?

Thanks for any advice

Some extra information:
Primers amplify in silico just 1 DNA fragment. A test showed that 2 genes bind also on regular DNA, but 1 of these 2 does not have any C noise.
By doing the direct sequencing again, the C noise sometimes disappears. But on the other sequences that are run on the same plate, the C noise stays. I already tried to rule out contamination issues by using new primer aliquots, products, septa,....

I think it is your base caller compensating for the lack of C (or G in the reverse strand) in your direct sequencing result. because of the low signal, it's expecting something there and the background noise is amplified, When doing direct sequencing experiments you really need to calibrate to a standard, i.e.: known methylation of the amplicon you are interested in. Otherwise you are taking stabs on what the methylation value is in your sample.