Sure, you can do this with a few genomic set operations with BEDOPS tools:

1. Convert your reads to sorted BED with bam2bed
2. Use bedmap --range to find elements within the first, smaller range boundary (set A)
3. Use bedmap --range again to find elements within the second, larger range boundary (set B)
4. Use bedops --not-element-of to filter out elements from set B that overlap elements in set A

What's left at the end are your elements between the ranges specified for sets A and B.

For instance, to get reads that are between 200 and 300 bases from one another:

$ bam2bed < reads.bam > reads.bed
$ bedmap --range 200 --count --echo-map --delim '\t' reads.bed | awk '$1>1' | cut -f2- | tr ';' '\n' | sort-bed - > reads_within_200_bases.bed
$ bedmap --range 300 --count --echo-map --delim '\t' reads.bed | awk '$1>1' | cut -f2- | tr ';' '\n' | sort-bed - > reads_within_300_bases.bed
$ bedops --not-element-of 1 reads_within_300_bases.bed reads_within_200_bases.bed > answer.bed

The file answer.bed has reads between 200 and 300 bases of one another.