Good that the issue was solved .
As said, the dd for windows by John Newbigin is a very good tool but it is not really-really dd, since it has been re-written it may behave strangely with strange, rare setups.
In theory dd is (should be) "sector size agnostic" but there may be (actually there are) some limitations due to the NT technology/OS.
The DSFOK package contains (among others) two different tools, dsfo and dsfi, that basically mean dsf-out and dsf-in, one gets data out of a device/file, the other gets data into device/file.
To give you an example of a possible issue that may be connected to what you experienced, dsfi cannot write on device less than a whole sector, and I believe this is due to the way the device is accessed under NT, if you need to change a single byte on device you dsfo the related sector to file, dsfi the single byte to the copied sector, then dsfi back the whole copied sector to the device, overwriting the whole original sector.
What is needed to "pseudo-wipe" a device in NT based systems largely depends, usually it is enough even to delete the "magic bytes" (55 AA), i.e. the last two bytes of the first sector of a hard disk like (partitioned) device to have it shown in Disk Manager as "need to initialize",
But other systems/OS may have completely different ideas on what to look for/what to check for.
To give you an example (only seemingly unrelated) check what kind of checks different OS/drivers do on floppies:
But - as another example - I have a couple machines that have a BIOS (Insyde) that checks the actual code in the MBR and decides whether to boot or not based on that (i.e. the DOS/2K/XP MBR code boots just fine, the grub4dos doesn't (actually later the grub4dos MBR code was changed so that it allowed booting), the syslinux does in some versions but doesn't in others), I wouldn't be surprised if the good Apple guys invented something similar for the Ipod.
About the count parameter what is the problem?