dmitryk

As you figure out one program is different from another.

You can't recover anything from one pass erase. So the program you been using didn't done one pass erase. Btw, it is take significant amount of time to do one pass erase.(e.g. 750GB drive will take about 4-5hr at least.) So anything which took less then 20-30 min on 40GB is not erasing data.

DoD erase was created based on somebody proposal who had uneducated knowledge of how HDD function. Many variation of multipass erase will refer to flipping bits, writing different patterns etc, to make sure that data is has bits has been flipped to a different direction.....
Well the technology behind writing a single bit (actually a pattern of bits) is design to address ever growing density of the drive. Magnetic domains are located so close to each other, that they interact and detection (recovery) of the bits takes significant mathematical computation. Data on the drive is scrambled with various polynomials and will never going to be same as what you were intended to write there. Read/Write channel using Partial Response Maximum Likelihood (PRML) technology decode and encode sequence of bits with special pattern from/to drive. Every manufacture use it's own polynomial and ways to encode data on the disk. That as well differ between different models as well.
When we attempting to influence what is written on the drive with different patterns we won't be able to achieve that because of the variation of how data is scrambled on disk.
When we try to write multiple passes we trying to influence bit flipping and band erasure. It is no way to predict where the head will fly next time relative to the band. In a sense you can't predict exactly if head will go through the same path or not. E.g. you know that head will follow Highway 30 but you can't tell which lane it will take.
It is not possible to use drive electronics to read bands since it has been design for normal strength signal. The only to way to get to it is use Magnetic Force Microscopy. To image one disk under MFM it will take months to few years. After image is obtained data has to be descrambled, and not normal data but band data which already lost its signal strength. Bunch other additional processes has to be followed to reconstruct where particular sectors were located in the filesystem. I didn’t mention another problem …. How you will identify when that data was erased… was just yesterday when you run erase program or 2 years ago when you delete that file….

Check links to white papers we published...
1. http://forums.actionfront.com/showthread.php?t=19
2. http://www.actionfront.com/whitepape...20Preprint.pdf

and Wikipedia article on PRML http://en.wikipedia.org/wiki/PRML