It is not so expencive
For example, MO drives use this principle for reading information

I don't understand the point of these "patterns", drive will never write these "patterns" on the platters in "00000000000" or "11111111111" or "01010101010101" forms, because all user information will encoded with RLL code
For example: "00000000000000" will be somthing like this - "110100110111010"

It becomes VERY expensive when you need something "drive independant" : MO drives use that technique but if you want to recover data from HDDs you have to deal with many manufacturers each one with something peculiar.

About the RLL encoding, the result is the same : when you say to the drive "record FF hex" the poles on the magnetic surface will change accordingly , OK, with all the encoding and subsequent encoding of the bit patterns... the kernel of the question is : the PREVIOUS information is still available ? I asked many professionals and the answer was definitely NO if you make multiple passes with precise patterns.

You can try this : get a common VHS video tape, record a TV programme on it, then rewind, record a BLACK screen on it with no audio.
Then rewind and play. With some VCRs, according to the strength of the erase current, you can see a "ghost" image as a result of the "incomplete" erasure / demagnetisation of the previous information recorded , and probably a barely audible residual of the previously recorded audio track.
Repeat recording black on the tape for two or three times and the ghost image will be gone.
In stereo VHS VCRs , as the audio is recorded together with the video and there are often flying erase heads the result is better.
Try the same experiment with a DV (digital video) recorder, and you'll absolutely get NO ghost image and NO audio at the first pass.
If you record white noise on audio and video (for video is the same if you record an empty TV channel) the result is the same of recording random bit patterns on a digital media.... try to see the ghost image or reveal the previous audio track

For Doomer : if you need ABSOLUTE certaincy, do you think for every byte of data is sufficient to record all the possible bit combinations, then the RLL encoding will make the rest for us? Time-consuming, ok, but at this point I think NO humanly available device will bring the data back.

Anyway, does anyone of you have proof of data retrieval after such erasing procedures and/or white papers / documentation (not strange experiments or rumors please, only serious research) ? At this point the discussion makes me curious.

Regards.

I reply to Doomer and make a little explanation for whom it may concern:

MO drives use laser beam for READING but the reading technique of a MO need a REFLECTIVE film beneath the magnetic surface, as I know.
The polarisation of the light reflected is influenced by the magnetisation state (Keer effect), and accordingly to the pole (south or north) of the domain we'll detect two polarisation angles and two states, respectively 1 and 0 for S and N. But this principle need a reflective film. You think that the aluminum substrate of a platter could act as reflector ? and what about glass platters used on IBM HDDs ? Interesting....

so it is just how it made in MO disks, it is not only one way. If you ever saw the HDD platter you have observed - this is a mirror.
Unfortunately you right about decoding process - this is very expensive to get data from magnetic image and company needs years for research

If you just rewrite data once (with any pattarn) on modern HDDs, this data will lost. Does not matter how many "passes" you will conjure. Data rewrited once (on modern HDDs) is definitely UNRECOVERABLE

BlackST:

*****
A simple cycle of writing specific patterns of 1 and 0 (for example, 1010101010 , then 01010101 , then 11111111, then 00000000) could destroy even the toughest "memory" of the magnetic material, remember you have to make UNRECOGNISABLE the information.
I know no one has got the data back but a few bits on such HDDs.
******

Could you provide us with some more details about this recover of "a few bits"? Where can we find some evidence on this? Again, don't forget that data recovery is not about reading raw sequences of bits from disk surface, but about getting meaningful data at the interface, and which result from a long signal processing path starting at the heads.

*****
You can more easily reconstruct information stored on heavily degaussed magnetic tapes because you only have to amplify the signal and process it, but on the digital world the things are different.
****

If you are talking about backup tapes, forget about it. You won't get data back with signal amplification (let alone the problem of servo-tracks destruction).

**********
One issue could be the G-LIST sectors : during use, it's normal for every HDD to develop bad sectors that are automatically locked out and replaced with spare ones. The "bad" sectors are added to a list but the information is still there, and one sector could contain 512 bytes of data, sufficient for storing your personal info. !!!
***********

Data in soft bad sectors usually may be recovered to some extent. Data lost by surface damage is gone forever. If a sector has just partial surface damage, I really don't know if the intact data can be recovered. I would appreciate if someone in this forum could shed some light on this subject.

*******
The P-LIST (primary list of defects discovered at factory level) could not be important because when you start using a new drive, those sectors are already locked out.
********

You have to consider that sometimes people move bad sectors from G-list to P-list to free some space in the first one.

However, again I have not the experience and so I do not understand where does the controller go find more spare sectors. Are there more spare sectors than entries in G-list? Again, I would appreciate some light on this.

********
So, a correct roadmap for securely destroy the data (and make it NOT ECONOMICALLY recoverable and also unaccessible to the majority of all poor mortals ) could be clearing the G-LIST, then writing sequential patterns of 1 and 0 as I said above , and then writing random pattern characters on each sector.
*********

"random" is the opposite of "pattern", so there are no random patterns

Why does writing first a pattern followed by random characters would make it more dificult to recover data?

But you pointed out a very important fact: data erasing tools routinely miss sectors discarded by the controller, as well as HPA and DCO (for a good discussion on this subject, look in Google for "Computer forensics and the ATA interface", by Arne Vidström).

*******
I tried not to be so "technical" , further deep explanations go beyond the purpose of this forum
******

On the contrary my friend, the first purpose of this forum is sharing technical knowledge and some threads around here are heavily technical. So, if you want to go deeper on this, you thoughts are welcome for sure.

About recovering overwitten data: again, I advise consulting the links I post above. I forgot to add the link to Nicholas Major "Data removal and erasure from hard disk drives", which you can find in ActionFront web site.

Doomer:

****************
the modern HDDs do not use binary writing on the platters
*****************

The first thing you learn in any digital electronics course is the difference between analog and digital signals. The second one is that there are no perfect digital signals and they are all analog.

The distinction between analog and digital signals is related with pratical matters. The voltage produced by a synchronous or asynchronous generator is not perfectly sinusoidal. However, you can model the voltage as a sinusoidal one for pratical purposes. A DC generator does not produce a perfect DC voltage -- it will be a rippled one. However, for pratical purposes, you can consider it constant.

Signals at head level must be processed to make them obey to the standards at interface level. However, even at the head level, there are two distinct levels of voltage. The rest is distortion and noise which must be corrected. But using MFM you can see clearly the flux changes. And you need to: MFM is also used to study and design better heads.

Anyway, the problem is not on seeing the flux changes with MFM, but in the all rest. So, I agree with Doomer, because he bases his opinion on cold scientific facts rather than on wishful thinking. Overwritten data is gone.


Daniel

I would like to say digital signals are binary signals (have two active states)
analog signals have more than two active states, this is a difference

[/quote]
I would like to say digital signals are binary signals (have two active states)
analog signals have more than two active states, this is a difference[/quote]

Wrong my friend. "Binary" and "digital" is not the same. And you don't define "analog" using the term "state".

An analog signal varies continuously, while a digital one varies between levels. A binary signal is a digital signal which varies between only two levels, named "0" and "1" by convention.

While signals in digital electronic circuits are analog by nature but are considered digital when obeying to certain characteristics which make them acceptable for their purpose, you can find many examples of real digital signals, as the one sent by the speed level command of certain vacuum cleaners or the one sent by the temperature level button of certain electric or microwave ovens (I mean, when you can not vary the speed or temperature continuously but only between predetermined values).

Daniel

Time to make an experiment : I have some scrap MO drives with good laser pickup and also a lot of bad HDDs (electronics failure but I hope the platters are intact).
I am trying to find some documentation in order to make a "mock" , using the spindle motor of the bad HDD and in some way use the MO laser pickup in order to get some signals out .... (i have some time to dedicate to the cause ) . I just wanna see if the platter could act as a suitable reflector (I think NO but I'll try anyway) .
I'll begin trying with "small" HDDs (e.g. 170 MB.) because actual HDDs I think have too small tracks and too high recording density for the optics of a common MO... the purposal is to get something out from the PIN photodiode inside the pickup or out the internal detector (some pickups have only a high speed photodiode and you have to start from here, some others have internal detector and amplifier, often integrated... I opened some laser pickups and there was an integrated detector/amplifier with transparent molding, so the reflected laser goes directly into this IC and you get the amplified signal right out.
It's just an experiment, I'll keep you informed.
(as you can see the argument is shifting from forensic to electronics to physics... who knows next ?)

BTW I was studying an alternative concept for servo track writing without using pushpin or external laser interferometers, controllers and expensive equipment. Does anyone of you have investigated such possibility ?

Regards.

So,
I can't explain it better, my English is very poor
HDDs use PRML encoding for writing information on the platters, if you could explain these signals better - do it, please

http://en.wikipedia.org/wiki/PRML


Partial Response, Maximum Likelihood (PRML):
http://www.storagereview.com/guide2000/ ... aPRML.html

Hope this helps...