return to main page

Magnetic Tape, Physical Magnetic Data Format

Started - April 10, 2006

Not guaranteed yet - I forgot a lot. We should at least reach a majority before declaring the following reasonably valid.

Allen responds

Grant responds

We have been waving hands and smiling about the ultimate goal of 'magnetic' tape, the pattern of the physical 'bits' on the tape which are written under control of a computer, and later read as data or indications by a computer.

Time to get into the meat of the matter - assuring that the magnetic changes are correctly written onto the tape.

Fortunately for we steady-state-magnetically-blind humans, magnetic "lines of force" and magnetic changes can be visualized using iron or other magnetic powder. Magnetic powder tends to line up along "magnetic lines of force" without changing the "magnetic lines of force" too much. :-))

The magnetic changes of state of a recorded digital tape can "developed" similar to photographic film by dusting the tape with magnetic powder (or in the real world - dipping a section of tape into a low viscosity fluid containing magnetic powder). The resulting pattern of magnetic changes may be seen by humans as dense collections of magnetic powder:-)) This visualization can help assure that everything is OK - and is interesting and re-assuring.

First lets make a symbolic picture of the magnetic situation on a properly recorded piece of tape

• IBM compatible - the "industry standard" in the early 1960s
• 7 track, 6 data tracks, one parity track
• BCD - means among other things that a value of zero is encoded as some bits, as in a 1401
• even character parity - BCD format is recorded with even character parity
  - tape written by other systems in "binary format" were usually odd character parity
• NonReturn to Zero - NRZ - when the tape was moving, the write heads magnetized when writing
  - a data one bit flipped the direction of magnetization
• at the end of the 'data record', one character position was bypassed,
  - and then the write head magnetization was returned to original orientation,
  - effectively giving an even track parity character.
• The tape continued to move almost a half inch, into the forming 3/4 inch "inter-record" gap
  - the tape comes to a stop and the current to the write heads turned off.

We are not getting into other features such as

- reflective tape marks for beginning and end of data

- volume labels (just another data record to the hardware)

- end of file marks, (IBM used at least two kinds)

- a special character, with the following track character parity of course

- 3 and 1/2 inches of 'blank' tape - recorded by write heads in the default direction

-

Got that? Maybe a picture will help. Looking at the magnetic oxide side of the tape:

The large arrows indicate direction of tape movement

Here is a three character BCD record consisting of "B", followed by a "6", then a "1".

The zeros are place holders, no flux change, the "1"s are magnetic flux change.

Click on image for larger version.

Now - how are we going to make/buy some "developing solution"? Assuming this magnetic tape visualization is a lost art, and the solutions are not even available on e-bay? ???

In the goode olde daze we could have captured some iron filings from a grind wheel and put them into some carbon tet cleaning solution. Now what solution can we use? ( I should have been dead of a miserable death long ago!! Too bad to die - only the good die young? ) Is there a source of finely divided iron? Would 50 mesh work? http://www.scientificsonline.com/Product.asp_Q_pn_E_3060176 , http://www.sciencekit.com/category.asp_Q_c_E_756151 ,

Here is a solution to the population problem - and also get a Darwinian Award - spray the results of a spinning grind wheel working on steel into an open pail of gasoline. :-)) Results guaranteed :-))

Allen Palmer responded to the above - politely, thank you

Ed. The question of 'developing' (as we use to call it) the pattern on mag tape is simple. We used the same iron powder as in the clutches suspended in IBM cleaning fluid. Any thing to dries quickly - alcohol will do. What you want is the finest iron powder you can find. Just move the tape up & down in the solution and then hold up to dry like a strip of photographic film. * NonReturn to Zero - NRZ - when the tape was moving, the write heads magnetized when writing - a data one bit flipped the direction of magnetization comment. remember that we have an 'erase head' that is positioned prior to the 'write head' . Ed I will give you a better explanation next Wednesday at the museum. Ed says - Ah yes, and the debates about AC vs DC erase. And the business of the write heads being wider than the read heads. OOHH the fog of time - Well, at least painful things hurt less :-)) while writing a record, current passes through all head tracks at all times , flipping when you wanted to write a bit. That is what NRZ means. It never comes to zero or no current status during writing a record. Allen j Palmer

The above will be incorporated into the body - Real Soon Now -

Grant Saviers responded to the above - politely ;-))

There was another gadget that was easier to use made by 3M. The magnetic particles were in a low viscosity fluid sandwiched between a thin piece of non magnetic stainless (like 0.001" thick or less) and a piece of glass. One would rub the SS "genie like" and this would suspend the particles. The image was developed by placing the SS against the tape. Both techniques worked well enough that individual bits could be read at 200bpi. At 556bpi, it is a struggle and forget it at 800bpi. b.t.w. when I operated a 1401 in 1966 we always used 556 as 800 wasn't adequately reliable with the tapes of the era in a production environment. The advantage of the solvent process was that an entire record and interblock gaps can be seen (even if the individual bits can't be read) and measured. The "write wide-read narrow" was/is a very desireable feature in tape systems as it liberalizes a number of tolerances in head making and dynamic tape tracking in the tape path. The separate erase head benefits have been often debated - they "clean up" the tape but aren't needed to overwrite previously recorded data (in digital systems). Many excellent tape drives were built without an erase head. While IBM made NRZI the industry standard for a while, it was clearly inferior to PE (phase encoded or Manchester) systems of the day. The first computer tape drive was on the Univac I and it was PE as were subsequent Univac drives, until they were forced to be IBM compatible. Things came full circle when IBM switched to PE with 1600 bpi and then other phase modulation systems in later drives. Even though IBM was on the wrong technology track for a while, they did a superior job of making reliable NRZI systems. I've come to appreciate this in learning how the 1401 data detection circuits work. Grant One other comment on the iron powder: I think the mesh will need to be more on the order of 500 to 1000 or particles on the order of low 10's of microns (1 micron = 40 u"). Grinding doesn't make particles this small, frequently they are made by volatizing the metal in a flame and condensing out the particles. Welding fumes do get this small which is why special respirators are used. Ball milling can also make them from friable materials, e.g. ferrites. Grant Ed comments - yes - the particles seemed to gather into fine fuzz. You could easily see the clumps of fuzz at 200 characters to the inch. They must have been "very fine".

Comments to Ed Thelen ed@ed-thelen.org or Allen Palmer or Grant Saviers or ...

---

return to main page