# What is the maximum allowable error when aligning the magnetic heads of a floppy drive?

Today, I realigned the transplanted heads of a floppy drive by hand, and it felt like a herculean thing to do; as well as an experience I hope to never repeat.

Now that I'm finished, I want to quantify what I've actually done.

From what I understand, a 1.44MB floppy disk is 80 tracks at 135 tpi (tracks per inch). 1 inch is 25.4mm, so is the width of a track `25.4 / 135 = 0.18mm`?

How close do the heads need to be to properly be "aligned" with a track? And is there a degree of error or slop within the standard? How close to the center of the track does the head need to be for it to properly read it?

The specifications can be found in ECMA-125 standard.

The track location tolerance shall be within +/- 0.020 mm from the nominal position.

The distance between tracks is 0.1875 mm, but the width of a track is less, 0.115 mm +/- 0.008 mm.

• Whilst I appreciate a well sourced answer... I'm not sure what's in the spec necessarily conforms to what's allowable in the real world Dec 22, 2022 at 12:15
• @ScottishTapWater Would you expand on this?
– Zhro
Dec 22, 2022 at 14:08
• @Zhro - Well I've not got any experimental data to back this up, but in a hell of a lot of applications, things will still work quite well when something is out of spec. For instance, a huge number of USB devices don't meet the timing spec, but still just about work. There's often a physically acceptable tolerance for a lot of things that far exceeds the tolerance needed to meet the spec. I guess the difference is if you meet the spec, it will work but if you don't quite it might work. I'm sure it would be fine at +/- 0.025mm for instance even though that's out of spec Dec 22, 2022 at 15:52
• @ScottishTapWater It is entirely feasible for a floppy to be out of nominal alignment by a few mm, provided that the media is not expected to be interchangeable. What the ECMA specification is doing is specifying that if drives are aligned to that degree of accuracy then there will be a say 99.9% chance that a disc can be moved between drives. without an excessive number of read-retries. I'd also note that "back in the day" calibration discs were issued in hermetically-sealed envelopes, and (if one worked for a reputable company) had strict lifetime limitations. Dec 23, 2022 at 14:01
• @MarkMorganLloyd - Totally, I'm not disagreeing with that or that that's what the spec says. I'm just pointing out that a practical answer might be more useful than an answer from the spec in this case given it asks for the "maximum allowable", not nominal Dec 23, 2022 at 19:25

If you are reading floppies written with the same drive, it doesn't matter at all.

If you are reading floppies written with a different drive, then if those drives are differently aligned, the read head can start to pick up the flux from a neighbouring track. This materializes as "random noise" in the flux of the track you want to read, so as long as those signals are stronger, you'll still be able to read it. When the head is so far out of alignment that those signals get significantly distorted, you'll get read errors. It's even more fun if the head of the other drive wasn't properly aligned to start with.

So yes, there is a degree of slop, and actual results depend on a lot on which floppies you want to read, and where and how they were written.

I don't know how exactly you aligned your drive, but normally you do this with an oscilloscope and a floppy that is known to be written by a properly aligned drive, and then you can see directly the effects and the "slop" you get when moving the head, and how the signal distorts.

• This explains how it could be measured, but the OP's question just asks for actual specifications. Dec 21, 2022 at 14:17
• I well remember travelling to the uni with my final project work on a floppy disk only to find it cannot be read in the professors office. I had to go through the PCs in the students' lab to find one where the alignment is similar to my home PC to read the homework. Early 2000s. Dec 21, 2022 at 22:36

From what I understand, a 1.44MB floppy disk is 80 tracks at 135 tpi (tracks per inch). 1 inch is 25.4mm, so is the width of a track 25.4 / 135 = 0.18mm?

That's correct. To be compatible with other drives, you need to align to well below half the track width (if you're off by half a track, the head will read out a 50:50 mixture of two track signals, if it can read anything at all).

So, you should achieve something like 0.03mm accuracy or better.

• That's what it felt like. I could throw the heads off completely after aligning them, just by tightening the screws.
– Zhro
Dec 21, 2022 at 20:31