Why does TRS-80 Model III/4 corrupt inserted disks on power-on/off?

[gonk23]

Any experienced TRS-80 Model III/4 owner knows that the disk shouldn't be in the drive (with drive door down) when powering on/off the computer, otherwise the disk can become corrupted.

Does anyone know technically what's actually happening for the disk to become corrupted? Is the drive head physically/permanently damaging the disk? Or is it just random signal sent to the drive head which can randomly modify disk data? Or something else?

Are there other computers of the same era that had a similar behaviour?

 

[SomeGuy]

I thought that was mainly with the Model I. Off hand I don't recall any Model IIIs doing that, but that sort of thing could vary.

It is not physical damage. Either the default uninitialized state of the controller electronics may inadvertently activate the drive in write mode, or the electronics in the drive itself might send a random write signal to the head.

 

[pjhacnau]

We had a III then 4 in the 80s and it was still the "done thing" to open the drive doors when powering on. Having read about it in the last few years I think that, it there was still a problem with the III/4, the risk was much lower than for the Model 1. May have been (justified) paranoia rather than a real risk.

PJH

 

[gonk23]

Oh it was definitely an issue with my Model III. I remember back in the 80s when we got our cassette-based Model III upgraded to a one-disk system. When we got it home, we fired it up with the disk in the drive and immediately corrupted the only TRSDOS system disk we had -- had to go back to the Tandy store for them to recopy the system disk and tell us the proper way to start it up.

Looking at the TRS-80 Model III Disk System Owner's Manual, it does actually mention it on page 4:
"(6) Power Switch. All drives shouid be empty when you turn the Computer on or off. Otherwise, the information on the diskettes could be destroyed."

I wonder why many other non-TRS-80 systems didn't have this problem, and whether there was ever a hardware hack to the TRS-80 to stop it doing this.

 

[Patrick.B (TTR)]

its can do it because even at the time TRS-80 recognized this very fact in the user manual (as most people no one reads it before turning on their computer and even less nowadays. but here..
it is just not a "modern device like we were used to "bios boot systems later in Ibm systems that booted neutral and then loaded bios from chip then it read those instructions then it proceeded with "is there a drive (first light on a drive)
TRS-80 are "crude in that fashion when you flip the power switch it energies everything in one power-up, thus causing the head to possibly spike a little and "zap a sector on a diskette often the boot sector.
the trick simple and use a lot by older guys in a video like me, we have a floppy diskette in the drive with the latch open, turn on the computer, when "diskettes? or not show up for a few moments we close the latch and the system boots.
though in theory this still can cause damage to the floppy disk in the last 30 years I never lost a floppy in this fashion.
as for inserting while the disk is spinning that is correct never do that because the SSDD floppy drives (standard 180k drive) uses the underside head to read/write and when you push a floppy while the drive is spinning your risking to run the centre hub against the magnetic surface (though unlikely it can happen ) and scratch disk rendering it bad.

as reference

in the TRS-80 Model III disk system owner's manual on page 5

then on page 6 in case you missed it, the provide a "startup procedure"

I am positive this was repeated in the model 4 User manual, so I looked...

so in the introduction to your disk system for the model 4 on page 7 we find:

and if you missed and happen to slip to page 8 under handling diskettes you find on page 8

cheers....

 

[rittwage]

I'm unsure why just opening the drive door would help you much. I guess the felt pad is not pressed onto the top of the disk (or in the case of double-sided drives, the top head is disengaged) but the bottom head is still sitting on the disk surface.

Also, the disk is not spinning, so the damage would be limited to possibly only one very small magnetic flux area.

Anyway, it's just "best practice" to not have disks in the drive. Sooner or later, you will regret it if you don't.

 

[lowen]

...
Does anyone know technically what's actually happening for the disk to become corrupted? Is the drive head physically/permanently damaging the disk? Or is it just random signal sent to the drive head which can randomly modify disk data? Or something else?

The Tandon and Texas Peripherals (a joint Tandy/Tandon venture) drives that Tandy used in the Models I, III, and 4 (until the 4D) lacked a proper head load mechanism. The head was loaded (that is, in contact with the media with the correct pressure) whenever the drive was closed. Single-sided drives could relatively safely shave this extra cost since if the drive latch was closed without a disk the head would simply contact the load pad where the other side's head would normally be. It wasn't recommended to close the drive latch without a disk as I recall, but it also wouldn't bang two heads together, either, like doing the same thing with a double-side drive could, even though the actual heads were offset by a few tracks between side 0 and side 1.

So, during power up, ANY glitch into the write/erase head circuitry of the drive can cause a magnetic pulse wherever the head is; if the head is touching media, it can cause a small glitch to write a flux transition. Tandy engineered their computers in an interesting way: they would first set the desired retail price and the desired margin, and then the device would be designed and engineered to meet that cost point. That's one reason the original TRS-80 Model I lacked lowercase characters; the cost of the extra bit in video RAM (a relatively expensive 2102 1Kx1 static RAM chip) could have (along with many other tradeoffs) put the cost over the threshold. It was less costly to recommend users "don't power up the computer with a disk in the drive" than to spend the money across every computer shipped to have hardware that ensured a relatively glitch-free startup.

Are there other computers of the same era that had a similar behaviour?

Well, I looked in some manuals to see if any other manufacturer ever made this recommendation. IBM, at least, in their IBM PC Guide to Operations, recommends the exact opposite:

 

[gonk23]

@Patrick Bureau - thanks for digging up those manual references -- proof for the disbelievers!

@lowen - thanks, that was what I was after. So presumably people that had upgraded to double-sided drives (or even installed 3.5" drives) wouldn't have this problem.

 

[ngtwolf]

Well, I looked in some manuals to see if any other manufacturer ever made this recommendation. IBM, at least, in their IBM PC Guide to Operations, recommends the exact opposite:

Unless a machine, like the Tandys, specifically call for removing the disks, it's generally better practice to leave a disk or one of those cardboard inserts in the drive to limit wear on the drive or issues in travel. This is more specific to 3.5" drives where the spring is continuously stretched while it's opened. I've had a few bad 3.5" drives due to the stretched springs from sitting for decades in that position.