Old McDonald's Farm by Edit

contribute new/updated software missing from online archives
Post Reply
User avatar
scarybeasts
Posts: 797
Joined: Tue Feb 06, 2018 7:44 am
Contact:

Old McDonald's Farm by Edit

Post by scarybeasts »

Hi,

This disc has a bit of a story. I spotted this possibly MIA title as part of a batch of discs (see bottom left):

omf_in_batch.jpg

billcarr2005 picked it up, but it doesn't load and appears to have wrecked catalog sectors -- even a Greaseweazle sees only noise. So I started trying to recover the bits with my new scope.

signal_collapse_catalog_sectors.png

As can be seen, the signal collapses to almost nothing in a couple of places at the start (and mirrored at the end). These are the catalog sectors -- oh dear! This is the weakest signal I've seen but by some miracle, one of my drives is able to get something almost legible. So I proceeded to recover the bits best I could. billcarr2005 noted that other EDIT titles seem to use the same engine and files so with this boost via cross-referencing, and billcarr2005 checking my proposed SSD, we have a recovered image. It's attached. We think it's ok but it's hard to be sure because the title itself doesn't seem to do all that much.

omf_screenshot.png

You'll need to run with the disc image writable, otherwise it'll fail to load with a write protect error. The setup is a bit haphazard: the boot sequence writes the current value of PAGE into a file called "page". In fact, it's this write to the original disc that failed in some way at the physical layer and wrecked the disc. The physical layer failure mode is the same as we encountered when recovering the source code to "The Living Daylights". We don't know exactly what goes wrong and would love any experts to weigh in about how disc writes might fail in this way. (We considered alignment as a likely failure mode but I've manually misaligned one of my drives heavily in both directions, and the signal gets no stronger.)


Cheers
Chris
Attachments
omf.ssd.zip
(25.88 KiB) Downloaded 18 times
User avatar
Diminished
Posts: 641
Joined: Fri Dec 08, 2017 9:47 pm
Contact:

Re: Old McDonald's Farm by Edit

Post by Diminished »

That is fascinating.

This is such a stupid question I'm almost embarrassed to ask it, but I'm guessing there's no visible evidence of any physical damage to that region of the disc?

I'm just wondering about bizarre realities in which the write head's current amplifier somehow went loco, and started putting an order of magnitude too much current through the write head, somehow knackering the oxide in one way or another.

Ugh. I suppose it might even be possible to test that hypothesis. But it might cost a drive.

Another thing I'd love to try would be to see if you can write successfully to that region using a known sane drive, but that's likely out of the question for obvious reasons. Maybe if a less precious disc turns up with the same failure mode? Perhaps it would be an idea to start to collect as many unreadable discs as possible?
User avatar
scarybeasts
Posts: 797
Joined: Tue Feb 06, 2018 7:44 am
Contact:

Re: Old McDonald's Farm by Edit

Post by scarybeasts »

Diminished wrote:
Tue May 11, 2021 4:18 pm
That is fascinating.
Isn't it? :)
This is such a stupid question I'm almost embarrassed to ask it, but I'm guessing there's no visible evidence of any physical damage to that region of the disc?
There's a slight bit of general wear around the region of track 0, but nothing perfectly localized to two sector bodies. The sector headers, including the one in-between the two wrecked sectors, are 100% fine.
I'm just wondering about bizarre realities in which the write head's current amplifier somehow went loco, and started putting an order of magnitude too much current through the write head, somehow knackering the oxide in one way or another.
Is that possible? A failed resistor or something?
I was imagining possibly the opposite scenario: a feeble amount of current making it through.


I should add an addendum to my original post: one of the more amusing parts of this story is how it wasn't particularly necessary to dive in with an oscilloscope. Since it's the catalog sectors that are broken, they can be rebuilt by hand by comparing catalog and sector content with a similar EDIT title.


Cheers
Chris
User avatar
Diminished
Posts: 641
Joined: Fri Dec 08, 2017 9:47 pm
Contact:

Re: Old McDonald's Farm by Edit

Post by Diminished »

scarybeasts wrote:
Tue May 11, 2021 8:35 pm
Is that possible? A failed resistor or something?
I was imagining possibly the opposite scenario: a feeble amount of current making it through.
You asked for experts, and I'm not one. I should change my name to Dunning-Kruger. [-X

Here's another theory: Maybe the write current is fine, but the output stage is no longer receiving a meaningful signal due to something having blown up earlier in the signal chain. So what gets written to the disc is quiescent (a.k.a. a big load of nothing), or high-frequency noise.

That theory could at least be tested without destroying anything -- you'd need to temporarily disconnect (or maybe ground) the input to the amplifier that drives the write head. I actually have a spare drive here that I picked up for free from an ABUG meet, but it seems to have an integrated controller which stuffs all the gory details onto a single IC, so I don't think I can try it with this. Maybe if you could find an old one that has less integration.

Hopefully someone who actually knows what they're talking about will be along shortly
User avatar
BigEd
Posts: 4099
Joined: Sun Jan 24, 2010 10:24 am
Location: West Country
Contact:

Re: Old McDonald's Farm by Edit

Post by BigEd »

So, it sounds like the great majority of the disk is original, written by duplication equipment, and then just a few sectors are written the first time a user runs the program - written by their own drive. If that's so, it's enough to suppose that the first user's drive was broken in some way. In particular, perhaps, it had too little current in the write head, or some crud between the write head and the disk - would that do it?
User avatar
scarybeasts
Posts: 797
Joined: Tue Feb 06, 2018 7:44 am
Contact:

Re: Old McDonald's Farm by Edit

Post by scarybeasts »

BigEd wrote:
Tue May 11, 2021 10:14 pm
So, it sounds like the great majority of the disk is original, written by duplication equipment, and then just a few sectors are written the first time a user runs the program - written by their own drive.
I think it's every time the user runs the program. !BOOT chains "DEFS", which does this on line 20:

T=OPENUP"page":Y=PAGE:PRINT#T,Y [...]

So IMHO it's a hazard waiting to go off.
If that's so, it's enough to suppose that the first user's drive was broken in some way. In particular, perhaps, it had too little current in the write head, or some crud between the write head and the disk - would that do it?
I'm not sure about a dirty head, but maybe it's possible. For the write to proceed, a successful read of each sector header has to occur. Also, the eventually written data is weak and noisy -- but pretty uniform. On the read side of things, a dirty head usually has a signal that's all over the place.


Cheers
Chris
User avatar
scarybeasts
Posts: 797
Joined: Tue Feb 06, 2018 7:44 am
Contact:

Re: Old McDonald's Farm by Edit

Post by scarybeasts »

Diminished wrote:
Tue May 11, 2021 9:47 pm
scarybeasts wrote:
Tue May 11, 2021 8:35 pm
Is that possible? A failed resistor or something?
I was imagining possibly the opposite scenario: a feeble amount of current making it through.
You asked for experts, and I'm not one. I should change my name to Dunning-Kruger. [-X

Here's another theory: Maybe the write current is fine, but the output stage is no longer receiving a meaningful signal due to something having blown up earlier in the signal chain. So what gets written to the disc is quiescent (a.k.a. a big load of nothing), or high-frequency noise.
This is an interesting thought. I think what you're asking is: what does a sector on the disc surface look like after a big load of nothing is written on top of it? Is there a faint remnant left of the old data?
That theory could at least be tested without destroying anything -- you'd need to temporarily disconnect (or maybe ground) the input to the amplifier that drives the write head. I actually have a spare drive here that I picked up for free from an ABUG meet, but it seems to have an integrated controller which stuffs all the gory details onto a single IC, so I don't think I can try it with this. Maybe if you could find an old one that has less integration.

Hopefully someone who actually knows what they're talking about will be along shortly
This can probably be tested directly with a Greaseweazle and a quickly hacked HFE to include zero flux pulses in a certain region.
Or you could do it directly from a beeb + 8271 with the "open the write gate outside a command" trick :-)


Cheers
Chris
User avatar
BigEd
Posts: 4099
Joined: Sun Jan 24, 2010 10:24 am
Location: West Country
Contact:

Re: Old McDonald's Farm by Edit

Post by BigEd »

Ah yes, good point, the write operation will need to have successfully read the sector header.

I wonder if a malfunction of the erase signal would do the job.
https://www.retrotechnology.com/herbs_s ... html#erase

Is the weak part of the signal precisely the data that the user's write will have written? Just a few sector contents, and the sector headers still pristine as-duplicated?
User avatar
scarybeasts
Posts: 797
Joined: Tue Feb 06, 2018 7:44 am
Contact:

Re: Old McDonald's Farm by Edit

Post by scarybeasts »

BigEd wrote:
Tue May 11, 2021 10:35 pm
Ah yes, good point, the write operation will need to have successfully read the sector header.

I wonder if a malfunction of the erase signal would do the job.
https://www.retrotechnology.com/herbs_s ... html#erase
I'd expect an erase signal failure to leave a strong signal, but with some possible corruption due to mixing with old data.
Is the weak part of the signal precisely the data that the user's write will have written? Just a few sector contents, and the sector headers still pristine as-duplicated?
Yes. The sector header signal is strong and well formed.


Cheers
Chris
User avatar
Diminished
Posts: 641
Joined: Fri Dec 08, 2017 9:47 pm
Contact:

Re: Old McDonald's Farm by Edit

Post by Diminished »

scarybeasts wrote:
Tue May 11, 2021 10:29 pm
This is an interesting thought. I think what you're asking is: what does a sector on the disc surface look like after a big load of nothing is written on top of it? Is there a faint remnant left of the old data?
Yeah.

I wonder if anyone in security circles would know. I'm sure that recovering deleted data from floppies was an important criminal forensic technique, once upon a time.
philpem
Posts: 693
Joined: Fri Apr 04, 2014 6:42 pm
Contact:

Re: Old McDonald's Farm by Edit

Post by philpem »

scarybeasts wrote:
Tue May 11, 2021 10:46 pm
I'd expect an erase signal failure to leave a strong signal, but with some possible corruption due to mixing with old data.
If you were about 180 degrees out of (signal) phase, you'd get destructive interference... weakening the read signal.
User avatar
scarybeasts
Posts: 797
Joined: Tue Feb 06, 2018 7:44 am
Contact:

Re: Old McDonald's Farm by Edit

Post by scarybeasts »

Diminished wrote:
Tue May 11, 2021 11:08 pm
scarybeasts wrote:
Tue May 11, 2021 10:29 pm
This is an interesting thought. I think what you're asking is: what does a sector on the disc surface look like after a big load of nothing is written on top of it? Is there a faint remnant left of the old data?
Yeah.
I think this could be it!

I wrote a sector with a bunch of "0" bits in the middle -- represented by 8us-spaced pulses on the disc surface. I then re-wrote the track but with no flux changes in that same area. It looks like this (full track view top, transition point zoom-in bottom):

nfa_overwrite_8us_1.png

And zooming in to 10mV/div and focusing on just one of the inverted pair of signals:

nfa_overwrite_8us_2.png

You can still clearly see the signal peaking every 8us. Surprisingly, It appears to have retained about 10% of its original strength. The signal picked up a bunch of noise in the process. The test point is post-filter on this drive, so the noise appears to have filtered to something like 2us cycles -- perhaps suggesting a 500kHz low-pass filter?

Caveats apply: you can bet this will differ wildly between drives. But it's the most plausible explanation so far as to what happened to these discs.


Cheers
Chris
User avatar
flaxcottage
Posts: 4635
Joined: Thu Dec 13, 2012 8:46 pm
Location: Derbyshire
Contact:

Re: Old McDonald's Farm by Edit

Post by flaxcottage »

I've had a look at this title. I have seen it before and have it somewhere but can't find it at the moment. (Isn't that always the problem! :lol: :? )

Technically the programming is not very good. This process of writing PAGE to a file so that the software knows its original value to get back to the menu screen by reading it before CHAINing the menu is bonkers. The file "page" on the disc will constantly be overwritten many times during running the title and it will always be in the same place. It is not surprising that the file was corrupted.

There does not seem to be a valid reason for saving page in this way - a system variable could be used far more effectively and safely.

However, the title does work well using an emulator.

From an SD card, USB stick, CF card, etc. one could store up problems since the number of writes to these devices is limited and the write will always be in the same location.

Thanks for posting the title. =D> I've worked out how it can be controlled though the manual is really necessary for full use.
- John

Check out the Educational Software Archive at www.flaxcottage.com
User avatar
flaxcottage
Posts: 4635
Joined: Thu Dec 13, 2012 8:46 pm
Location: Derbyshire
Contact:

Re: Old McDonald's Farm by Edit

Post by flaxcottage »

This title has now been uploaded to the educational archive.

The write of PAGE to the disc has been removed and the "page" file deleted. The image is now 200K in size although it will fit onto a 100K SSD image if required. The download zip file also includes some of the key presses needed to control the software.
- John

Check out the Educational Software Archive at www.flaxcottage.com
User avatar
Diminished
Posts: 641
Joined: Fri Dec 08, 2017 9:47 pm
Contact:

Re: Old McDonald's Farm by Edit

Post by Diminished »

scarybeasts wrote:
Wed May 12, 2021 1:03 am
I wrote a sector with a bunch of "0" bits in the middle -- represented by 8us-spaced pulses on the disc surface. I then re-wrote the track but with no flux changes in that same area. It looks like this (full track view top, transition point zoom-in bottom):
Interesting stuff. Thanks for going to the trouble of checking it.
User avatar
scarybeasts
Posts: 797
Joined: Tue Feb 06, 2018 7:44 am
Contact:

Re: Old McDonald's Farm by Edit

Post by scarybeasts »

flaxcottage wrote:
Wed May 12, 2021 11:30 am
I've had a look at this title. I have seen it before and have it somewhere but can't find it at the moment. (Isn't that always the problem! :lol: :? )
Ah, per Chris Jones on Twitter, this title was already posted to the forums:
viewtopic.php?f=2&t=7190&p=259236#p259236

I totally missed it and had thought Old Macdonald's Farm was MIA.

The disc catalog layout is very close between the two images. One interesting difference is the DFS title. It's OLD.MAC in the previous post and 1187V1.0 in the image billcarr2005 and I worked on.

The actual story book on the discs is very different though! Page 1 of the OLD.MAC version is called "The Ghost Farm" and there are all the animal graphics scattered around. It seems possible this version was edited by a schoolkid :) The 1187V1.0 version seems like a more plausible default story book...


Cheers
Chris
User avatar
billcarr2005
Posts: 1692
Joined: Fri Sep 09, 2005 4:01 pm
Location: UK
Contact:

Re: Old McDonald's Farm by Edit

Post by billcarr2005 »

I commented on that thread too :lol:
then promptly forgot about it :oops:

but as per viewtopic.php?p=259083#p259083 the discs imaged weren't original, so it's good to have a "clean" copy! :)
nicolagiacobbe
Posts: 37
Joined: Tue Jul 03, 2007 10:40 am
Location: italy
Contact:

Re: Old McDonald's Farm by Edit

Post by nicolagiacobbe »

I have seen that scarybeasts reached the hackaday fame. The web page of it make a really superb history.

Willing to give my 2c I wonder if the raw data dump of defective sections could be made available, I'd like to try to apply some DSP techniques and see what could happen.
User avatar
scarybeasts
Posts: 797
Joined: Tue Feb 06, 2018 7:44 am
Contact:

Re: Old McDonald's Farm by Edit

Post by scarybeasts »

nicolagiacobbe wrote:
Fri May 21, 2021 2:34 pm
I have seen that scarybeasts reached the hackaday fame. The web page of it make a really superb history.

Willing to give my 2c I wonder if the raw data dump of defective sections could be made available, I'd like to try to apply some DSP techniques and see what could happen.
That would be amazing. Would love to hear how you get on.

Track 0, containing two very faint sectors, is here: https://drive.google.com/file/d/1CgrO3B ... sp=sharing

Audacity will load this via Tools -> Sample Data Import and you can export as a WAV from there if that's a preferred format. (Make sure to use at least 16-bit, as the sample doesn't use the full volume range.)

The FM encoded sector data for the first sector starts at sample 579960 or so. The correct ASCII decode for the first few bytes is, I believe, 1187V1.0


Cheers
Chris
Post Reply

Return to “archive submissions”