This page contains information about imaged Atari FD that I have been preserved using a KryoFlux device. It is related to the KryoFlux page.
Examples of Analysis with the KFAnalyze Program
Following are some examples of output from the KFAnalyze program. The raw file corresponding to the track analyzed are provided so you can run the KFAnalyze program by yourself on these examples. Note that these test cases has been chosen because they cover several interesting disk copy protections mechanisms. For more information you can go to this page Atari Floppy Disk Protection / Preservation as well as the page from Markus Fritze.
Note: you can click on any image to get a full size display.
Turrican
Turrican contains many interesting protection mechanisms. Here we are looking at track 7. We first look at the layout of all sectors of track 7. As we can see (you need to click on the image to look at a full size picture) the first sector 3 does not start at the begining of the track. This is because the last sector (at bottom) 5 starts close to the index and therefore has data over the index. We can also see that sectors 0, 16, and 1 overlaps (sector within sector). And the first two (sector 0 and 16) have fuzzy bytes as indicated with the orange color.
The following plot show sector 0 of the track 7. As we can see we have a huge section of the sector without any flux transitions. Of course this violates the MFM rules and causes the track to be decoded incorrectly.
Back to the topDungeon Master
This games uses several protection mechanisms to render the FD impossible to copy with standard Atari Hardware. For detail analysis of the Dungeon Master & Chaos Strike Back protections please refer to the DM Protection document, the detailed analysis of the Dungeon Master and Chaos Strikes Back for Atari ST Floppy Disks as well as the US patent “Copy Protection for computer Disc 4,849,836”
The game “Dungeon Master” uses the following protection mechanisms:
- Invalid Sector Number: Track 0 the sector 8 is numbered 247.
- Fuzzy bits & Sector with bad Data: Track 0 sector 7.
All the Protections are located in track 0 of the FD.
Here is the Layout of track 0 as analyzed by the KFAnalyze program:
****************************************************************************
Track Layout Information: 6258 Bytes - length=199.981 ms
ID Good/Bad=10/0 - Data Good/Bad=9/1 - Sync Good/Bad =20/0
****************************************************************************
GAP1 56 bytes length=1818.28 us
-----------+------------------+-----------+---------------------------+------------
GAP2 |ID |GAP3 |DATA |GAP4
Bt Lgt |Sct Pos Lgt CRC|Bt Lgt BS|Bt Lgt CRC TMV BRD Clk |Bt Lgt BS
-----------+------------------+-----------+---------------------------+------------
15 458 |1 2276 223 OK |37 1179 0|515 16433 OK 0 0 3.99|41 1307 0
15 478 |2 21899 223 OK |37 1179 0|515 16493 OK 0 0 4.00|41 1308 0
15 477 |3 41581 222 OK |37 1177 0|515 16452 OK 0 0 3.99|41 1305 0
15 477 |4 61217 222 OK |37 1174 0|515 16392 OK 0 0 3.98|41 1302 0
15 475 |5 80785 222 OK |37 1171 0|515 16451 OK 0 0 3.99|41 1313 0
15 480 |6 100425 224 OK |37 1186 0|515 16525 OK 0 0 4.01|41 1308 0
15 477 |7 120148 222 OK |37 1174 0|515 16506 BAD 0 495 4.01|41 1313 0
15 479 |247 139845 223 OK |37 1179 0|515 16410 OK 0 0 3.98|41 1304 0
15 476 |9 159441 222 OK |37 1173 0|515 16418 OK 0 0 3.98|41 1311 0
15 480 |10 179047 223 OK |37 1181 0|515 16536 OK 0 0 4.01|93 2991 0
-----------+------------------+-----------+---------------------------+------------
As you can see in sector 7 we have a lot of border bits (BRD), aka bits in uncertain area, and therefore this results in Fuzzy bytes when reading this sector several times. These border bits are obtained by using a sliding pattern of the flux transitions close to the border of the inspection window. This is shown in the following diagram (click on the image for a full size display) - Click here for a high definition eps graphic file.
If we zoom on the start of the sector we can see that the beginning uses normal timing but after the position 122000 ns we have the bit transitions gradually sliding to the border of the inspection window (close to the 5000 ns line). We can see that we have a pattern that looks like a sine wave and this implies that many bits are at the border of the inspection window (shown as green dots). More information on US Patent 4,849,836.
As explained in the WD1772 DPLL Input Circuitry, having transitions at the border of the inspection windows will result in random value latched by the DPLL data separator and therefore these bits can be considered as Fuzzy Bits. Reading this sector several times will results in different values returned due to the floppy disk rotation speed fluctuations.
Back to the topPopulous
Populous from Electronic Arts uses the several protection mechanisms. We are going to look at the usage of long sector from the Rob Northern copylock protection. Here we can see that Track 0 sector 6 is has a “long data sector” of 17197µs which is about 4.2% above a normal sector of 16480µs. We can see that this is done by changing the clock from 4µs to 4.2µs after the begining of sector 6 and get back to normal value after the end of the sector (click on the image for a full size display).
Back to the topD50 Editor music utility from DrTs
The D50 Sound Module Editor program from DrT uses many protection mechanisms; Here we are looking at track 0. In the last sector of the track we can see that the flux transitions looks random. This is typical of an unformatted track. So we can see that the beginning of the sector looks almost normal and the end of the sector is unformatted. I fwe look more carefully at sector 10 we can see that the flux transitions follow a strange pattern at the begining (with many border bits) and after that the flux transitions look pseudo-random with a lot of timing violations. As you can see the PLL of the WD1772 DPLL gets totally crazy and this results in many fuzzy bits. Note that the violations seems to be done symetrically so thatr the overall resulting lenght of the track is not changed.
Back to the topColorado
The colorado games uses the Macrodos protection from Speedlock. In this protection a sector is divided into four segment and inside this segment the clock is raised and lowered (intra-sector bit rate variation). The sector look like this
Theme Park Mystery
Use a Rob Northern protection with two overlapping sector at end of track that reads as data over index and with fuzzy bits. Not however than this is not realy Data Over Index has there is no room for these sectors at the begining of the track. So it is fake DOI.
Why do they also read with fuzzy bits. Well if you look at the flux transitions in sector 11 you will see that it contains a Non Flux Area. The NFA starts at about 194350µs position and endup very close to the end of the track (200000µs). As for Turican presented above the sector 12 is in fact synchronized on the clock bits of the sector 11 and therefore shifted by an half clock. Therefore even if the two sectors overlap their content is totally different (as usual click on picture for full size image)
Back to the topTest Cases
I do not provide complete images of FD for copyright reason. However I do provide tracks from several games to illustrate some of the protection mechanisms used in the Atari World. You can analyze these images with the KFAnalyze program.
- Populous track 00 (Rob Nothern protection)
- Turrican track 07 (no flux transition area)
- Dungeon Master track 00
- D50 Editor (half formatted sector)
- Colorado track 1 (Macrodos protection)