Ape Cosplay
Professional
Everytime I work on a board for a repair I look for undumped devices including PROMs or PLDs.
This time I completed a Smash T.V. repair for our dear @Mitsurugi-w, then proceeded to check which PLDs were already dumped on the PLD archive (https://wiki.pldarchive.co.uk/index.php?title=Main_Page). Only 3 of them were on the main board: U40 (PLS153) and U41 & 42 (PAL16L8 ) even though labels didn't match my board (seemingly a mistake on the PLD archive page). Reason for that is the remaining ones are an unusual kind of beasts, namely EP600s (5C060 equivalents), and they were locked of course. They are a bit more complex than PALs/GALs with 16 macrocells and 2 dedicated clock inputs (1 for each bank of 8 macrocells).
For my first try I targetted U52 since it's registered (U53 seems to be purely combinatorial), I was a bit concerned about those two clocks, but the dumping software, due to some overlooking during development, allows to define more than 1 clock pin, though they will get the same signal/state, but luckily they were tied together on the board.
The second thing is I haven't reversed the fusemap yet, so only equations will be reversed. That's actually not too much of a problem as the version of the MAX+plus II software I have supports EP600/EP610 devices under the CLASSIC category, therefore I should be able to generate a jed file.
Brute-forcing and equation reversing went well.
MAX+plus II managed to compile the code and generate a jed file.
Which I tested in a real EP600
Success!
[EDIT 1]
U52 (A-5346-3044-5) needs 14 macrocells, 6 of them being used for the watchdog generation. However, from the schematics it turns out that signal feeds pin 11 (!WDI) of the MAX691 but the corresponding output (pin 14 = !WDOUT) has been left unconnected. This has been confirmed on real hardware, so by dropping the watchdog logic it could fit in a GAL22V10 with a small adapter to reroute the signals.
[EDIT 2]
I have now reversed the 3 other EP600 devices from that board, jed files provided are for EP600s (in the next post) for direct drop in replacement:
- U53 (A-5346-3044-6): purely combinatorial, 10 MCs, fits in a GAL22V10 by rerouting pins 3, 4, 14 and 23 (tested). It is also interesting that the AB direction selection signal for U94/U110 (pin 15) is a fixed 1.
- U8 (A-5346-3044-1): that one was the biggest surprise, despite having both clock inputs connected they're actually not used. So as many other signals... In the end it only requires 5 MCs (4 inputs, 5 outputs) and is purely combinatorial as well. Using the same adapter than for U53 I have been able to test it in a GAL22V10 but with more rerouting it would also fit in a GAL20V8 or GAL16V8.
- U65 (A-5346-3044-7): the last Midway's joke. This time registered, but not using the LD12~15 inputs and with floating (permanent hi-Z) EXT_T1 output (pin 5). Needs 6 MCs, and a lot of rerouting to fit in a GAL22V10.
[EDIT 3]
The last PLD to be reversed on Smash T.V. is U115 (A-5346-3044-8 ): EP900 in DIP40, 24 MCs but I've done bigger chips before (like the MACH111 on the Batman Forever sound PCB with 32 MCs), the only limitation being one output can't be dependent of more than 21 other signals. However I don't want to further risk a board that isn't mine, we're already lucky to have all four EP600s now.
Chips/PCBs donations are welcome if you want to support the cause (contact @Darksoft , @Mitsurugi-w or myself in PM).
This time I completed a Smash T.V. repair for our dear @Mitsurugi-w, then proceeded to check which PLDs were already dumped on the PLD archive (https://wiki.pldarchive.co.uk/index.php?title=Main_Page). Only 3 of them were on the main board: U40 (PLS153) and U41 & 42 (PAL16L8 ) even though labels didn't match my board (seemingly a mistake on the PLD archive page). Reason for that is the remaining ones are an unusual kind of beasts, namely EP600s (5C060 equivalents), and they were locked of course. They are a bit more complex than PALs/GALs with 16 macrocells and 2 dedicated clock inputs (1 for each bank of 8 macrocells).
For my first try I targetted U52 since it's registered (U53 seems to be purely combinatorial), I was a bit concerned about those two clocks, but the dumping software, due to some overlooking during development, allows to define more than 1 clock pin, though they will get the same signal/state, but luckily they were tied together on the board.
The second thing is I haven't reversed the fusemap yet, so only equations will be reversed. That's actually not too much of a problem as the version of the MAX+plus II software I have supports EP600/EP610 devices under the CLASSIC category, therefore I should be able to generate a jed file.
Brute-forcing and equation reversing went well.
MAX+plus II managed to compile the code and generate a jed file.
Which I tested in a real EP600
Success!
[EDIT 1]
U52 (A-5346-3044-5) needs 14 macrocells, 6 of them being used for the watchdog generation. However, from the schematics it turns out that signal feeds pin 11 (!WDI) of the MAX691 but the corresponding output (pin 14 = !WDOUT) has been left unconnected. This has been confirmed on real hardware, so by dropping the watchdog logic it could fit in a GAL22V10 with a small adapter to reroute the signals.
[EDIT 2]
I have now reversed the 3 other EP600 devices from that board, jed files provided are for EP600s (in the next post) for direct drop in replacement:
- U53 (A-5346-3044-6): purely combinatorial, 10 MCs, fits in a GAL22V10 by rerouting pins 3, 4, 14 and 23 (tested). It is also interesting that the AB direction selection signal for U94/U110 (pin 15) is a fixed 1.
- U8 (A-5346-3044-1): that one was the biggest surprise, despite having both clock inputs connected they're actually not used. So as many other signals... In the end it only requires 5 MCs (4 inputs, 5 outputs) and is purely combinatorial as well. Using the same adapter than for U53 I have been able to test it in a GAL22V10 but with more rerouting it would also fit in a GAL20V8 or GAL16V8.
- U65 (A-5346-3044-7): the last Midway's joke. This time registered, but not using the LD12~15 inputs and with floating (permanent hi-Z) EXT_T1 output (pin 5). Needs 6 MCs, and a lot of rerouting to fit in a GAL22V10.
[EDIT 3]
The last PLD to be reversed on Smash T.V. is U115 (A-5346-3044-8 ): EP900 in DIP40, 24 MCs but I've done bigger chips before (like the MACH111 on the Batman Forever sound PCB with 32 MCs), the only limitation being one output can't be dependent of more than 21 other signals. However I don't want to further risk a board that isn't mine, we're already lucky to have all four EP600s now.
Chips/PCBs donations are welcome if you want to support the cause (contact @Darksoft , @Mitsurugi-w or myself in PM).
Last edited:
