obsolete's SE/30 Reloaded Build

[Bolle]

My suggestion would be to target the IIcx GLUE first

II, IIx, IIcx and SE/30 GLUEs are identical. The video part of the SE/30 is done entirely in the 4 PALs that act like a PDS card to the rest of the system.

 

[Melkhior]

II, IIx, IIcx and SE/30 GLUEs are identical. The video part of the SE/30 is done entirely in the 4 PALs that act like a PDS card to the rest of the system.

All the better. The II and IIx don't have subtle difference due to the slightly weird memories? The nice thing about the IIcx is that in addition to the simplicity, it's bog-standard FPM memory and I believe works out-ot-the-box with the the IIsi's 32-bit clean memory. If the Iix and SE/30 have the exact same GLUE and can use the same well-known ROM (might need to disable some stuff if some hardware is missing at first), then they're also fine.

I suggest staying a long way from the original II, as otherwise one needs to deal with the complexity of the external MMU. Not worth the hassle vs. doing a '030 system. For me the value of the II is limited to the original board, if one could trace + reverse engineer the discrete NuBus circuitry it would help with recreating a NuChip at a later stage. Probably could be done without it, but additional intel is always useful.

 

[Bolle]

The II and IIx don't have subtle difference due to the slightly weird memories?

Nope, they fixed the memory issue externally in a PAL on the IIcx and SE/30.
I do have one of my Mac II boards modified in a similar fashion and it runs just fine with high >1MB RAM SIMMs.

 

[croissantking]

I do have one of my Mac II boards modified in a similar fashion and it runs just fine with high >1MB RAM SIMMs.

How does the mod look?

 

[obsolete]

Heya Obsolete. I have LOTS of 27C512's (256's too, but did I mention I have a lot of 512's?). Send me a message if you need a couple and I'll send them your way gratis.

@Callan Hmm, looking at the datasheets, replacing a 27C64 with a 27C512 seems like it would be slightly tricky, but doable. From the SE/30 schematic, pins 27 and 1 are tied high, which are A15 and A14 on the 27C512, so the ROM image would need to be written at offset 0xC000. Also, because pin 26 (A13 on the 27C512) appears to be floating (haven't checked on the HW yet, just looking at the schematic), it seems like it would be a good idea to write another copy at 0xE000, because A13's state is going to be indeterminate unless I stick a pulldown resistor on it. Actually, it would be easier to pull it up, just bridge it with its neighbor A14 on pin 27. Then I could just write a single copy of the ROM starting at 0xE000, though the end of the chip.

Could someone please check my logic here?

 

[Callan]

@Callan Hmm, looking at the datasheets, replacing a 27C64 with a 27C512 seems like it would be slightly tricky, but doable. From the SE/30 schematic, pins 27 and 1 are tied high, which are A15 and A14 on the 27C512, so the ROM image would need to be written at offset 0xC000. Also, because pin 26 (A13 on the 27C512) appears to be floating (haven't checked on the HW yet, just looking at the schematic), it seems like it would be a good idea to write another copy at 0xE000, because A13's state is going to be indeterminate unless I stick a pulldown resistor on it. Actually, it would be easier to pull it up, just bridge it with its neighbor A14 on pin 27. Then I could just write a single copy of the ROM starting at 0xE000, though the end of the chip.

Could someone please check my logic here?

I have 2764's too. I have pretty much any eprom you need bud (2516/2716 up). Byproduct of being a game tech for over 25 yrs. Let me know what size / speed you need and I'll send you a couple. I can program them for you too if you send me the bin.

 

[obsolete]

@Melkhior With the relatively large footprint of the 84-pin PLCC, I think a carrier PCB with small surface-mounted FPGA, level-shifters, etc. would be feasible vs. an ASIC.

The IIcx also has the ASC (344S0063-A), but I guess if you can live without a SWIM, you can live without sound too?

 

[obsolete]

I have 2764's too. I have pretty much any eprom you need bud (2516/2716 up). Byproduct of being a game tech for over 25 yrs. Let me know what size / speed you need and I'll send you a couple. I can program them for you too if you send me the bin.

You've got mail!

 

[Melkhior]

The IIcx also has the ASC (344S0063-A), but I guess if you can live without a SWIM, you can live without sound too?

Forgot about those but yes, the assumption is sound isn't a critical feature so can be ignored in a first draft. And it's one of the thing for which I did write a driver (for HDMI audio), so there's a path to an alternate solution to add sound later.

 

[croissantking]

@Callan Hmm, looking at the datasheets, replacing a 27C64 with a 27C512 seems like it would be slightly tricky, but doable. From the SE/30 schematic, pins 27 and 1 are tied high, which are A15 and A14 on the 27C512, so the ROM image would need to be written at offset 0xC000. Also, because pin 26 (A13 on the 27C512) appears to be floating (haven't checked on the HW yet, just looking at the schematic), it seems like it would be a good idea to write another copy at 0xE000, because A13's state is going to be indeterminate unless I stick a pulldown resistor on it. Actually, it would be easier to pull it up, just bridge it with its neighbor A14 on pin 27. Then I could just write a single copy of the ROM starting at 0xE000, though the end of the chip.

Could someone please check my logic here?

You're along the right lines I guess, since burning the VROM onto a 27C256 (without adjusting any offsets, etc.) does not work in my SE/30.

I have one more of these chips to experiment with, but I since they're write once, I guess the first one is junk.

 

[Melkhior]

I have one more of these chips to experiment with, but I since they're write once, I guess the first one is junk.

You can try 'bending away' the problematic address pin (so they are no longer in the socket/PCB) and soldering them using some patch wire to low instead of high (or the reverse). if the chip is toast anyway, it doesn't risk much, you just have to make sure you don't short anything.

 

[obsolete]

You're along the right lines I guess, since burning the VROM onto a 27C256 (without adjusting any offsets, etc.) does not work in my SE/30.

I have one more of these chips to experiment with, but I since they're write once, I guess the first one is junk.

Let's see, on a 27C256 you've got A14 tied high and A13 floating, so in theory you'd need two copies of the ROM, one at 0x4000 and one at 0x6000, to account for the indeterminate state of A13.

Again, I'd like someone with some more experience to verify that my thinking here is correct.

 

[obsolete]

You can try 'bending away' the problematic address pin (so they are no longer in the socket/PCB) and soldering them using some patch wire to low instead of high (or the reverse). if the chip is toast anyway, it doesn't risk much, you just have to make sure you don't short anything.

True, if you were to bend away pins 26 and 27 (A13 and A14) and ground them, it should work.

 

[obsolete]

Forgot about those but yes, the assumption is sound isn't a critical feature so can be ignored in a first draft. And it's one of the thing for which I did write a driver (for HDMI audio), so there's a path to an alternate solution to add sound later.

I forgot about that project, very cool!

 

[croissantking]

Let's see, on a 27C256 you've got A14 tied high and A13 floating, so in theory you'd need two copies of the ROM, one at 0x4000 and one at 0x6000, to account for the indeterminate state of A13.

Again, I'd like someone with some more experience to verify that my thinking here is correct.

Well, I burned two more copies of the ROM, at these offsets, onto the chip I'd just incorrectly burned. Turns out you can write to the chip again, just not to previously burned sectors. So the chip now has three copies of the ROM - at 0x0000, 0x4000 and 0x6000 - and it works great, just verified.

 

[obsolete]

Well, I burned two more copies of the ROM, at these offsets, onto the chip I'd just incorrectly burned. Turns out you can write to the chip again, just not to previously burned sectors. So the chip now has three copies of the ROM - at 0x0000, 0x4000 and 0x6000 - and it works great, just verified.

Hooray!

 

[croissantking]

Hooray!

Yes, awesome!

I wonder if it's currently reading the copy of the ROM at 0x4000 or 0x6000

pin 26 (A13 on the 27C512) appears to be floating

Could this cause any issues, even with the two copies of the ROM?

 

[croissantking]

Right, so looking with a multimeter, A13 (pin 26) is at 0.5V with my SE/30 powered on. Therefore it's accessing the code starting at 0x4000, I take it.

 

[obsolete]

Right, so looking with a multimeter, A13 (pin 26) is at 0.5V with my SE/30 powered on. Therefore it's accessing the code starting at 0x4000, I take it.

Yeah, but just by touching it with a meter, you may have changed its state
Doesn't matter though, because with copies at both possible offsets, you're covered.

 

[croissantking]

Yeah, but just by touching it with a meter, you may have changed its state
Doesn't matter though, because with copies at both possible offsets, you're covered.

Is this quantum computing then?