Identify these weird EPROMs

[ieure]

I was given a fairly large box of ICs, dating from the 1970s up to the 1990s.  A lot of stuff that seems more or less computer related; Z80s, 6502s, and various support chips; 74HC TTLs in both PDIP and SOIC packages; PALs; 1980s 2114/4116/6116 RAMs; and so forth.

But these four really jumped out at me.  I'm certainly no EE, but I've been inside way more than my share of electronics, and I've never seen anything quite like this.

[attach=1]

Given the '82-'83 date codes, my best guess is that they're very early production 27256 EPROMs made from four 2764s glued together, with the PAL decoding the two extra address lines.

Anyone seen these before, or know the part number?  Think they'd program as MBM27256s in my UniSite?

[Alex Eisenhut]

Beautiful. How much you want for one? Erasing one of these must have been fun. Ding! Time to flip it over!

[TheSteve]

Can't give any info on them but they are pretty cool. Have you tried to read one?

[ieure]

Can't give any info on them but they are pretty cool. Have you tried to read one?

Hmmm, while I don't think I have any use for them, I'm also not sure I want to sell them.  But since I haven't the slightest idea what they'd be worth, it's a moot point.

[ieure]

Can't give any info on them but they are pretty cool. Have you tried to read one?

I haven't.  Given condition and that most of the parts in this lot are NOS (some in factory packaging), I expect they're blank.  But I'll give it a whirl tomorrow.

[pcprogrammer]

They do look cool indeed.

I have used my fair share of 2764 eproms from that era, but never came across these where they are combined to a bigger one. I tried looking for the PAL1018 datasheet, but no dice. The top sides show a marking with 8331, but that gives no hits either.

Some company went to extremes to make these. The F on the 2764's might indicate Fairchild semiconductor.

[Nusa]

Try reading them before you worry about erasing and programming them.

[AndyBeez]

They look military in application. These certainly would have cost $$$$$s in the day. Check out that SMT capacitor. A really cool and potentially profitable find.

[PA0PBZ]

I tried looking for the PAL1018 datasheet, but no dice.

PAL10L8 , see attachment. They look cool indeed and I guess op is right, it combines the 4 2764 chips into a bigger one. If it's compatible with 27256 has to be seen.

[RoGeorge]

Never seen anything alike before, looks cool! 

[Alex Eisenhut]

I assume 8331 is the date code of the combined thing, the Fujitsu EPROMs and the MMI PAL. I've seen this type of hybrid approach for "4132" or "4332" 32K x 1 DRAMs (in an Apple II I think?), and of course the kind of cost-is-no-object workstations from the 1980s.

https://www.x86-guide.net/xhoba/en/collection_divers/Dec-VAX-V11-ROM-RAM-chip-cpu-no291.html

But I've never seen something populated with active circuitry top and bottom.

OP, 50$ US + shipping for one?

[wraper]

These are not EPROMs but some custom modules consisting of 4 2764 EPROMs and PAL. I really doubt about them working as 27256, you don't need PAL for that.

[Benta]

These are not EPROMs but some custom modules consisting of 4 2764 EPROMs and PAL. I really doubt about them working as 27256, you don't need PAL for that.

Actually you do need the super-simple PAL for the chip selects. A simple 74xx decoder could probably also have done it, but there was not 100% agreement on the 2764 pinout back then (upper address and V_PP). A PAL was safer.

It's "intermediate" 27256 EPROMs from before the real thing became available. Notice how the dies fill the windows completely, meaning it's the maximum die size for the stepper camera and the lead frame.
The "real" 27256s are a couple of geometry shrinks away, probably 2...3 years in those days.
Expensive solution, but apparently necessary.

And extremely cool! 

[wraper]

These are not EPROMs but some custom modules consisting of 4 2764 EPROMs and PAL. I really doubt about them working as 27256, you don't need PAL for that.

Actually you do need the super-simple PAL for the chip selects. A simple 74xx decoder could probably also have done it, but there was not 100% agreement on the 2764 pinout back then (upper address and V_PP). A PAL was safer.

How PAL is supposed to help with Vpp? There is no voltage compatibility. Can you give an example of 2764 with a different pinout? Half of 74LS139 or 74LS138 would do the job.

[TheSteve]

I feel like they were probably the worlds first DS/HD (Double Sided High Density) EPROM's.

[rsjsouza]

Pretty cool find. They are most probably not a simple larger EPROM but instead a custom hybrid IC - so common in those days such as the Sanyo STKs. What is unique about those is they are not fully enclosed in plastic/epoxy packaging of other common devices.

[Benta]

How PAL is supposed to help with Vpp? There is no voltage compatibility. Can you give an example of 2764 with a different pinout? Half of 74LS139 or 74LS138 would do the job.

What's up with you, why do you carry such a chip on your shoulder?
You take a longer post, pick one teeny piece out of it and blow it out of all proportion.

I never said that VPP is part of the PAL circuit. The 2764 pinout in 1982 was not homogenous concerning upper address, VPP and chip selects.

On top of that, no such thing as 74LS or 74HC existed in SO-packages or equivalent. It was DIP back then (or die or MIL-packaging). Period.
The PAL10L8 was probably chosen for its availability in CLCC package.

[wraper]

How PAL is supposed to help with Vpp? There is no voltage compatibility. Can you give an example of 2764 with a different pinout? Half of 74LS139 or 74LS138 would do the job.

What's up with you, why do you carry such a chip on your shoulder?
You take a longer post, pick one teeny piece out of it and blow it out of all proportion.

I never said that VPP is part of the PAL circuit. The 2764 pinout in 1982 was not homogenous concerning upper address, VPP and chip selects.

On top of that, no such thing as 74LS or 74HC existed in SO-packages or equivalent. It was DIP back then (or die or MIL-packaging). Period.
The PAL10L8 was probably chosen for its availability in CLCC package.

Look, you made a flawed argument in my opinion. Not to say argument about my post you quoted. If PAL does not fully solve pinout mismatch, there is no point using it for EPROM interchangeability. Secondly, identical 54 series existed in ceramic SMD package similar to this PAL. Look into datasheet https://www.ti.com/lit/ds/symlink/sn54ls138.pdf?ts=1675288001960&ref_url=https%253A%252F%252Fwww.ti.com%252Fproduct%252FSN54LS138

[Benta]

Yeah, yeah, you're right and I have my peace.
Feel better?

[BrianHG]

How PAL is supposed to help with Vpp? There is no voltage compatibility. Can you give an example of 2764 with a different pinout? Half of 74LS139 or 74LS138 would do the job.

What's up with you, why do you carry such a chip on your shoulder?
You take a longer post, pick one teeny piece out of it and blow it out of all proportion.

I never said that VPP is part of the PAL circuit. The 2764 pinout in 1982 was not homogenous concerning upper address, VPP and chip selects.

On top of that, no such thing as 74LS or 74HC existed in SO-packages or equivalent. It was DIP back then (or die or MIL-packaging). Period.
The PAL10L8 was probably chosen for its availability in CLCC package.

Look, you made a flawed argument in my opinion. Not to say argument about my post you quoted. If PAL does not fully solve pinout mismatch, there is no point using it for EPROM interchangeability. Secondly, identical 54 series existed in ceramic SMD package similar to this PAL. Look into datasheet https://www.ti.com/lit/ds/symlink/sn54ls138.pdf?ts=1675288001960&ref_url=https%253A%252F%252Fwww.ti.com%252Fproduct%252FSN54LS138

You cheap slow 74LS138 added up to 41ns of propagation delay to what was already a slow 200ns eprom.  How fast was the ' MMI PAL10L8-2ML ' when doing a simple 2 to 4 address decode mux?  IE: only 3 of 8 product terms for each output...  I bet it was a crap load faster...  Also, don't forget you may be also emulating the OE and CE, not just the upper addresses as a control input element.  IE: when programming 1 of 4, you will need to operate the CE while keeping all OEs disabled.  When reading, the OE are preferable for address switching over using the CEs as the CEs do power-down the eprom and slow things down to awaken.  This added switch in functional wiring/control makes a PAL far better suited while you would need some more additional logic ICs to wire all this up.

[ieure]

I feel like they were probably the worlds first DS/HD (Double Sided High Density) EPROM's.

Quad-core EPROMs!

[wraper]

You cheap slow 74LS138 added up to 41ns of propagation delay to what was already a slow 200ns eprom.  How fast was the ' MMI PAL10L8-2ML ' when doing a simple 2 to 4 address decode mux?  IE: only 3 of 8 product terms for each output...  I bet it was a crap load faster... 

This PAL has larger propagation delay than 74LS139. If needed faster there were 74/54 S, AS, H, F series too.

How fast was the ' MMI PAL10L8-2ML ' when doing a simple 2 to 4 address decode mux?

Its speed does not depend on what's it's doing as it's a very simple device.

Also, don't forget you may be also emulating the OE and CE, not just the upper addresses as a control input element.  IE: when programming 1 of 4, you will need to operate the CE while keeping all OEs disabled.  When reading, the OE are preferable for address switching over using the CEs as the CEs do power-down the eprom and slow things down to awaken.  This added switch in functional wiring/control makes a PAL far better suited while you would need some more additional logic ICs to wire all this up.

Well, you should switch only either CE or OE, with other connected together between EPROMs and connected to external pin. You can switch OE or CE with 74LS139 and use it's enable pin as external OE or CE.
EDIT: regarding programming while switching OE, not CE for better speed, yes PAL makes sense. For example you could use second half of LS139 connected to CE. However a little bit of additional logic would be needed for enabling all CE outside of programming.

[wraper]

BTW I never claimed It cannot possibly be 27256. My argument is that using PAL for that is unnecessary, wasteful and requires more work as you need to program PAL before attaching to the package.

[wraper]

While thinking more, I realized there is not enough pins in the package for 27256 having PGM input. And I fail to see how PAL would solve this problem to be able to program 2764 within module.
EDIT: maybe it's what those apparently resistors are for. Maybe a voltage divider from Vpp to PAL input to activate an alternate PGM input mode for CE pin.

[amyk]

You're not the only one on the Internet to speak of such a thing: https://forums.arcade-museum.com/threads/show-n-tell-time-take-a-look-mbm2764-20-fujitsu-odds-are-you-never-seen-one-like-this-before.478300/