OT: stacked (DIP) chips

[BigEd]

I thought this should be of interest: I collected some links to designs using stacked chips - mostly DIP chips - to get high density or to expand an existing circuit board with minimal work.

Richard Goedeken wanted to switch kernal ROMs in his Plus 4
http://www.fascinationsoftware.com/FS/h ... witch.html


Duane Elscott made 4 towers of 16 SRAMs for a wirewrapped graphics board. From
http://web.archive.org/web/200406112215 ... tt/mem.htm
"If you're wondering why one of the stacks in the picture is taller than the rest, those were a different brand of ICs, with a slightly thicker body."


Timo Noko stacked 35 DRAMs to provide tagged lisp node storage in an Osborne luggable.
http://anycpu.org/forum/viewtopic.php?t=235&p=1456


Garth Wilson stacks four 512Kx8 SRAMs on a DIP header to get 2Mx8, with a 74AC139 stood on edge at the end to select the right chip.
viewtopic.php?p=7564#p7564


Garth Wilson's stack of 8 chips with a side order of decoder
viewtopic.php?f=4&t=2322&start=30


(More from Garth on this page)

Duane Degn stacks 8 serial chips to make a byte serial memory
http://forums.parallax.com/discussion/1 ... it-version


Quinn Dunki's Veronica machine piggybacks two RAMs
http://quinndunki.com/blondihacks/?p=1322


IBM's own 64k memory expansion for the 5150 PC used stacked chips.
http://www.minuszerodegrees.net/5150/ea ... _early.htm


And again: this pair of 64Kbit, plastic-packaged DRAMs in DIPs have been stacked and soldered together at the Texas Instruments factory to produce a 128Kbit device with a small footprint. This product is destined for a socket in an IBM PC/AT clone that had broken the 640K barrier"
https://eda360insider.wordpress.com/201 ... packaging/


Stacked chips in John Draper’s 24k Apple II, from "Captain Crunch’s Memory Modification" at
http://www.willegal.net/blog/?p=7246


Ronald Dekker stacks EPROM and SRAM in his "tiny but powerful stand alone BASIC microcomputer, not larger than the size of a package of cigarettes built entirely from salvaged components, just for fun!"
See
http://www.dos4ever.com/8031board/8031board.html


"This early 1980s “advanced” calculator board included dual in-line packages (DIPs) stacked on top of each other."
http://electronicdesign.com/embedded/pa ... rospective


Dave Curran discovered this stacked rom version switcher when refurbishing a ZX80
http://blog.tynemouthsoftware.co.uk/201 ... epair.html


[Previously posted on G+]

[Dr Jefyll]

Wow! -- some great work there. The byte serial memory is especially clever. And the 35-bit DRAM... Well!

I myself have often stacked chips, usually in unremarkable (though highly expedient) ways. But the photos below show a slight departure from the norm.

What we see is a Z80 SBC that features an 'hc574 used as an 8-bit output port. But because it was sometimes necessary to update fewer than all 8 bits there was a need to know what the existing data was. At first I managed that by maintaining a duplicate copy of the port's contents in RAM -- a workable but slightly cumbersome solution. It'd be better if the port itself could be read back...

Readback capability took the form of an 'hc245 soldered on top of the '574. As you can see, it was necessary to stagger the pins to make the appropriate connections line up!

-- Jeff

[BigDumbDinosaur]

Dr Jefyll wrote:

Readback capability took the form of an 'hc245 soldered on top of the '574. As you can see, it was necessary to stagger the pins to make the appropriate connections line up!

Yuh got all the angles covered.

[BigDumbDinosaur]

BigEd wrote:

I thought this should be of interest: I collected some links to designs using stacked chips - mostly DIP chips - to get high density or to expand an existing circuit board with minimal work.

Duane Elscott made 4 towers of 16 SRAMs for a wirewrapped graphics board. From
http://www.infinetivity.com/~delscott/mem.htm
"If you're wondering why one of the stacks in the picture is taller than the rest, those were a different brand of ICs, with a slightly thicker body."

Reminds me of a public housing project for some reason.

[whartung]

BigEd wrote:

Timo Noko stacked 35 DRAMs to provide tagged lisp node storage in an Osborne luggable.
http://anycpu.org/forum/viewtopic.php?t=235&p=1456

Nice. Seems like he embedded his own linear accelerator in there as well. Maybe he salvaged it from one of the Ghostbusters back packs.

Those photos are eerily creepy. Like photos of insects.

[Oneironaut]

Those are some fine examples of RAM Stacks indeed!

Here is my 256K Stack, made with 15ns 32K SRAMs and a 3-8 decoder.
Great for dropping on a breadboard when you are out of 512K DIPs....



Brad

[BigEd]

These are neither stacked nor DIP, but I thought it fit in here. This is the 8-way stacked QIP-bearing "module" found in a 1985 Cray-2: 300 modules, 750 chips on each, several hundred watts from each module, immersed in 200 gallons of pumped inert coolant. AFAICT, most of the memory chips (75000 of them) would be 256kbit. The logic chips are 16-gate programmable. Four heavy duty CPUs running at 240MHz. More detail here.

[whartung]

I'm going to have to say that I'm happy that I don't have to carry around 200 gallons of inert coolant to keep my iPhone from melting in to my leg every time I get a text.

[cbmeeks]

What would be awesome is if someone made a complete SBC (without the board) using nothing but stacked IC's.

And no micro-contollers or PLD's.

[Alarm Siren]

Well, the problem you'd have there is the different package sizes and pin layouts. Stacked RAMs are relatively easy since only one pin needs to be seperately routed, but stacking a RAM, ROM, I/O, 6502 and Glue Logic would be a Frankenstein's Monster, methinks, not to mention a PITA to build.

[cbmeeks]

Alarm Siren wrote:

Well, the problem you'd have there is the different package sizes and pin layouts. Stacked RAMs are relatively easy since only one pin needs to be seperately routed, but stacking a RAM, ROM, I/O, 6502 and Glue Logic would be a Frankenstein's Monster, methinks, not to mention a PITA to build.

Exactly. Which is why it would be awesome.

[GARTHWILSON]

That's quite a collection!

I seem to remember a small board from decades ago where memory was stacked on top of a 65xx IC (whether µP or I/O I can't remember).

Brad, I think I remember your stack which is probably in your picture, but your link to it just gives an error 404 message.

Here's a stack of four 74AC14's I used for driving a voltage multiplier to supply several watts:

74AC14x4.jpg (33.07 KiB) Viewed 22512 times

Here's a dual MAX153 A/D converter:

DoubleMAX153.jpg (28.04 KiB) Viewed 22512 times

This was for a workbench computer, but I have not implemented it yet.  The MAX153 is an 8-bit half-flash converter that's pretty quick.  If you run a 6502 at only 1MHz, you can put it on the bus and read it like a memory location.  When you select it, the conversion begins, and data is ready before the end of the phase-2 cycle; IOW, there's no need to start a conversion and come back later when it's done.  You can do the whole thing in one cycle.  I have a singe MAX153 on my main workbench computer, although there I go through a VIA since the clock is too fast to put the A/D right on the bus.  [Edit, 7/16/22:  I was just shopping for more parts, and found that the MAX153 has gotten super expensive, like $24US!  The somewhat slower ADC0820 is one-quarter to one-third the price.  The MAX114 seems to be the same thing inside as the MAX153 but is about half the price and yet has four input channels.  The MAX118 is the same thing for eight channels instead of four.  IOW, my two stacked converters above are now unnecessary.  I'm finding that all the fast A/D converters are expensive now though.  My purpose for the '153 is speed, and getting exact timing on the readings; so if you don't need the speed but you want more resolution, you you could use SPI or I²C A/D converters, perhaps putting them in SPI-10 or I2C-6 or 65SIB devices you plug into those ports on your computer.]

My stacked 512Kx8 SRAMs in Ed's pictures above were things that eventually led up to the 4Mx8 10ns 5V SRAM module I offer, with parts on both sides:

ModuleStraightPins.jpg (40.16 KiB) Viewed 22512 times

I have three different connector options.

Stacking made good use of space; but besides being kind of messy, it was too labor-intensive to do a lot of, like to supply modules to others for their projects.  Putting parts on both sides can be done with thru-hole parts too, if you stagger the rows of IC pins.  You can either use socket strips on the first side so the other side can be soldered before putting the IC in the first side, or you have to be able to solder the ICs from the top of the pins instead of under the board.

[cbmeeks]

As Dave would say, all of the electrons are going to fall out!

[Alarm Siren]

cbmeeks wrote:

Alarm Siren wrote:

Well, the problem you'd have there is the different package sizes and pin layouts. Stacked RAMs are relatively easy since only one pin needs to be seperately routed, but stacking a RAM, ROM, I/O, 6502 and Glue Logic would be a Frankenstein's Monster, methinks, not to mention a PITA to build.

Exactly. Which is why it would be awesome.

Challenge Accepted.

[cbmeeks]

Alarm Siren wrote:

Challenge Accepted.

Now, I would say PLCC-44 only for MPU and I/O but I think the Geneva Convention prohibits that kind of torcher.