I have several retro machines and add-on cards for the Amiga that use ZIP RAM. This vertically mounted chip design enjoyed a brief popularity in the early 1990s, in between the original DIP DRAM and the rise of the SIMM memory.

The advantage with ZIP over DIP was obvious - more RAM chips in the same board area. Of course, you needed different sockets.

SIMM offered the same space advantage, but with a more complicated socket and with the RAM chips having to be soldered onto a PCB. Overall, seems like a more costly way to go than ZIP RAM, but SIMM obviously won in the market.

So why were the added manufacturing costs for SIMMs not an issue and why did they quickly replace ZIPs?

Note: I get that SIMM insertion and removal is relatively easy, but that ease-of-upgrade option only makes sense if it was a major consumer selling point. My recollection is that most motherboard RAM upgrades happened at the point-of-sale, with vendors likely choosing the cheapest solution.

Right off the top, SIMM type modules have higher pin densities and are mechanically far more robust than that ZIP chip package. It'd be relatively easy to bend pins getting these things in an out (possibly worse than DIPs even), and protecting the pins in transport looks even harder than doing so with DIP chips.

Remember that many retailers are unlikely to admit to having broken the pin off a chip - they'll return it as defective instead.

Further, each ZIP is one chip, no more, no less. With a SIMM module being a PCB with chips mounted on it, you can build modules in various ways to get the desired density and to use the most cost effective parts available at the time.

In the end, as busses widenened, the advantages of being able to use multiple chips probably were the final determinant.

  • 6
    Zip is far worse than DIP IMHO, I have had some that had broken pins, some that had bent up pins, and if something doesn't work, it is extremely hard to see where the problem is. And, they are a nuisance to plug in correctly. – tofro Oct 5 at 16:25

Separate memory modules on a standardized interface provide several advantages:

  • Reduced overall production cost
  • Greater flexibility in RAM selection
  • expansion way beyond what a classic (pre-SIMM) motherboard would support.
  • Reduced motherboard design cost
  • Greatly reduced motherboard production cost
  • Faster innovation cycles due to defined and spacious interface definition
  • Move of RAM choice and investment to point of sales (or later)
  • More option and reduced investment for retailers.
  • Modules will be usable across different brands and boards.

And maybe most important

  • Greatly reduced cost of memory cards (in addition to reduced mainboard cost)

The advantage with ZIP over DIP was obvious - more RAM chips in the same board area. Of course, you needed different sockets.

That's why it was added, and especially used in graphics cards, where space was always a premium - at least when trying to provide upper end features.

SIMM offered the same space advantage, but with a more complicated socket and with the RAM chips having to be soldered onto a PCB. Overall, seems like a more costly way to go than ZIP RAM, but SIMM obviously won in the market.

Not really, as it reduced the amount of parts. Instead of having a socket for each and every ram chip a single socket was needed holding 8 (or sometimes up to 32) chips. Quite a reduction from the point of a motherboard manufacturer. The memory modules itself could carry the chips without any socket. Further it allowed to separate RAM mounting technology (thruhole or SMD) from motherboard manufacturing. A SIMM Module could use any kind of RAM, no matter if it is DIP, SIP, ZIP or SMD like TQFP or BGA.

It further decoupled RAM pinout from motherboard layout. Now larger or smaller RAMs from various manufacturers could be incorporated. This include pinouts that were not known/available at the point the motherboard was designed.

And most important, it removed the need to design a motherboard for a specific amount of memory and, as follow up, the design of specific memory cards. A motherboard can only hold sockets for a specific amount of RAM of a specific pinout. When they are filled, some kind of memory or I/O bus card must be made to add more memory.

Using memory modules is, well, modularization. Much like having a nice, clear API, designed with an upgrade path in mind allows independent development and sales.

So why were the added manufacturing costs for SIMMs not an issue and why did they quickly replace ZIPs?

Because manufacturing costs were not higher but lower. Lower for all parties involved, Motherboard and RAM module manufacturers as well as retailers.

  • On the motherboard side, buying one (complex) socket is less expensive then adding 32 (or more) single chip sockets.
  • Similar, placing one (or even 4) SIMM sockets is faster and less expensive than 32 separate sockets.
  • For the RAM manufacturer the board design will be way less expensive than for a traditional (large) memory expansion board (like ISA Memory cards for the PC).
  • Similar board production will cost way less.
  • For the retailer taking the RAM (socket/space) cost out of the motherboard reduces upfront investment on motherboards, as well as on RAM modules

Note: I get that SIMM insertion and removal is relatively easy, but that ease-of-upgrade option only makes sense if it was a major consumer selling point. My recollection is that most motherboard RAM upgrades happened at the point-of-sale, with vendors likely choosing the cheapest solution.

The profit of vendors/resellers configuring a PC is quite dependent on the time invested to configure a PC to the customers specs, and inserting 8, 16 or more flimsy ZIP chips without bending a pin and good insertion and so on, is a great burden compared with clipping in one or two SIMMs. Time is money.

Stock cost is as well important for a vendor/reseller. Modularized structure does enable a considerable lower stock by only holding a few different memory modules instead of many different chips and even more model specific memory expansion boards.

  • Another format my first 80386 machine used was, IIRC, SIP. A circuit board which plugged into IIRC a 30-pin inline socket and was electrically equivalent to a SIMM. I think SIP sockets were a tiny bit cheaper than SIMM sockets, but the cost difference would have been eaten up by the extra assembly cost of putting the pins on the board. – supercat Oct 5 at 22:42
  • @supercat Jup, that SIP (BTW, there are also SIP chips) 'socket' for memory boards was added out of the same considerations - they just not went for improving a card edge format (as SIMM is) but used the more reliable (and proven) pin and socket mechanics. Another argument I heared for them (back then) was that these modules could also be soldered in for a factory installed memory, saving the sockets in total. SIMMs did beat them already in the short run by their ove all lower price. – Raffzahn Oct 6 at 10:46

The ZIP package was mostly replaced by better packages. Ignoring for a moment the variety of memory modules (DIMM etc) most memories moved to SOP, SOIC, TSOP or BGA for higher pin density, improved mechanical stability and lower package height.

None of the new packages suffer from the bending torsion that made ZIP packages prone to damage. Further they do not require the package to maintain the complex bending of pins that ZIP requires.

All of these result in lower manufacturing costs and fewer system failures.

Could you imagine trying to do a Raspberry PI with ZIP memory? Me neither.

  • 7
    I think this answers what really happened: SMD packages. – pipe Oct 5 at 22:47
  • The Question is about SIMM modules, not is a single boarder with soldered on RAM like the RasPI, so there's not relation to what was asked - and more important, SIMM do predate the use of SMD RAMs in PCs - 30 pin, as well as 72 pin ones where fitted with various kinds of DIP RAM in the begining: commons.wikimedia.org/wiki/File:Edoram.jpg – Raffzahn Oct 6 at 14:15
  • 1
    @Raffzahn Actually the question is about ZIP packages. The image in your comment shows surface mounted SOJ packages on a memory module. No DIP packages anywhere. Here's a link: digchip.com/datasheets/parts/datasheet/517/VG2617405EJ-6.php – Peter Camilleri Oct 6 at 14:57
  • @PeterCamilleri You may want to reread, as the question is about why ZIP has been replaced by SIMM. Next, SOJ ust still a DIP format with inbound bend pins - but more important, just pointing that out doesn't change the primary point, that SIMM was introduced with thruhole components, not SMD - check the very next picture from wiki ( en.wikipedia.org/wiki/SIMM#/media/… ) - saves me posting a picture of several shoeboxes of DIP SIMM in my workshop :)) – Raffzahn Oct 6 at 15:04
  • @Raffzahn The first SIMMs were apparently developed for flip-chip and SMT (J-lead) technology. While through-hole was made, I believe that the intent was always allow DRAM to go to SMT. – user71659 Oct 6 at 18:12

Your Answer

 

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Not the answer you're looking for? Browse other questions tagged or ask your own question.