vantreeck@logic.DEC (10/25/85)
>Newsgroups: net.micro.mac >Path: decwrl!decvax!bellcore!petrus!sabre!zeta!epsilon!gamma!ulysses!mhuxr!mhuxt!houxm!hou2d!xxajtxx >Subject: Cheap RAM upgrade >Posted: 23 Oct 85 22:19:16 GMT >Organization: AT&T Bell Labs, Holmdel NJ >1. The motherboard should have sockets into which the high-density chips can > be inserted. There are a several reasons why high volume manufactures of PCs do not ship socketed boards: 1) It allows the user to order an unpopulated version of the computer in order to stick in cheaper RAMs, thereby depriving the manufacturer of some profit. 2) It raises the cost of board -- sockets aren't for free. 3) Increases the width of the board, possibly increasing the volume (expense) of the computer cabinet. This is mostly a problem when memory is in backplane slots - which the Mac doesn't have. 4) HIGH volume manufacturing techniques don't seem to be able to produce socketed memory boards that are as reliable as unsocketed memory boards, lower yields of good boards and lower MTBF mean higher cost to you -- at least that was the case a few years ago. Perhaps there are better techniques and sockets in use today. 5) RAM chips are easily damaged by static, and most users don't know how to properly ground themselves, the memory board, and the RAM chips while installing the chips. 6) Only a very small segment of the business world wants to install there own chips. Installing one's own chips is mostly the concern of people who want to own a computer but have difficulty affording it. It's like someone on a $20,000 a year income, owning a new Porche or Jaguar, and wanting inexpensive doit-yourself maintenance kits. Such people aren't in the mainstream of the customer base, and aren't likely to see the manufacturer design the product with them in mind. -George
timp@gcc-milo.ARPA (Tim Peacock) (10/26/85)
In reference to the reliability of sockets: I seem to recall the rule of thumb being: The addition of a socketed chip, in place of a soldered chip has the same effect upon reliability as adding another chip. I believe this comes from an ATT manual on reliability. Thus, if Apple (for example) was to add socketed RAMs, it would be equivalent, in terms of reliability, to adding 16 more chips to the digital board. All in all, I'll take the soldered version. Tim Peacock
callen@ada-uts.UUCP (10/28/85)
> 5) RAM chips are easily damaged by static, and most users don't know > how to properly ground themselves, the memory board, and the RAM > chips while installing the chips. I personally think this static business gets blown out of proportion. I've handled A LOT of 256k DRAMs without blowing ONE, and I don't take any elaborate precautions - I just don't do anything blatantly stupid (like work in a polyester shirt under a wool sweater). > 6) Only a very small segment of the business world wants to install > there own chips. Installing one's own chips is mostly the concern > of people who want to own a computer but have difficulty affording > it. It's like someone on a $20,000 a year income, owning a new > Porche or Jaguar, and wanting inexpensive doit-yourself maintenance > kits. Such people aren't in the mainstream of the customer base, > and aren't likely to see the manufacturer design the product with > them in mind. I don't know who YOU work for, but MY employer DOES get empty boards and populate them, and therefore save bo-ku bucks that can be better used to, say, give me a raise (:-). LOTS of business types are wising up to the big memory scam. >-George - Jerry ...ihnp4!inmet!ada-uts!callen
hogan@rosevax.UUCP (Andy Hogan) (11/07/85)
>> 5) RAM chips are easily damaged by static, and most users don't know >> how to properly ground themselves, the memory board, and the RAM >> chips while installing the chips. > >I personally think this static business gets blown out of proportion. >I've handled A LOT of 256k DRAMs without blowing ONE, and I don't take >any elaborate precautions - I just don't do anything blatantly stupid >(like work in a polyester shirt under a wool sweater). There is a widespread but completely mistaken understanding of ESD (static) damage to integrated circuits which says it is an all-or-nothing proposition. You can cause significant damage to an IC by static without "blowing" it completely. This damage takes the form of altered dynamic or static electrical performance (ie, input current, output current, i/o impedance, speed, etc.) and fairly often does not IMMEDIATELY affect the operation of the circuit. However, such damage has been shown to significantly shorten the useful lifetime of the IC. So it is impossible to tell, immediately after installing an IC in a circuit, whether it has been damaged by ESD or not. Also, the word "easily" in the first post is a relative term. 256K RAMs are indeed "easily" damaged with respect to certain other technologies, such as small-scale TTL chips. TTL is not destroyed by static until the applied charge comes from a source of a few kilovolts, whereas a CMOS RAM can be utterly destroyed by a source of less than one KV. Damaged-but-not-destroyed levels show roughly the same magnitude difference. Any system which prevents the buildup of a static field lower than damage levels is sufficient. The choice of prevention measures is an economic one. For one person, with one machine, careful handling, humidification and wearing cotton clothes may be enough. For more than one, you have to make a judgement. We tend to install our own (cheap) chips in PCs here because we have workstations that are available for such occasional use and that are fully protected, but this equipment is moderately expensive, and I doubt we would buy much of it just for our PCs. However, good grounding wrist straps can be had for $15--20 and that cost is easily justified. -- Andy Hogan Rosemount, Inc. Mpls MN path: ...ihnp4!stolaf!umn-cs!mmm!rosevax!hogan Working is not a synonym for Quality.
hogan@rosevax.UUCP (Andy Hogan) (11/07/85)
>In reference to the reliability of sockets: >I seem to recall the rule of thumb being: >The addition of a socketed chip, in place of a soldered chip has the >same effect upon reliability as adding another chip. I believe this comes >from an ATT manual on reliability. > >Thus, if Apple (for example) was to add socketed RAMs, it would be >equivalent, in terms of reliability, to adding 16 more chips to the >digital board. All in all, I'll take the soldered version. > > Tim Peacock Hmm. This seems like an old rule, but is probably still good if you substitute "another small- to medium-scale integrated TTL chip" for "another chip". Newer technologies and larger scales of integration have greatly decreased the reliability of chips vs. sockets. Socket manufacturers have not stood still either (they know that they get bad press on reliability) so they have increased socket reliability. Socket reliability is only one factor. When production quantities are in the region that Apple works in, the cost of the sockets (and installing them, etc.) becomes far greater than the cost savings in repair and other cost advantages of socketing. So there is enormous pressure on a large manufacturer not to use sockets. From Apple's point of view, they are a burden. From MY point of view, they were necessary when I did the Dr. Dobbs upgrade, because they allow me to replace dead chips easily and (potentially) allow me to add in 1 Meg chips when they are available (with some more hacking and new Apple ROMs which I optimistically assume I can get (1/2 :-) ). I also got very good sockets at an enormously low cost.... In any case, sockets for ICs (particularly memory and other expensive ones) are something that should be designed in until a total cost analysis pushes them out. When that happens, the manufacturer should be aware that he is trading cost against problems that are very tough to measure, such as customer satisfaction, and weigh that in the decision also. -- Andy Hogan Rosemount, Inc. Mpls MN path: ...ihnp4!stolaf!umn-cs!mmm!rosevax!hogan Working is not a synonym for Quality.