Talk:RCA 1802

Timeline Conflict?

The article states that it was "introduced early in 1976". However, it also says it was used on the Viking Mars landers. These were launched in 1975 and probably needed to be based on plans that couldn't change much into the manufacturing stage. Thus, such a chip would probably have had to exist around 1973. And, I doubt NASA would base a billion-dollar probe on a prototype. Perhaps it was the 1801 chipset/model? Or perhaps it was designed around the 1801 chipset but switched to 1802 late in the project because the 1802 is compatible with the 1801 set. --Tablizer (talk) 05:11, 15 July 2008 (UTC)

I'm fixing it to 1974, based on http://jbayko.sasktelwebsite.net/cpu2.html#Sec2Part1 and Microprocessor Report, August 5, 1996, Volume 10, Number 10. --IMneme (talk) 01:17, 19 December 2010 (UTC)

Note that it was NOT used in the Viking. This has been corrected with citations.--WillBo (talk) 08:15, 17 October 2016 (UTC)

Origins

old RCA semi manual has cd4057 cmos logic chips that implemented a 4-bit register/alu slice. cosmac elf kit dates back to 1975. Key to space is SOS or silicon-on-saphire, which reduces latch-up on radiation exposure. Shjacks45 (talk) 11:03, 23 August 2008 (UTC)

https://datasheetspdf.com/mobile-datasheet/CD4057A.html TaylorLeem (talk) 19:17, 23 May 2021 (UTC)

OK, so what is the question to discuss? I have a slight memory that there was a SOS CDP1802, but only slight. If there is, should it be discussed here? I do suspect that the CD4057 is too far off for this article. Gah4 (talk) 00:59, 25 May 2021 (UTC)

CD4057 was bit slice part of 1801 ALU design . Explains 1802 design. Silicon on Saphire is a Wikipedia article and should be linked. 2600:100F:A020:F420:E8D3:194F:55B:4A2D (talk) 04:34, 2 July 2024 (UTC)

The CD4057 was not part of the CDP1801. The CDP1801 was two fully-custom chip layouts, which were not made up of other chips. Brouhaha (talk) 08:19, 28 August 2024 (UTC)

CPU architecture infobox?

Would be a nice addition, since the 1802 instruction set was mentioned in the Popular Electronic article as having been designed by a single person.. —Preceding unsigned comment added by Wikip rhyre (talk • contribs) 12:43, 11 March 2010 (UTC)

1802 in Voyager craft

According to the official Voyager2 Twitter feed, the processor used in the Voyager craft are actually based on the 1802 203.56.250.55 (talk) 23:34, 19 May 2010 (UTC)

If so, NASA has forgotten their own history, and is relying on unreliable web sources such as the earlier versions of the Wikipedia 1802 and Voyager program pages. The actual NASA technical reports on the Voyager describe the computer as a slightly updated version of the Viking orbiter computer, and the NASA technical reports on Viking describe the processor as something MUCH different than an 1802. The 1802 wasn't even announced, much less shipping, when the Viking was designed. The references to these NASA reports are on the Viking program and Voyager program pages. --Brouhaha (talk) 20:33, 13 September 2010 (UTC)

This has been correctly referenced and mentioned on the page, with citations, as well as having been corrected across the Internet where possible by several 1802 enthusiasts, although there are places where it cannot be corrected. (old articles, etc.) --WillBo (talk) —Preceding undated comment added 20:13, 12 January 2017 (UTC)

I believe that the Integer cosine transform (ICT was designed around the arithmetic capabilities of the 1802, for Galileo when the big antenna didn't open up. Gah4 (talk) 01:10, 25 May 2021 (UTC)

References

Complementary Silicon/Metal-oxide Semiconductor?

Complementary Silicon/Metal-oxide Semiconductor would be abbreviated COSMOS, not COSMAC. That's how it's bolded in the article, too. Does the acronym really stand for something else? 69.54.60.34 (talk) 17:57, 13 September 2010 (UTC)

COSMOS was RCA's acronym for their general CMOS process, not specifically the 1802. COSMAC was the acronym for the 1802 family. --Brouhaha (talk) 17:08, 2 November 2010 (UTC)

Undefined Instruction

I recall that in a computer I built around the 1802 in 1977 that the undefined instruction hex code 68 was essentially "lethal" to code it RAM. It wrote 68 hex over the next instruction. When the next instruction was executed the process repeated filling RAM with 68 Hex. This would be worth mentioning if someone can confirm my recollection. — Preceding unsigned comment added by 173.212.187.148 (talk) 18:30, 26 August 2015 (UTC)

The 68 hex instruction is "undefined" for the 1802, and used for extended instruction set instructions for 1804, 1805, & 1806. I have seen discussions about what happens when the instruction is executed in the 1802, which I think is as mentioned above, but it is unknown if that is universal for all (several) source company microprocessor chips. It is not "lethal", however. --WillBo (talk) 20:04, 12 January 2017 (UTC)

Package Markings

None of the chip markings/suffixes have any citations documenting. Citation Needed markings applied. If you have documentation, please cite. Encyclopedic content must be verifiable per Wikipedia standards. — Preceding unsigned comment added by 169.252.4.21 (talk) 08:27, 14 October 2016 (UTC)

Much of the information in this section was gathered from MANY source documents, so citing is problematic. I am in the process of creating a web page about this information on my website ( http://www.MDCCCII.com/ ), and will end up using it as the citation. It will have citations for its information. That should be good enough. I did find a decent citation for the second request, so I added that. I'm about as close to an expert on the 1802 as anyone can be, and I assure you I only add information to this page that I KNOW is correct. MOST of the citations on the page (and associated information) were added by me. I don't know when I will get my 'summary' page completed. But hopefully not too far in the future. It's a lot of work to do it well and right. (just compiling the information in this section was a lot of work) I only added this section because someone added some scant similar information in the wrong place, so I deleted it and added this more expansive information, knowing that I didn't have a citation for it yet. — Preceding unsigned comment added by WillBo (talk • contribs) 07:54, 17 October 2016 (UTC)

Citations

Someone went through and added "citation needed" in many of the sections, and multiple paragraphs within sections. Most of these can point to the same document (particularly those in the "Technical Description" section), the "User Manual for the CDP1802 COSMAC Microprocessor" (MPM-201x), an online copy of which can be found here:

https://archive.org/details/bitsavers_rca1802MPMualNov77_10819092

So that is the citation I intend to use. I don't know if I should add this citation to every paragraph, or only the first paragraph of the section, or only the last paragraph of the section.

Note that there is a Reference #39, "User Manual for the CDP1802 COSMAC Microprocessor", but that is all it has, no link to the actual document. It is used once in the article. I didn't add that, and I don't know if it is valid or not to do it that way.

— Preceding unsigned comment added by WillBo (talk • contribs) 08:07, 17 October 2016 (UTC)

It is important to note that your contributions to the article sometimes constitute a liberal concoction of facts found in the manual. That makes it harder to assert that the source actually supports the statement, and easier to accidentally write things that are not supported by the source.

I just found an statement without a reference that directly contradicts an explicit statement in the manual and replaced it with something that reflects the content of the manual. Not the whole section, just one sentence. For all I know, the article may be full of such mistakes, and those are much easier to find when each individual statement can easily be tracked back to a published document, because now I don't know whether this is from some other document that contradicts the manual or just a misunderstanding of the manual. That is a good argument for either providing citations where they have been marked as needed or just ignoring them until someone else does. 84.219.234.209 (talk) 20:01, 24 October 2018 (UTC)

Better Die Shot

Here is a better die shot than the one used in this page. I'm not sure how the copyright works here, is it possible to use this? Illiteration (talk) 15:19, 12 January 2017 (UTC)

Afaik, it has to be either in the public domain or the copyright owner has to give permission. --WillBo (talk) 19:57, 12 January 2017 (UTC)

Another application

The 1802 was used in underwater equipment deployed in the lagoon at Kwajalein (the downrange end of the Pacific Missile Test Range). The equipment detected the acoustic signature of a warhead strike on the ocean surface, recorded an accurate timestamp, and later reported its results by acoustic telemetry. We probably cannot document this because the source material is likely classified if it still exists at all (this was back in the 80s). But it was an interesting application. N.B. - I have never held a security clearance of any type. — Preceding unsigned comment added by 50.43.56.108 (talk) 02:12, 18 June 2017 (UTC)

where I saw chip X

After a few instances, a list of things that used chip X is no longer illuminating for an article about chip X. It woudl be much more worth while to say *why* X was used, what the alternatives were at the time, and show the particular advantages chip X had in that application. --Wtshymanski (talk) 17:51, 16 September 2017 (UTC)

I disagree. Historical information is ALWAYS valid and of value. It gives the reader a clear insight into how extensive the use of the CPU was, and documents and archives that important information. ALSO, some of the changes you made where you removed valuable information was tantamount to vandalism. I did A LOT of work on this article and spent MANY hours gathering and verifying this information. I don't appreciate you 'savaging' that work. MANY of the changes you made were COMPLETELY arbitrary and unnecessary and not only did NOT make the article better, but weakened and lessened it. It really deserves an UNDO. --WillBo (talk) 18:24, 16 September 2017 (UTC)

As I said above, a list of primary references to uses of X is no more than a clipping service. Some analysis of *why* this processor was chosen would be of far more value to an encyclopedia article. --Wtshymanski (talk) 19:03, 16 September 2017 (UTC)

You simply do not know what you are talking about. Your shoddy "editing" of the article information and deletion and removal of valuable information suggests that you should be banned from editing this article before you do even more damage than you have already done. --WillBo (talk) 19:38, 18 September 2017 (UTC)

This is deletionist nonsense. Suppose the benefits of having this information here are insignificant. What, then, are the benefits of removing it? What are the negative side effects of having information most people don't care about? Is it really that hard to scroll past something? Sure, you could find some far-fetched reason why unimportant information could be inconvenient, but the reasons why the lack thereof would be inconvenient far outweigh that, considering the incredibly small inconvenience it causes.

I personally find it interesting, so I'd rather it be kept, and the article's already pretty long so it's not a massive inconvenience. If you think that the article needs an explanation of why the chip was used in those applications, you can do the research and add it yourself, or at least try to convince others that it's important to do so, but removing a bunch of interesting historical information doesn't seem useful to me.

It seems that most of your edits are removals, so the issue is clearly bigger than this article. You ought to really consider who's adding more to Wikipedia; you, rapidly removing information and making everything more brief, or somebody focusing more on adding things. A lot of your revisions (going back a long time, I think) seem to remove actual information for the sake of brevity, claiming things are irrelevant or redundant when they clearly aren't, or that it's just so inconvenient to scroll past something for 1 second. I know that it's hard to go back on a general stance you've had for years, and not all removals are bad, but you should try to be less enthusiastic about deleting others' work. Illiteration (talk) 17:12, 22 September 2017 (UTC)

Wikipedia has infinite space and budget, but human readers looking at an article will die in only 70 or 80 years or so - an article that wastes human lifespan by being too long should be edited. You'll have to clarify what you mean by "most of my edits" - do you have a time span in mind, or over the whole 13 years? I do delete a lot of vandalism. I've no idea what the red numbers and green numbers sum to over the years. It might even be near zero? In the last 500 edits I count about 322 green edits, adding text. I don't know how zero net byte change edits rank me on your scale of evil. But that is not "mostly" deletions. --Wtshymanski (talk) 18:50, 22 September 2017 (UTC)

Thanks for pointing out that mistake. I exaggerated how much deletion you do, and I probably should have provided examples, but I'm lazy and that's probably better suited for a discussion on your talk page. To address your point about limited life spans, people don't have to read sections they're not interested in. Of course, it's a good idea to make articles easy for people to skim through, but the section you deleted did nothing to stop people from reading only what they wanted, so why remove it? --Illiteration (talk) 19:54, 22 September 2017 (UTC)

Article Is Under Attack

Some idiot(s) (there is no other more appropriate description for them) went through and SAVAGED this article and completely ruined it. They took out so much important information I can only assume they are near-moronic, or are purposely attacking the article to ruin it, because only someone like that would do what has been done. They should be prevented from changing this article in the future, and it should be restored to the great article it once was. Someone in charge from Wikipedia needs to deal with this issue. WillBo (talk) 04:10, 10 November 2017 (UTC)

I assume that you mean this edit: https://en.wikipedia.org/w/index.php?title=RCA_1802&oldid=800942447

Please assume good intentions here. The list that was removed may fall within the definition of a "laundry list". There is a WikiProject to remove these from articles: https://en.wikipedia.org/wiki/Wikipedia:WikiProject_Laundromat. That article makes a good case for why they may be considered undesirable. The user that removed the list probably intended for the change to increase the quality of the article. If you disagree, it's better to make a good case for why you think they should be there and revert the change than resort to name calling.

The same is true for the citation requests that you removed. You removed them with good intentions, but the user that added them (and me personally) think that they made sense. We have conflicting opinions but good intentions all the same. 84.219.234.209 (talk) 20:21, 24 October 2018 (UTC)

Link in Intro is dead

Does anyone know if the statement "It is currently being manufactured by Intersil Corporation as a high-reliability microprocessor." is still correct? — Preceding unsigned comment added by Emulate-atg (talk • contribs) 06:08, 31 October 2018 (UTC)

Speed grades and their introduction etc.

Right now all this is pretty vague in the article... there's mention at some random point that the chip "was available" with speeds "from 3.2 to 6.4MHz", and much further down it suggests that "later" versions of the chip could run at "4 to 5 MHz" at 5V, and only "upto 6.4MHz" at a pretty edgy (for CMOS) 10V, which if I didn't know better I'd cry foul over and call a blatant bit of unofficial overclocking. There's no mention of what speed it introed at, what the actual interstitial ratings are, whether that redline 6.4MHz is even official, or when any of the higher ratings became available (if indeed they weren't available from day one). I suspect, for one thing, the actual COSMAC computers ran at <1.8MHz - instead of more like 3.6MHz - because, like very nearly every other CPU of the period, they simply couldn't withstand that high frequency, and might not have been good for any better than maybe 2.4MHz to start with.

(About the only thing that could run up to 5MHz in the 70's was the later Intel chips and maybe the initial releases of the MC68000... everything else, including Intel and Motorola's lesser chips, was languishing around 4, 2.5, 2.0, or even just 1MHz; people dig at the IBM PC for running at a paltry 4.77MHz, but that was actually pushing the envelope somewhat for its time of release... if the 1802 could manage even 3.2MHz it would have been nearer the upper end of the market rather than the lower, but 2.4 wouldn't have been particularly embarrassing)

Obviously, this is totally unacceptable for a CPU page in general, and definitely not for an early and very long-lived one with many industrial and scientific uses and microcontroller-like derivatives (even if it didn't become popular in the SoHo realm). There will be many perfectly good, precise, and more than likely clearly dated spec sheets and other references for it out there which contain this and probably a lot of other missing details that I haven't personally picked up on yet. If we've been able to raid them for things like the pinout and its register layout, how come we haven't got the full skinny on the device, more than 40 years after it hit the market, as a hobbyist's darling? 146.199.60.36 (talk) 22:54, 30 August 2019 (UTC)

Ugh. It gets worse. One of the main cited pages in the actual spec section was just a bunch of chip photos on CPU-Collection, and the two datasheets appended in "external links" were completely mislabelled, as well as at odds with each other. The "original" sheet (which bears a very clear "1997" datestamp) was anything but, being for the "A", "AC" and "BC" derivatives, the latter two of which (origin date still unknown) seem to be the source of the "4 to 5MHz" and "6.4MHz" claims respectively; the AC can apparently hit 5MHz in general use, which isn't really surprising for a late-90s chip, even a low power embedded one... what's weirder is the BC needing not just 10V, but 10.5V to reach max speed, though that might well be its maximum voltage tolerance rather than an outright requirement, because it's still not exactly an extreme speed even for an embedded chip from the era (Palmpilots were already coming with 8MHz 68000-derivatives running at 3V - though I suppose the difference may lie in the radiation and other environmental hardening?). The straight "A" caps out at just 3.2MHz, supporting my initial hypothesis, and this is borne out, bizarrely, by the eleven years later (but now 11 years old) AC/3 spec sheet (until now, listed as "current") somewhat schizophrenically claiming all three of a "3.2MHz", "3.6MHz", and "4.5us cycle" (= 3.556MHz with 16-clock machine cycles... that is, a little higher than 2x 1.76064, but certainly just a little lower than 3.579MHz) maximum operating speed, at a vanilla 5V. It doesn't help that these come from the "cosmacelf" website with filenames that suggest the original link labels were telling the truth. What a mess.

(Update: Antique Chip Collector page, as rough as it looks and as patchy as its data is, suggests the original speed grades were 2.5MHz and maybe 5MHz (or, that came in later?), with 6.4MHz versions not appearing until 1981, quite a bit after the time of the Elf and VIP; I guess we can assume 3.2MHz became established sometime in-between, though there were likely still lower and intermediate binnings. Quite where "4" came from, though?) 146.199.60.36 (talk) 23:18, 30 August 2019 (UTC)

My (printed, copyright 1977) MPG-180A COSMAC Microprocessor Product Guide, on Page 8, states quite clearly that the CDP1802D was rated at from 4 to 12V for supply voltage, and that the maximum clock rate was 3.2MHz at 5V, 4MHz if Vdd was 10V, and 6.4MHz if both Vcc and Vdd were 10V. One assumes that at 12V it could go even a bit faster. The lesser CDP1802CD package was rated from 4 to 6V Vcc/Vdd, and the same 3.2MHz clock rating at 5V. The CD was rated at 10mW power consumption, the D at 100mW. (Presumably a function of the product of voltage and frequency, it seems likely the D would consume the same power as the CD at the same operating conditions.) No other packages/ratings are mentioned, just these two. But they cover, and right from the beginning, all the speed ratings that seem to be talked about.

Of course, getting everything else in the computer to also run at 10-12V was nearly impossible, especially for hobbyists, so a 4MHz 1802 system was about the best you could hope for. (Common memories of the time also required +12V, which could also have provided Vdd for the processor, enabling that slightly higher clock speed.) While RCA also claimed RAM and ROM devices in the CDP18XX family, rated at the same voltages as the CPU, procuring them might not have been easy. 1802 systems didn't require ROM, and high-voltage 4000-series CMOS was there for jellybean logic, so the real 10V problems were: RAM, and the (common) keyboard scanner. System design is non-trivial. ISC PB (talk) 21:36, 31 March 2020 (UTC)

Recent edits

As part of a cleanup effort for various 8-bit CPUs, I re-wrote the lede of this article. I found it in a rather confused state - repetitive, containing trivia like part numbers, lacking the fact that it was the first CMOS CPU, and far too long - according to the MOS the lede should generally be four paragraphs except for very long articles, which this isn't. However, my work was RVed, with a comment that it removed hard work, thereby introducing most of the problems again. Unless someone can offer specific points as to what was wrong with my version, I would be inclined to re-instate the edits. @WillBo: Maury Markowitz (talk) 00:32, 27 January 2020 (UTC)

I STRONGLY disagree. It is NOT too long, it is NOT repetitive, and it reads fairly well, or better than that. In NO WAY could anyone who knows how to write complain that it was in a "rather confused state". And "part numbers" are NOT "trivia", ESPECIALLY when talking about integrated circuits. As I stated, you basically ravaged the page section that was there and removed important information. I am a professional writer and I know how to write well. There are no problems with the section as it was and is again. PLEASE do not make the page WORSE. --WillBo (talk) 06:53, 27 January 2020 (UTC)

cycles

There are comparisons of clock cycles to other contemporary (or not) processors. The comparisons aren't very useful without clock speed values. One can often trade off cycles for speed in a design. (More recently, this often comes up comparing Intel and AMD processors.) As well as I remember, though 1970's CMOS was pretty slow, though that depended on supply voltage. Gah4 (talk) 18:09, 27 August 2020 (UTC)

Ten volt 1802 was faster. 2600:100F:A020:F420:E8D3:194F:55B:4A2D (talk) 04:49, 2 July 2024 (UTC)

In the 1802, all instructions take either two or three "machine cycles", but each machine cycle takes eight clock cycles, so that instructions take either 16 or 24 clock cycles. (The added instructions in the 1804/5/6 can take more than three machine cycles.) The 1802 machine cycles and clock cycles are directly analagous to the 8080/8085/Z80 machine states and T-states. In general, the 1802 takes more clock pulses than almost any other microprocessor to achieve generally similar results. This is because the 1802 internally is a bit-serial processor. The clock frequency supported by the CDP1802 was higher than that of some contemporary processors (especially at 10V), but that wasn't enough to overcome the 8 clocks per machine cycle problem. --Brouhaha (talk) 08:35, 28 August 2024 (UTC)

VHDL cores

The Emulators and Simulators section has long referenced a VHDL core purportedly written by Scott Baker, with a link to a nearly empty github repo that doesn't actually contain such a VHDL core. I don't know whether Baker has actually written one, but I don't think it should be cited unless a more suitable link can be provided.

I am not myself making such a change to the article; my doing so might be construed as a conflict of interest as I am the author of another VHDL core that is referenced. I would request that another editor please look into it, and if warranted, edit the section accordingly. --Brouhaha (talk) 07:16, 3 June 2021 (UTC)

Requesting again that the link to Scott Baker's purported VHDL core, not present at that link, be removed. I'm not removing it myself, but request than another editor check it, and either remove the link from the article, or reply here as to why they don't think removing that link is warranted. --Brouhaha (talk) 08:23, 28 August 2024 (UTC)

Code Samples

While I get the love for the processor, and its historical significance, I'm minded to suggest that this section either goes or is totally rewritten.

1. There's only one of them - it's a code sample. More importantly...

2. It's largely meaningless to anyone who's not already familiar with the chip. I thought I was 'knowledgeable enough to read "pseudo-code" and is minimally familiar with assembly and machine language programming' having used, for 8-bit CPUs, 6502, 8080, Z-80 and 6809, but clearly that's not enough.

"GHI 0 .. SET UP R6" - what does the GHI instruction do, and why does having it be zero set up R6?

"SEX 0 .. (X=0 ALREADY)" - if it's zero already, what is gained by setting it to be zero again?

etc. Replace with a link to the reference manuals? Lovingboth (talk) 15:18, 30 November 2021 (UTC)

1802 and 1804/5/6 differences

Some crucial differences here particularly lack of DMA bootloader (manual DMA for switches or paper tape loading of programs) on 1804 series. The value of the additional instructions should be noted, especially program branching and interrupt handling. The 1804 was used in embedded systems (like Fluke meters et al) so didn't get the visibility that desktop processors got. Note the only systems mentioned were hobbyist toys. 2600:100F:A020:F420:E8D3:194F:55B:4A2D (talk) 04:47, 2 July 2024 (UTC)

Note also that the 1804A/1805A/1806A contain more instructions than their 1804/1805/1806 (non-A) counterparts. The "A" in this case refers to a fairly significant mask change to add instructions. --Brouhaha (talk) 08:26, 28 August 2024 (UTC)