Yeah I think the 8200 manual is the mystery document. I fed it into Gemini and it says it's happy to fully convert the entire document to markdown. Says it did it and then the file doesn't show up in the files list. Lies. In the interactive section it's only partial. More lies.
After a lot of verbal abuse it eventually admitted that the document is too big for it to convert all at once. But what it does convert it does a pretty good job. So you may want to play with giving it small page ranges of the PDF and see what it can do. Just don't believe it did what it says until you compare line by line until you're satisfied because it's a naughty liar. Now the other tech ref has several pages that have missing/ garbled portions like on page - 99 -, - 117 - - 119 - probably a bunch of others but I haven't reviewed every page. And of course the ocr is bad throughout... I really need this document in markdown too. -- John. On Fri, Oct 9, 2026 at 5:28 PM B 9 <[email protected]> wrote: > It will be exciting to have a clean copy, but now I'm also realizing that > you are talking about the Technical *Manual* > <https://www.web8201.net/Files/LIBRARY_web8201/NEC8201A-TechRef.pdf>, not > the Technical > <https://www.1000bit.it/support/manuali/manuali.asp?id=115>*Notes > <https://www.1000bit.it/support/manuali/manuali.asp?id=115>.* The > document for the PC-8201 that I find the most intriguing (and least > legible) is the Notes, which has this table of contents: > [image: contents.png] > What's so interesting about the Technical Notes? Well, for one it actually > defines the Floating Point notation, which I needed to document the > PC-8201's .BA file format. (Unlike all other Model T's, the computers from > NEC stored numbers in BASIC programs as binary instead of ASCII). Another > thing is that the Technical Notes is the mysterious "another manual" > referred to by the Technical Manual when it says, > > NOTE: The another technical manual for PC-8201A has >> been available already. There are many information >> about the RAM file handling routines in ROM #0 in it. >> For example, •oPEN RAM FILES", 'KILL ASCII FILE", >> 'READ A CHARACTER FROM A RAM FILE' and 'CLOSE ALL >> FILES'. If you will use your application or >> subroutine with ROM #0, you had better refer to that >> manual. > > > > The *Technical Notes* also has other interesting historical curiosities. > For example there's a page that says, > > File: [MasahiroN.Docs.PPC]PPCInfo.Doc > Mathematical Routines in PC-8201 ROM > Draft 1.0 > May 27, 1983 > > The filename makes me wonder if the internal product name for the PC-8201 > was "Portable PC". Also, was there ever a non-draft version released? > Perhaps in Japanese? > > Another thing it has been useful for: I had been having trouble with the > calculations in my NEC SuperROM spreadsheet tests going awry. Gary told me > that NEC's wacky math was likely due to backwards compatibility. I think he > was right as, in the Technical Notes, Masahiro specified that the PC-8201's > floating point notation was to be "EXACTLY" the same as on the PC-8000 from > 1979. What I don't get yet is that I don't see anything inherently wrong > with NEC's format. It can theoretically represent numbers with 16 times > more precision than your standard 64-bit double, with a tradeoff of having > thousands of times less range. (See notes at end.) So, now I'm curious if > the weird mathematical curiosities are due to speed optimizations specific > to the PC-8201 or if the PC-8000 and PC-8800 also do strange things like > this: > > DEFDBL X : X=10^2 : ?X > 99.99999237060547 > > Have you ever seen 10² calculated with such precision before? > > —b9 > > P.S. Notes to make my hand waving about floats more real: > > A floating point value equals ±(1+f)·2ⁿ, where f is fractional (or > mantissa) part and n is the exponent. > > IEEE-754 8-byte float > > 63 62 52 51 0 > |-------------------------------------------------. > |s| 11 bits |msb 52 bit mantissa lsb| > `-------------------------------------------------' > | | `---------------- mantissa > | `-------------------------------- biased exponent (3FFh) > `------------------------------------- sign bit > > NEC PC-8000 double precision float > > 63 62 56 55 0 > |-------------------------------------------------. > |s| 7 bits |msb 56 bit mantissa lsb| > `-------------------------------------------------' > | | `---------------- mantissa > | `---------------------------------- biased exponent (80h) > `------------------------------------- sign bit > > > > On Fri, Oct 9, 2026 at 10:37 AM Kurt McCullum <[email protected]> wrote: > >> No, it's all white original paper. Lots of fold outs in there. Looks >> like I have a second copy of those sheets as well so I should be able to >> get you some good scans. >> >> Kurt >> >> On Fri, Oct 9, 2026, at 12:33 PM, B 9 wrote: >> >> No rush, Kurt. >> >> Is the Green NEC book the one with about 200 pages of which the last half >> appears to have been printed on tractor-feed greenbar paper >> <https://www.pdp8online.com/images/greenbar.shtml>? >> >> I've been trying to figure out the puzzle pieces from different poor >> quality scans like they're the dead sea scrolls for years. Just last week I >> had been talking with Gary about the mathematical routines in the ROMs >> which lead me back to the second half of the PC-8200 Technical Notes. Even >> the parts which aren't missing are very hard to read. I tried to merge the >> scans into a more legible PDF to upload to archive.org, but gave up. I >> finally decided it would be less of a pain to just retype the whole thing. >> I was seriously considering doing that and was reading up on Leiden >> bracket notation >> <https://papyri.info/docs/leiden_plus#leiden-conventions> so I could >> properly emend the lacunae <https://arxiv.org/html/2407.00250v1>. A good >> quality scan will save me so much time and effort! >> >> —b9 >> >> On Fri, Oct 9, 2026 at 8:48 AM Kurt McCullum <[email protected]> wrote: >> >> >> I can do that. The only problem will be with the pages that unfold to >> show circuit diagrams but I will figure that out. Might take a couple days >> though. This weekend is rather busy for me. >> >> Kurt >> >> On Fri, Oct 9, 2026, at 10:23 AM, B 9 wrote: >> >> Kurt, if you could scan the original, it would be great! I've been trying >> to merge and decipher the two (and a half) versions of the tech ref I've >> found online and it is nigh impossible. The quality is just too poor. >> >> —b9 >> >> On Fri, Oct 9, 2026 at 6:30 AM Kurt McCullum <[email protected]> wrote: >> >> >> I have two version of the Tech ref. One is a photocopy and the other an >> original hard copy in the green NEC folder. >> >> I can check those for any content you might need. >> >> Kurt >> On Fri, Oct 9, 2026, at 7:24 AM, Gary Weber wrote: >> >> Hi John, >> I may have a different PDF available. What's the missing content? Are >> you missing whole pages or diagrams? >> >> Have a look at https://www.web8201.net/tech.asp and see if the second >> document you're referencing is present. >> >> On Thu, Oct 8, 2026 at 8:58 PM John R. Hogerhuis <[email protected]> >> wrote: >> >> Is there only one PDF version of the 8201A tech ref? It has some missing >> content. >> >> Also the document refers to another tech ref that covers the file system >> in even more detail, don't know if anyone has found that. >> >> I ripped a version of it into an .md file so my local AI rig can utilize >> it. Unfortunately converting it to .md (marksown) leaves a lot to be >> desired... the ASCII art is mangled and the embedded OCR'd text is riddled >> with errors. >> >> If anyone has got a better copy or one with better OCR that would be >> helpful. Successfully converted to .md would be even better.' >> >> Next I'll be looking for or creating similar documents .md Inside the >> Model 100 and Hidden Secrets, the user manual and Lien's book to feed into >> the AI spaghetti flinger. >> >> -- John. >> >>
