hp 9815a/s repair



#2

I have corresponded 2 years ago concerning "hp 9815a repair". Last week we obtained an hp 9815s calculator via ebay. It is in better shape than our 9815a, but the 9815s does have some memory issues. Rather than displaying ~3800 programmable steps, it displays 904. Also the storage registers A->E and 003->009 work perfectly, but F->J, and 000, 001 and 002 already are filled with the number 2.02 x 10^33 when the unit is turned on. All other functions seem to work. Obviously some of the memory chips (hp 1818-0438) are faulty, but some chips are still active. Does anyone know which memory chips are responsible for which functions? Which memory chip holds storage registers A->J, which holds registers 000->009, and which hold the remaining 3800 programmable steps? We have Tony Duell's schematics for the hp 9815a (Australian HP Museum), but are not sure that these schematics cover the 9815s. Does anyone know which of the 8 memory chips does what?

And where do we find replacement 1818-0438 memory chips?

We also are seeking the Utility and Test tape cartridge for the 9815a/s, or a copy, or a print-out of the Utility and Test programs, so that we can make a test tape of our own. Apparently every 9815a/s came with a Utility and Test tape, so at least one 9815 owner must still have a Utility and Test tape cartridge that works - hopefully!!??!!

Thank you, most very kindly.

Frank Simpson franksimpson4@gmail.com


#3

I have a 9815a expanded to 9815s spec. The only difference between the two machines is surely the interface card and the extra ram card sitting middle top. So to reduce your machine to a 9815a, remove the extra ram card and then you've got a match for Tony's schematic. I would expect the base register location to be the same between both macine types anyway as the s just has more extended ram.

Don't know if that helps...

Mark

#4

Copying the Utility and Test tape is something that I have been unable to do after many attempts, years ago. There's some sort of special copy protection that it makes use of.


#5

Thank you, Katie,
Can the Utility and Test tape program be printed out via list, or similar?


#6

Quote:
Can the Utility and Test tape program be printed out via list, or similar?

I'm afraid not. Most of the programs are hidden and are in machine language. They display and print patterns that are not possible to generate via user programming. I suppose that it wouldn't be impossible to hack the tape read/write code given that the 9815 used a 6800 microprocessor, but there must be an easier way to copy this tape.

Just curious, has anyone here ever figured out how to run their own 6800 code on the 9815A/S? I used to write in 6502 machine language for my PET computer way back when I didn't have an assembler and had most the op codes memorized. It might be run to try that (a few bytes of it) on a 9815.


#7

Although it is a long while now, I used to be completely fluent in the 6800 (I detested the 6502!). Sadly, I know I have lost all my original opcode docs as I've been trying to find them for over 20 years but there must be docs on the net. If anyone can get a binary dump of these tapes, I wouldn't mind seeing what I can do with it given time. The results could be very interesting as they would show entry points and how to load and run your own 6800 code from tape.

One of the things that always picqued my interest in the 9815 was that it used a 6800 - it would be fun delving into this more even if we are a long way off doing this as of now.

Mark


#8

Hi Mark,

The starting point would be to get a binary dump of the ROMs, then disassembling the code shouldn't be too difficult since we have the partial listing in the patent. Any clues about how to read the binary without desoldering the ROM chips?

Joel Setton


#9

Hi Joel,

No, wouldn't know how to do that.Back in those days, you tended to get branded as a software or hardware person and I was always software. I'm out of my depth here but all I can think of is a method of loading 6800 code via tape can be established, then a simple program to dump the entire 64K memory map. But seriously, I'm guessing a lot here.

Are the patent docs freely available on-line?

Mark


#10

Hi Mark,

The patents are available from Eric Smith's website. And they are public domain by now, so they are "highly-recommended" reading. I have found the one on the 9815A to be a fascinating description of the machine.

Joel Setton

#11

The HP 9815's RAMs are most likely Intel 2111a parts. They're not the eight white chips that I think you're referring to. Those white ceramic parts are ROMs and contain the calculator's firmware. There are actually six RAMs on the motherboard. Each RAM is a 256x4-bit chip. They are black, have 18 pins, and most likely have a big, lower-case "i" for Intel stamped on them, although they may well have an HP part number instead of 2111a. I think they're up by themselves at the top of the motherboard between the tape drive and printer cutouts. It takes two of these chips to make one 8-bit byte and there are three banks on the motherboard. Sorry, I can't tell you which one is bad from your description. I did a quick search and you can still find these on eBay, but the chip is way obsolete. Some just sold on eBay: http://cgi.ebay.com.my/ws/eBayISAPI.dll?ViewItem&item=260454127351 for way cheap. You can also use AMD 9111a parts, I think.


#12

Hi Steve,

I recall I worked with Intel 2101A, 256x4 static RAM chips (=one kilobit); are those compatible with the 2111A you mentioned?

Regards


#13

Intel 2101A parts are earlier devices. They have 22 pins instead of 18 and require multiple power supplies where the 2111A parts only require 5V. The HP 9815A was based on the Motorola 6800 microprocessor, which was the first commercial microprocessor to run on just 5V, hence the desire to use 5V RAMs as well: to simplify the HP 9815A's power supply.


#14

The 2101 is a 5V-only part. The reason it has more pins is that it has separate data inputs and outputs.


#15

My mistake. Thanks for clarifying.

#16

Since the 9815S is newer and has more RAM, I'd expect it to use newer, higher-density RAM chips than the 9815A did. Possibly the 2114. But I don't have a 9815S handy to check.


#17

The additional RAM was added with a daughter card, I believe. Someone stated that in the archives earlier. Otherwise, there would have been two different motherboard designs, which there weren't. In addition, there would have to have been 2-Kbit SRAMs, which there weren't at the time. We designed those machines with the biggest commercially viable chips available at design time, and the IC generations were decidedly further apart in time than they are now. The HP 9825A started with 4-Kbit DRAMs and then graduated to 16-Kbit devices two to three years later in the HP 9825T. Note to all of you born in the 1980s or later: I'm sure this discussion of Kbit DRAMs is hilarious with the multi-Gbit SDRAM devices we have now, but the first commercially successful DRAM (from Intel) only appeared in 1971. It held all of 1 Kbit. Few people know that it was a bear to bring to production and it was only through the help of the friendly engineers of HP Loveland, who desperately needed the chips to make the HP 98x0 series of desktop calculators real that the chips made it to market in 1971.


#18

The 9815S was introduced four years later than the 9815A. The 2114 4-kbit SRAM (1Kx4) was certainly readily commercially available before the 9815S shipped.

I was of the impression that the 9815S used a different PWB than the 9815A, but I'll admit that it's been quite a while since I looked at the innards of either model, so it's quite possible that I'm mistaken about that.

Even if the 9815S used the same main PWB as the 9815A, it's still possible that it used a newer memory expansion board with denser RAM chips than the original optional memory expansion for the 9815A. In fact, I'd be extremely surprised if it *wasn't* a new board, because the 9815S had as standard *twice* as much memory as the 9815A could handle as an option. This means either a different memory board was used, or a different main board that could accept two memory boards.


#19

Folks,

The two CPU boards are different: The one in the 9815A uses the 2111 RAM chips (with a capacity of 256 x 4) and seven 2kx8 ROM chips; the board in the 9815S uses 2114 memory chips (each of which with a capacity of 1k x 4) and two 8Kx8 ROM chips. As seen in the 9815S I have, the CPU is not an upgraded 9815A, it is a redesigned board. The 9815S gets its larger memory from the "large-capacity" 2114s chips and not from an additional RAM expansion board.

I also believe that at some point in time, HP stopped making the old 9815A CPU board and used the newer 9815S boards to service defective 9815As. In the repair process, these calculators became "hybrids", with a 9815S CPU and its larger memory, but without the two explansion slots which are standard in the 9815S.

Joel Setton

#20

Steve, thank you for your comments. Yes, it is fun when we talk about bits and kilobits. I use to comment those facts to my engineering students at the university; for instance that all HP35 code fit in 768 x 10 bit words... some of them expect no less than megabytes even for the most simple task!

I still have (in working condition) my homebrew RCA 1802 board, with 256 bytes made from 2 5101 SRAM... I once "upgraded" it by adding a 2 kbyte CMOS SRAM, all battery-backed.

#21

Here are some AMD 9111As for $5 each on eBay: http://cgi.ebay.com/Vintage-INTEL-P2101A-4-256X4-RAM-2101A-22-PIN-DIP_W0QQitemZ320252853049QQcmdZViewItemQQptZLH_DefaultDomain_0?hash=item4a908ebb39. I think they'll work.

#22

http://www.jameco.com/webapp/wcs/stores/servlet/ProductDisplay?jameco_page=46&langId=-1&productId=38931&drill_parents=category_root%24%2410%24%241050&drill_children=10%24%241050%24%24105090&catalogId=10001&storeId=10001&search_type=all&drill_displays=ICs+%26+Semiconductors+%2F+Memory+%2F+SRAM&ddkey=http:StoreCatalogDrillDownView

#23

Hi Frank,

The assignment of memory location addresses to the various storage registers in the 9815A (i.e. X/Y/Z/T in the stack, user program storage, etc...) is very well documented in the 9815A patent (US 4,089,059) which you can read from Eric Smith's website. This patent is a "must-read" for all those interested in the 9815A.

This should help you find the addresses of the defective RAM locations, which will point to a pair of RAM chips since RAM locations are 8-bit wide and each chips stores 4 bits.

I hope this helps, but please let me know if it doesn't!

Joel Setton


#24

Thank you, thank you, for your many wonderful, helpful responses!!!

Since there seems to be some debate on the differences between the "a" and the "s" models, here is the scoop on the boards and memory chips for our hp 9815a and our hp 9815s, opened, side-by-side.

9815a - with both the 001 option (extended memory to 2008 programmable steps) and the 002 option (HP-1B interface).

Numbers on CPU Board:
88809F
REV B
09815-66510
4699
RAM chips on CPU: SIX 1818-0140 (these cross reference to Synertek SYP2112A-4).
RAM chips on 001 extended memory board: TWELVE 1818-0140.
All RAM chips are 16-pin with 8 pins per side.
Numbers on 001 option board:
88809L
REV A
09815-66581
525
OS
ROM chips on CPU: SEVEN, white, 28-pin chips with 14 pins per side, with an open slot
on the board for an 8th chip.
NOTE: The hp 1818-0140 RAM chips cross reference to 2112, NOT 2111.

9815s
Numbers on CPU Board:
88809F
REV C
09815-66514
3936
RAM chips on CPU: EIGHT 1818-0438 (these cross reference to TMS2114-45NL).
No provision on CPU board for attachment of 001 memory option board.
These RAM chips are 18-pin with 9 pins per side.
ROM chips on CPU: SEVEN, white, 28-pin chips with 14 pins per side, with an open slot
on the board for an 8th chip.

Clarification of confusions on the forum:
1) The RAM chips on the 9815a cross reference to 2112, and the RAM chips on the
9815s cross reference to 2114. Although everyone is talking about 2111 chips, 2111's
are not to be seen on either of our calculators.
2) Tony Duell mentions that the 9815a uses 7 ROM chips, but the 9815s uses a pair of
ROM chips, http://www.hpmuseum.org/cgi-sys/cgiwrap/hpmuseum/archv014.cgi?read=66265.
Both our 9815a and our 9815s have 7 ROM chips. Perhaps our 9815s is a transition model.
3) Our 9815s DEFINITELY has a redesigned CPU board compared to our 9815a (compare
the serial numbers listed above). Our "s" version has no provision for
attachment of the "a" version's extended memory board. In the region of the
memory chips, the circuitry of the boards, and the layout of the chips, are quite
different. Plus the 9815s has eight 18-pin RAM chips on its CPU, while the 9815a
has six 16-pin RAM chips on its CPU.

I'll let you know if/when we get either of these running at 100%, and how we solved it.

Frank

Edited: 26 Sept 2009, 3:53 p.m. after one or more responses were posted


#25

Thanks for posting all the details. If it's not too much trouble, could you check the part numbers of the ROMs in both units? If either one has part numbers other than 1818-2602 through 1818-2608, that would be very interesting to know.

You should be able to find some 2114 chips, but they've got a reputation for being unreliable, so pulls or even NOS parts aren't necessarily going to be good. If all else fails, you could kludge up something with newer RAMs such as the 6116. That was a 2K*8 CMOS static RAM originally introduced by Hitachi around 1980, but equivalent parts are still made to this day (unlike the 2114).

Thanks!
Eric


#26

Hello Eric,

The ROMs on my 9815 are one 1818-1056A and one 1818-1057A. Each of these chips is 8K bytes, so that's a total of 16K bytes. However, the top 2K are disabled by the missing jumper which enables/disables the optional character set.

The CPU board part-number is 09815-69512 which is technically the same as a 09815-60512. The 69xxx prefix (as opposed to 60xxx) means the board was sold by HP as a service part.

Joel Setton


#27

Dear Eric and Joel, Saturday, 9/26/2009

Here are the numbers of the white ROM IC's for our 9815a and our 9815s. As for the position of the chips, I am viewing the keyboard section inverted (i.e. upside down), and rotated so the white ROM chips are at the top. I then read left-to-right for the uppermost bank of four ROM chips, followed by, left-to-right, the lower bank of one empty slot, followed by three ROM chips.

hp 9815a
Upper bank:
G1818, 2605, 2693
H1818, 2602, 1363
D1818, 2603, 2696
B1818, 2604, 2801
Lower bank:
empty slot
B1818, 2608, 2668
D1818, 2607, 2783
E1818, 2606, 2455

hp 9815s
Upper bank:
H1818, 2605, 5640, SINGAPORE
H1818, 2602, 2581, SINGAPORE
D1818, 2603, 5454
B1818, 2604, 5174, SINGAPORE
Lower bank:
empty slot
B1818, 2608, 5633
D1818, 2607, 6149, SINGAPORE
E1818, 2606, 5915

General question: Why has 'most EVERYONE spoken of 2111 memory chips on the 9815a, whilst our 9815a has 2112's (hp 1818-0140)? Is our 9815a unique, or (heaven forbid!) are some recollections foggy?

Thank you, Eric, for the information on possible replacements for the unreliable 2114's, and thank you, Joel, for the patent reference for memory assignments for the RAM chips. Maybe we can order some replacement RAM's soon.

Endless details: Also Joel, when you refer to the two ROMs of your 9815, I assume that yours is an "s" version? So it is interesting that our 9815s has 7 white ROM chips, like our 9815a, while your 9815s has two ROM chips. Our "s" CPU board is REV C, 09815-66514, while you notice that your board is the service part 09815-69512.

Frank

Edited: 26 Sept 2009, 3:41 p.m.


#28

My incorrect reference to the 2111a was caused by the photo of the 9815 motherboard on this site. It shows 18-pin chips for RAMs. That led me to incorrectly state that the chips on your board were 2111a parts. Clearly, not correct. From now on, I'll leave it to the experts.

#29

The 2112 (and AMD Am9112) are similar to the 2111 (and AMD Am9111), but have only a single chip select and no output disable pin. It's likely that they either used exactly the same die, or a die with only a metalization change. The 2112 was probably slightly less expensive due to the smaller package.

With regard to my previous suggestion of the 6116 as a possible replacement, you should be aware that it is not even *close* to being pin compatible. The 2114 is in an 18-pin DIP (0.3" row spacing), while the 6116 is most commonly found in a 24-pin DIP (0.6" row spacing). Were I in need of replacements for a lot of 2114 chips, I'd probably lay out a small PCB with a surface-mount SRAM like an ISSI IS61C256AL-12TLI (28-TSOP package) on one side, and two 9-pin surface-mount male headers on the other.

Hmmm... I wonder if enough people need 2114 replacements that I could sell some? I need a bunch myself.

On the other hand, Unicorn Electronics does have several 2114 variants in their catalog


#30

Jameco (www.jameco.com), a mail-order company in the SF Bay area, sells 2114s for about $2 each. Just search on 2114.

#31

Hi Frank, Eric and all,

The ROM part-numbers in Frank's message show that his 9815A and 9815S have the same set of ROMs, so I think we can reverse-engineer the history of 9815x CPU boards:

- The 09815-66510, first generation used the seven-chip ROM set, 1818-2602 to 1818-2608. An additional 2K ROM chip can be added for optional characters. The RAM chips were six 2111s, and the CPU board has a connector for the RAM expansion board. RAM size is 768 bytes.

- The second-generation 09815-66514, uses the same 7-chip ROM set, but increases RAM size with eight 2114 chips to a capacity of 4K bytes. Since this is the maximum addressing capacity permitted by the firmware architecture, additional RAM cannot be used and the connector for the memory expansion accessory is deleted. This board is used to produce the first 9815S.

- The third generation is the 09815-60512, with keeps the same 4K of RAM in 2114 chips, but the seven+1 ROMs are replaced with two 8Kx8 chips, one 1818-1056A and one 1818-1057A.

The backward-compatibility of the newer boards, which can be used in all generation of 9815x calculators, resulted in HP using the newer boards to repair failed 9815As. This is what happened to mine, which was manufactured in 1977 as a 9815A, but was later fitted with a new board. Since the calculator had Option 001 (two expansion slots), this repair essentially turned it into a 9815S.

It would be interesting to know if the switch from the seven-ROM set (1818-2602 to 1818-2608) the two-ROM set (1818-1056 and 1818-1057) included any firmware changes, bug fixes etc... Is anyone aware of such changes?

Joel Setton


#32

Maybe my one contribution to this whole exchange is to question, and hopefully put to rest, the notion that 2111 RAM chips were used in the 9815a. My 9815a here in my hands has 2112's - NOT 2111's. Apparently mine is not a one-off prototype. I am noticing that almost every mention in the forum of RAM in the 9815a, even by the expert Tony Duell back to circa 2004, says 2111. However, when I have looked up the 2111 at Jameco during these last few days, the 2111 has 18 pins. The 9815a here in my hands has 16 pins on each RAM chip. When I cross-reference the hp numbers of the 6 RAM chips on mine's CPU Board, and the 12 RAM chips of its 001 extended memory option board, every one of which is 16 pin, and every one of which is 1818-0140, I obtain SYP2112A-4. I have obtained this cross reference SYP2112A-4 at TWO different sites on the internet, by typing "hp 1818-0140" into google, as the girl on the telephone at hp did, and as she advised I should do, also.

Further, it was thoughtfully pointed out by Steve in Message #26 of this forum page, that there is a photograph of the entrails of an hp 9815a on the hp museum's internet site. I've gone to the hp museum's photograph. It really is of a 9815a, as its CPU board has the proper gold connectors for the 001 memory option. I unfortunately cannot decipher the code numbers on the chips of this photo; however by squinting and recounting several times, I find 16 pins on each RAM chip, 8 per side, not 18 pins/9 per side. I presume therefore that the RAM chips of the hp museum's 9815a are likewise 2112, not 2111.

#33

Hello Joel, One more bit of curiosity. If your calculator really was initially a 9815a that later acquired an "s" board during servicing, I am wondering what its stickers read under the printer roll lid, at the back. Our 9815a has two stickers in this location that read, "option 001", and "option 002". While our 9815s has a single sticker here that reads, "9815 S". Might yours have all three stickers, perhaps?

I'm also noticing REV letters on the CPU boards. Our 9815a CPU board is REV B. Our 9815s CPU board is REV C. Both of these boards have seven ROM chips. Your CPU has dual ROM chips. Though it necessitates removing a few screws, etc., to open your case, so you may not want to risk a look, I also am wondering what REV letter your CPU board has. This letter might verify your reverse engineered theory, somewhat, by confirming that our REV C is an earlier "s" version, while your board is later.

Frank


#34

http://store.americanmicrosemiconductor.com/tms2114-45nl.html

http://www.americanmicrosemi.com/information/spec/?ss_pn=TMS2114-45NL

I have just ordered all 14 of these RAM chips. The number TMS2114-45NL cross-references to the original hp 1818-0438 NMOS 450 ns RAM chip of the hp 9815s. I am not sure who the manufacturer is. Hermes tells me that American Microsemiconductor can order more because their supplier has 8000 in stock, though he cannot predict the cost of their next batch. Hermes can be contacted at hermes@americanmicrosemi.com, phone# 1-973-377-9566 or 1-877-372-6652.

One giant leap for human kind!

#35

Hi Frank,

I had a look at the CPU board, it's an 09815-69512, A-2244 which means it's the first version of the 09815-6x512 copper (revision A) and the last engineering change to that board was made in the 44th week of 1982. Of course this is not the date of production, the most recent chips on the board have 1984 datecodes. The board has 4K memory in eight 2114 chips, and two 8Kx8 ROM chips.

The revision letters were incremented when a board went through a change which did not affect "fit, form and function", in other words when its feature set and compatibility with the outside world are unaffected by the change. When the feature set changes, for example when replacing the 2112 SRAMs with 2114s and removing the memory expansion connector, the new board is not an exact replacement for the old one, so it gets a new part-number and its revision letter is reset to A.

Comparing the revision letters of two different boards is probably not going to tell us a lot, because production of different types of boards may have overlapped during some time.

On the outside, my 9815 started life as an "A" and included the optional I/O backplane. It has just one sticker, "Option 002". Then it probably failed and got repaired with a new 09815-69512 CPU board, with 2114 RAMs and two 8Kx8 chips. This repair upgraded the machine to an exact 9815S.

Does this help?

One last word: when it comes to HP calculators, curiosity is not very far from passion!

Joel


Possibly Related Threads...
Thread Author Replies Views Last Post
  9815A/S option ROM Joel Setton (France) 0 429 03-08-2011, 12:43 PM
Last Post: Joel Setton (France)
  9815A manual Joel Setton (France) 0 409 05-21-2010, 04:25 PM
Last Post: Joel Setton (France)
  hp 9815a/s program tapes frankabc 3 770 10-28-2009, 09:46 AM
Last Post: Joel Setton (France)
  hp 9815a repair frankabc 6 1,070 10-16-2007, 12:27 AM
Last Post: frankabc
  HP 9815A repair frankabc 7 1,201 08-25-2007, 02:26 AM
Last Post: frankabc
  HEWLETT PACKARD 9815A Sale !!! gileno 2 646 02-22-2007, 05:43 PM
Last Post: Steve Borowsky
  9815A up for auction Les Wright 3 757 01-15-2007, 02:49 PM
Last Post: Steve Borowsky
  Synthetic Programming on the HP-9815A/S John Ioannidis 9 1,194 12-04-2005, 02:00 AM
Last Post: John Ioannidis
  HP 9815A or 9815S? Ken 5 841 12-02-2004, 09:56 PM
Last Post: Ken Delano
  HP 9815A Fabrizio 1 492 04-01-2003, 07:00 PM
Last Post: Tony Duell (UK)

Forum Jump: