Be warned that I have (alas) nevr seen one of these units, so what i am going to say is going to be very general. But it might help Nice find! I asusme this is likely to be a ROM chip (i.e. same number of pins as the other ROMs on this PCB). My guess (being optimistic) is that the board was made ofre sensible number of ROM chips (8, 16, etc) but fewer were actually neeeded to store the firmware (c.f early HP9815 CPU boards which have space for 8 ROM chips but only 7 are ever fitted). If you are genuinley missing a ROM, you are going to have _big_ problems. What sort of RAMs are used? Some older static RAM chips are noted for unreliability, and can cause 'interesting' faults. Do you know if all those LEDs are controlled by the processor? Can you trace them back to an output poret. The point being that the fact they are doing something would seem to imply the processor was working, and that it was executing some code. The CLOCK LED is curious to me. I would have thought, if that is controlled by the processsor (rather htan some hardware power-fail circuit) then it would be turned on after the 'core' (ROM, RAM, CPU, etc) self-tests. which impleis those might well have passed. Oh for a schematic.. Argh!. Such manuals annoy me. 'I am the service engineer damnit. Just tell me what the real fault is' :-) What sort of floppy drive does it use? Is there a standard interface (like 50 pin SA800 thingy)? Does it make any attempt to, for example, restore the heads to cylinder 0? Not a bad idea, but the fact it turns the fault light off and on again would seem to indicate the CPU was running. I assume you don;t have a logic analyser. If you do, I'd trigger it on the FAULT LED turning on, and trace the code backwards from there. -tony

Bob R. wrote: it Yes, the unpopulated socket is a 24-pin, near the edge of the ROM board. It's a relief to know that the fault being reported is not due to a missing chip. actually Wow! Ten? With as huge and heavy as these things are, I can imagine that you must have a lot of storage space! That'd be most appreciated! I don't want to put pressure on Al, but it'd be great if the 4907 Service Manual could be on the priority list to be scanned and put on Bitsavers. Schematics and other service information would make the job much easier. Tony D. asked what type of RAM is used. It uses 18 MK4127 (Mostek) 4Kx1 Dynamic RAM chips with 150ns access time. They are all socketed. I was able to find the datasheet for the parts online. They need to be refreshed a maximum of every 2ms. I figure that if worse comes to worse, I could simply build a little circuit with the address latches (the address lines are multiplexed) to grab the address, then use fast Static RAM to replace the DRAM. Wouldn't be too tough to build, and would get rid of the reliance on DRAM. I'm willing to bet that the problem is probably with the DRAM. It would make sense that the firmware would do a quick RAM test at initialization after perhaps performing a ROM checksum test, and if either test fails, lighting the FAULT light. Tony also asked how the fault light is driven. The four LEDs are connected to the main controller board with a ribbon cable, that plugs into the controller board with a harmonica connector. It looks like the LEDs are driven by a latch that is probably driven by the 6800. So, like Tony posited, the microprocessor is probably running, and is actually detecting some fault condition that prevents it from getting through its power-on self tests. It would have been nice if the LEDs could somehow have been used to indicate the part of self test routines that was failing -- it'd at least be a good clue for troubleshooting. On another topic related to the 4051...many of the 1/4" (Tektronix-branded) cartridge tapes that came with the machine have busted drive bands. The tapes are OK, but they won't work in the 4051's drive because of the broken drive band. I've re-banded one of them ("Version 5 Loads-Of-Fun Game Tape (Binary Version") and it works good now, after I did a little o'scope-assisted tweaking of the tape head azimuth to get the signal strength a little stronger. The games tape has a lot of really neat games that appear to be derivations of various common BASIC games, but modified for use on the 4051 with its graphics capabilities. There's a fun little artillery game when you take turns trying to launch artillery shells at your opponent (or the computer). There's also a game called "Weather Wars" where you try to destroy your enemy with lightning strikes and tornados. There is a BASIC version of Chess (not graphical) that appears to be a re-write in BASIC of a minicomputer chess program (the board listing looks a lot like the Chess for the PDP/8 under OS-8). It's painfully slow to play...computer takes about 6 minutes to make its move. It'd play faster (probably about 3 times faster) on a 4052 that uses the bit slice implementation of the 6800). There's a music box program that uses 6800 machine language code (the CALL EXEC function) to manipulate the bit that drives the speaker to play music. Anyone know of a source for the drive bands in these 1/4" cartridges? I've got a stock of spare DC300 and DC600 tapes that have good drive bands, but I don't want to pillage them for drive bands to try to repair the 4051 cartridges that have broken bands. Rick Bensene The Old Calculator Museum http://oldcalculatormuseum.com

On 15 May 2009 at 13:27, Al Kossow wrote: Google for "Plastibands". Use the 4" size. Your local office supply place may stock them. They work like a charm. The smaller 2" size works for DC2000-size carts and would probably also work for the DC1000 mini-carts. Cheers, Chuck

On Fri, May 15, 2009 at 11:27 PM, Mike Loewen <mloewen at cpumagic.scol.pa.us> wrote: Nice board. I have no information and no docs on this board, but I know a bit about the 68000. The daughter card should be easy enough to untangle - it's just an address comparator, a couple of buffers and the off-board connectors. If I had that rig, the first thing I'd do is to draw a schematic of the daughter card at least, to understand how data gets off the ZPU. -ethan

On Fri, 2009-05-15 at 23:27 -0400, Mike Loewen wrote: Well, the daughterboard looks like it takes an EPROM and maps it into memory somewhere. There are a couple of octal buffers and a comparator on there, so presumably you set where it falls in the address space with the DIP switch. Gordon

On Fri, 15 May 2009, Mike Loewen wrote: After some more digging, I found a couple of references to this board. HSC is apparently Hallock Systems, as explained in this note: "As a side note, I once had a Hallock Systems CO-1668 68000 based co-processor on my old Xerox 820-1. It had 2 Meg of memory and was able to run CP/M-68K. It also supported a RAM drive under CP/M-80. The CO-1668 basically used the Z80 CP/M system as a file and I/O server. All of the CP/M-68K system calls were sent to the Z80 system and then converted to CP/M-80 calls and then the operation was completed. Kind of slow but it worked." Still looking for documentation... Mike Loewen mloewen at cpumagic.scol.pa.us Old Technology http://sturgeon.css.psu.edu/~mloewen/Oldtech/

On Mon, 18 May 2009, Frank McConnell wrote: Thanks for the info, Frank. Someone else who has one is sending me the manuals to scan, which I will post on my site (and send to Al) for download. Mike Loewen mloewen at cpumagic.scol.pa.us Old Technology http://sturgeon.css.psu.edu/~mloewen/Oldtech/

Some things developed today on this front. The UPS man came with another load of stuff from the donor of the 4051 and 4907. This batch of packages included a bunch of documentation, including -- an original copy of the 4907 Service Manual! So, the documentation problem is solved. So, I spent some time reading through it. It says that the most common cause of the FAULT light being on is a RAM failure. Unfortunately, to diagnose it, a special fixture is required, which is essentially a front panel for the system, including address and data toggle switches, examine/deposit/start, stop, reset, and various other controls, as well as LEDs for display of the address and data busses. Finding one of these is probably pretty unlikely, at least in a timely fashion. Building one would be a bit time consuming. I don't have a logic analyzer, so trying to track down DRAM failures with just a scope would be tedious. Thus, I'm more prone to doing some component substation to see if I can just eliminate the cause. With the service manual saying that that the FAULT light being on is most likely indicating a RAM problem, I started looking at the datasheet for the MK4027 chips used in the 4907. I don't have any of these (or equivalents) on hand, but I then looked at the datasheet for the MK4116 (a 16K x 1 DRAM), which, I happen to have a bunch of. The only difference between the two devices is that the MK4116 replaces Pin 13 (~CS) on the MK4027 with A6. All of the other pinouts are identical, and the voltages and general timing specifications are the same between the two chips. What I'm wondering is if I could just pull all of the MK4027's and replace them with MK4116's. In the schematic for the RAM section of the 4907 controller shows the ~CS pin on all of the MK4027's is wired to ground. This would have the effect, with a 4116 plugged in place of the 4027, of forcing the A6 address line 0 at all times, which, if my brain is working properly (there are days when I wonder), should not have any effect on the 4116's ability to serve as a replacement for the 4027s. Any thoughts or guidance? It might be a quick solution to the problem without having to find replacement 4027s. Of course, this all assumes that the support logic around the DRAMS (refresh, address multiplexing, parity generation & detection, etc.) is all operating properly, which also could be the cause of the problem. Thanks to all who have responded. -Rick

The problem comes when the CS/ pin of the 4027 is used for something. In that case, if you replace the 4027s with 4116s, you are likely to end up with a bus contention, since the 4116s will probably not be disabled at the right time. But if CS/ is grounded, the selection must be done in some other way (gating the CAS/ line is the most obvious). In which case there is no problem. It should work fine. -tony

Bob Rosenbloom confessed: You have HOW MANY? I have been looking for a 4907 for at least twelve years, and I've never seen any evidence that any even survived! Are any of them for sale? Which side of the Atlantic are you/they on? What would you like to receive in trade? Sky's the limit... Philip.