In article <m11joav-000IyNC@p850ug1>, ard(a)p850ug1.demon.co.uk (Tony Duell) wrote: In my (commerical) experience more often missing than hidden. Vital details noted only in the margin of a single copy of the manual, scribbled after a telephone conversation with the designer, because no one ever thought to, or had time to, document them fully. Of course some things are hidden *as well*, but at least things which are merely hidden might eventually be released. I'm in an almost uniquely fortunate position of being able to generate usable microprogramming tools with very little effort. Only slightly. I have a software background and while I can (in principle) understand a processor down to the gate level, I have only a very vague idea of how things work electrically. I hope to remedy that this winter. Sometimes it's even worse. Try explaining to a client that no, we really *can't* write an assembler if they want to keep the instruction encodings secret. The previously-mentioned manual for this machine appears to be an architecture manual rather than a hardware manual as such (although, since it's in the form of a set of scans, I haven't had time to read much beyond the table of contents yet), so it isn't any help for this problem. Both it and the Daybreak manual do appear to document the microinstructions, though. The secondary end doesn't appear to be shorted, and the fuse blows anyway :-( Unless I've made a mistake, things look pretty simple (where + is a connection and -|- is not): L----Fi----+---+--switch------1(| t |)5--+----fuse2---A front 120V lt | +--fuse1----+--2(| r |)6--|--+---------B panel N----er--+-|---------------|--3(| n |)7--+ | | | +--4(| s |)8-----+ other fuse1 = 1/4 A 250V fuse2 = 3/4 A 250V switch = (part of) the main power switch. trns = "DALE IPL-2329-25 105P80342 8445"; properties not marked. Fuse1 is the one that blows instantly. AB leads to the front panel only. With the power switch off, and the front panel disconnected (i.e. no load except the meter), a 1A fuse at fuse1 lasts about two seconds, during which time I measure about 15V across AB. -- Kevin Schoedel schoedel(a)kw.igs.net

On 1999/11/06 at 12:20am +0000, Tony Duell wrote: I'll time myself once I find the encodings :-) For the Daybreak, that might be the hardest part. I understand the 80186 loads it on boot, but how or whether one might make things happen in a running system I don't (yet:-) know. I think I can find a way to convert them en masse to another format. They total some 65M, however, so if as I assume you are connecting by phone (you seem to start posting at 6pm...) you might not want to download the entire set. If you wish, I'm sure I'd be able to get a copy to you on CD-ROM or paper at some point in the next few months. That much only requires that I find and read the relevant sections, which I'll most likely do this weekend. I was wrong, of course. I had done it visually, but missed a trace under the body of the transformer, *and* mis-oriented the switch leads when I checked them. The wiring is actually absolutely straightforward: N------------+--1(| t |)5--+----fuse2--A front L--fuse1--+--|--2(| r |)6--|--+--------B panel | +--3(| n |)7--+ | +-----4(| s |)8-----+ (Sigh. I'm still at the point where I can't tell that obviously wrong things are wrong.) This seems to rule out everything but the transformer :-(

On Sat, 6 Nov 1999, Tony Duell wrote: ________O/_______ O\ ________O/_______ O\ Would it not, perhaps, be a good check to disconnect the paralleled windings and check them out individually to determine whether the two input or output windings might be bucking instead of properly paralleled. Resistance checks might also indicate if there were shorted windings in one. - don

At 7:19pm -0600 1999/11/07, Mitch Wright wrote: Depends what you mean by "laying around". We're still using the same basic software (with improvements over the years, of course). Some clients have rights to distribute them (in binary form) on their own terms, so there may be some available somewhere, but I don't know. Unfortunately I am stuck with enough NDAs that I'm not even sure which past example I could name. Purely out of curiosity (and not because I intend to sell mine; I don't) I checked ebay; someone sold a Daybreak for about $100 recently. Compared to ebay prices of other things, this suggests they are not *too* rare. I don't even want to *think* about how much I could sell the 8010 for there, if I were to invoke the "S" word (even though mine has probably always run Lisp, and is, by its memory and disk, clearly a late instance as well). Don't tell me; I want to keep it :-) A co-worker picked up the other of the two available Daybreaks; if he gets tired of it, I'll let him know you might be interested. On 1999/11/06 at 7:15pm +0000, Tony Duell wrote: Part of the manual I've read mentions that the 80186 loads the control store during booting "and debugging". I haven't yet seen how; I suspect the details are not in any manual I have. In most of North America, a flat line charge includes unlimited local calls; that's why people here throw around binaries with abandon. "The keyboard/interface uses the receive half of the i8251A interface chip" suggests it's one-way communication. "... receive from the keyboard as a differential signal pair by a 75176A receiver chip." I don't know; does this imply RS232-compatible levels? "... asynchronous serial interface with a data rate of 9600bps". I do not see the protocol documented, but by the time you get around to your machine, I should be able to hang my keyboard and mouse on some serial port and/or logic analyzer to work it out. After removing the transformer, I believe I've confirmed shorted turns on one half of the primary. Assuming no shorts in the secondaries, would this not mean that the voltage I measured (15V) is an upper bound on the correct voltage? [moved:] I'm not sure I can do this accurately enough; the outermost layers seem inextricably stuck in some hard resin-ish stuff. It appears to drive half of the front panel board -- the non-LED part. Now, there's a 5V regulator (SG340K-5) there. The output is clearly Vcc for the TTL on this half. The ground is clearly GND for the TTL and is connected to one line from the secondary. The input of the regulator -- I might be wrong again but I've double-checked -- is connected to the other line of the secondary through a single diode (in other words, it seems to be a half-wave); there's a 1000uF to ground. So, one of the things I need to do is keep the regulator happy, which seems straightforward; the question is whether the rest of the circuit imposes tighter constraints. I can't easily trace it all because some traces are hidden by the soldered-in TTL or the capacitor glued across vital parts of the back of the board, so I'm trying to go by what I can see and by considering why it's there. I can only think of two reasons why this might need AC rather than just using the main power supply. It might be a power-fail indicator, but I think that's implausible; since the machine has semiconductor memory and power-hungry disks, there's nothing it could *do* on power fail. And it does not seem consistent with the parts on the board. But if this *were* the case, it would presumably tightly constrain the required secondary voltage. The other possibility I can think of is that it's a line-based clock/timer. There are further diodes and resistors feeding (in a manner I can't fully trace) a pulse generator (74123) whose output appears on the cable from the front panel to the I/O board, and a trigger/inverter (7414) which, I think, drives counters (74LS163) leading to a flip/flop (74LS74) whose output likewise appears on the cable. Does this sound plausible? And would it impose any tight constraints on the transformer output? -- Kevin Schoedel schoedel(a)kw.igs.net

A non-text attachment was scrubbed... Name: not available Type: text Size: 692 bytes Desc: not available Url : http://www.classiccmp.org/pipermail/cctalk/attachments/19991108/7be2715f/at…

On 1999/11/08 at 8:00pm +0000, Tony Duell wrote: I actually paid more for mine, because it looks too much like a modern PC, and *less* for the 8010, because it's obviously too old to be good for anything. I still owe a summary of the microinstructions; unfortunately my 'main' machine is a 'classic' in its own right (i.e. it's old) and one of its disks, containing my notes, suddenly and catastrophically failed yesterday. Here is a very brief summary: 00 - 03 rA 2901 register number 04 - 07 rB 2901 register number 08 - 10 aS 2901 alu sources (A,Q; A,B; 0,Q; 0,B; 0A, D,A; D,Q; D,0) 11 - 13 aF 2901 alu op; the table in the manual is obviously botched 14 - 15 aD 2901 destination (values for rB, Q, Y bus) 16 - 19 Cin,enU,mem carry in; memory op (cycle dependent); U register op 20 - 23 fS determines interpretation of fY and fZ fields 24 - 27 fX misc operations (mostly seq call/ret, push/pop) 28 - 31 fY misc, encoded per fS, includes branch ops, or immediate constant 32 - 35 fZ misc, encoded per fS, or immediate constant 36 - 47 INIA immediate next address The next address is the INIA field optionally ORed with bits determined by branch ops in the previous instruction. Cycles are grouped in "clicks" of three cycles (c1, c2, c3) which determine the interpretation of the 'mem' bit: address, read, write. Five clicks form a "round". There are multiple (presumably six) microinstruction pointers ("tasks"), corresponding to clicks in a round. Each click in a round is associated permanently with and I/O service routine; during any particular click, either that I/O routine executes (if the device so requests) or else the macroinstruction interpreter executes. Besides the 2901 R registers, there is a 16 x 8-bit RH register set, grouped with the corresponding R registers to form memory addresses. There is also a 256-word external register set U, 16 words of which can be quickly accessed by a stack pointer. The large gate array on the CPU board is the sequencer/decoder; the smaller one is the bus controller. I didn't mean to suggest they wouldn't be worth saving. I'd never pass over a non- commodity-microprocessor machine at any tolerable price. (I've even bought uninteresting machines (cheaply) from that same source just to make sure that they know that 'big, old' computers are *saleable* and should not be scrapped. Anyone want three Sun 3/160's?) On mine, there are *very* few parts that have 733 numbers only. Nearly all have only obviously generic numbers (e.g. 74LSxx), or both generic and house numbers. The schematics appear to use only generic numbers. The documents contain some schematics, but in their current form (PDF bitmaps) it's really difficult to find anything in particular. Once I'm able to print them it should be possible to find things more easily. Keyboard reset. "unused" Yes. Can't find it. Noted on the schematic: 1 start bit, 8 data, 1 stop; 1 to 3 bytes of keyboard/mouse data per event. This is probably somewhere in the document as well, but I don't see it. I'll probably try that next. The machine ran for a few seconds on my initial trial (before the first fuse blew), when there obviously was already a problem, which suggests that the voltage is not too terribly critical. -- Kevin Schoedel schoedel(a)kw.igs.net

On 1999/11/09 at 2:16am +0000, Tony Duell wrote: I think it ranks as the second most complicated machine I've seen. Most I've seen are much more like the PERQ in spirit: alu, registers, simple sequencer, straightforward muxes. The internals don't appear to be documented in anything I've seen so far. I'll put it on my list of things to do when my brain isn't working. Should be done soon.... They're among the PDF files from Al Kossow's site. TechRef_b?_???.pdf and IOPschem.pdf at least contain schematics; I don't know how complete. Ghostscript appears to read the files fine and ought to be able to save the contents in Postscript or other form. If you have no better means, I'll eventually send you a printed copy. -- Kevin Schoedel schoedel(a)kw.igs.net

On 1999/11/09 at 7:18pm +0000, Tony Duell wrote: That's OK; I over-simplified the Daybreak :-) Some of the operations are weird, and only a fraction of encodable operations actually work, due to timing constraints. I don't fully understand it yet; I'll have to print the documents first. Hmm. The Daybreak has 8K control store, but only 12 address bits; everything I've read so far only mentions 12 bit addresses. This reminds of a device I once worked on, made by a company whose name you'd recognize. The part was *almost* a complete processor; it had a sequencer, ALU, registers, etc. but didn't quite nail down the instruction set. The manufacturer had designed an evaluation board; I was evidently the first person to try to program it. They had taken an extreme RISC approach, providing only one addressing mode: indirect through a register. The only way to get an arbitrary address into a register was, of course, to load it from an address in a register.... They changed the design. I haven't really looked at it yet. I think two of the gate arrays are involved, but I think it's just a dumb frame buffer. OK. Mine have about 10% 733 only, 60% standard numbers only, and 30% both, but naturally the same parts appear repeatedly. It seems that "leading" zeroes (after the W) don't matter. Here are the matching numbers from the Daybreak boards: 733W 0098 75189 733W 0318 74S00 733W 0319 74S04 733W 0321 74S260 733W 0339 7414 733W 0341 74279 733W 0351 74S257 733W 1523 74S189 733W 1550 AM27S07DC 733W 1606 74S10 733W 1611 74S08 733W 1616 74S138 733W 1619 74S20 733W 1620 74S64 733W 1621 74S51 733W 1624 74S138 733W 1625 74LS240 733W 1626 74LS244 733W 1630 74S175 733W 1633 74S240 733W 1634 74S241 733W 1638 74S280 733W 1640 74S374 733W 1643 74S02 733W 1644 74S11 733W 1646 74S32 733W 1647 74S38 733W 1648 74S86 733W 1652 74S157 733W 1663 74LS393 733W 1675 74LS74 733W 1676 74LS191 733W 1682 10124 733W 1698 74LS374 733W 1699 74S373 733W 1747 74LS259B 733W 1770 74LS163 733W 1771 74S74 733W 2136 74S37 733W 2137 74S151 733W 2369 MCA1300PKG/E5 733W 2380 MCA1300PLM/E4 733W 2538 CY7C122-25PC 733W 2550 C67402J 733W 2552 MK4501N-12 733W 3042 AM2901CPC I might have missed some, there might be errors here, and I haven't looked at the Dandelion boards. -- Kevin Schoedel schoedel(a)kw.igs.net