There's a procedure given in the manual, where you add chips a few at a time, power up, and test signals, etc. A word of warning. Firstlu, there's at least one mistake in that procedure (one test require a gate from a chip you've not fitted yet!). Secondly, one of the counters (IIRC) in the video chain is used in a marginal way. IIRC NatSemi chips work there, TI don't (or vice versa). The same chip is used somewhere else on the board, I had no end of fun tracking down a video problem when I accidentally swapped them round. -tony

On Thu, 2005-06-02 at 16:17 +0100, Witchy wrote: ta. Hmm... looks like it strictly needs 16 pins on the Nascom 2 - although I can't find a keyboard schematic yet (just one for the Nascom 1, which looks to be rather different) I suppose I could have the reset switch with the case rather than on the keyboard in order to save myself a line between keyboard and box - anyone know if the reset switch on the Nascom 2 keyboard does anything on the keyboard itself, or does it merely connect the CPU board's reset to ground (as with the Nascom 1 keyboard)? It's either that (and use a DA15) or use a DB25 (hence thicker cable) with lots of unused wires... hmmm... cheers Jules

I actually meant DB25 here, but DA26 would do as well :-) -tony

I would agree that a DB size shell with 16 pins is a DB16, but I would also claim that a DB shell with 25 pins fitted is a DB25 no matter how many (or few) of them are wired. -tony

On Thu, 2005-06-02 at 23:36 +0100, Tony Duell wrote: ta... nice to confirm it's not in the manual anyway. I think I'll trace out the keyboard schematic and see what the reset line does. If it's not needed by the keyboard itself, I'll move it to the CPU chassis and use DA15's for the keyboard. If it is needed, I'll have to go the DB25 route. I did wonder about DA26 connectors, but I'm not sure if those were about when the Nascom 2 was current (I'm trying to use period bits for everything). Plus I don't think I have any in the junk box anyway... Cutting out the right shaped holes in aluminium for these connectors is a right pain in the butt, so I don't want to have to do it more than once ;) ta for the info Jules

On Sat, 2005-06-04 at 00:29 +0100, Tony Duell wrote: 4 x 7400 ICs, 2 x LM3086, then a 7493, 74123 and 74145. Hmm, that's interesting. That's the exact same chips as on the Nascom 1 keyboard (which only uses 6 data lines, amongst other things. I'll definitely trace as much of a schematic as I need to I think as maybe the Nascom 2 keyboard doesn't use 8 data lines either (even though they're all present on the CPU board's connector) Aha, yes we have one of those at the museum - long time since I've peeked inside, but I seem to remember it was a pretty tricked-out system. Don't start that again ;-) Hmm, be nice to have one of those one day. In the past I've used steel mounting brackets from PC expansion cards that have the right connector hole as a template for cutting aluminium. Clamp them over the piece of work, then remove as much material as possible with a drill / snips before finishing off with a set of files. It works pretty well for putting the right holes in aluminium, just takes a while to do a nice job where a proper tool would be more or less instant. Presumably the punch tool has a limit as to how far from the edges of a piece it can reach? cheers Jules

That's the type I have. It's not a Q-Max brand, it's RS (http://www.rswww.com/ IIRC). I've never seen an odd-shaped genuine Q-Max cutter (I know they did make sqyare and rectangular ones, though). -tony

What, genuins Q-Max ones? I know RS sell (sold?) some odd/useful ones to fit common components (although I don't recall seeing any circular-with-flat ones). -tony

The Nascom 2 keyboard seems to use D0...D6 (in that the port is read, then ANDed with 0x7f). The '93 is the counter I refered to in my other message, the '145 is the decoder. I wonder what the other chips are for. It does use switches and not something more strange, I trust.... Yes, that's what I do for things other than DB shells. One day I'll buy the DE one (which is the next most common size that I use). No, it's not the rack-n-pinion press. These punches work by a central bolt. What you do is drill an 8mm hole in the middle of the place where you want the D connector. Bolt a template (supplied with the punch to that hole, then drill 2 3mm holes through it (these later become the mounting holes for the connector). Fit the 2 parts of the die to the panel -- they have pins that go through the 3mm holes. Fit the punch into the die with the bolt going through the 8mm hole. Fit the ball thrust bearing, washer and nut, then tighten the nut to pull the punch through the panel. As I said, if you want a DB hole done, and the aluminium is thin enough (see the spec on the RS web site, etc) I'll do it for you. -tony

At 00:29 04/06/2005, you wrote: You were building a Mullard 5/10 valve (tube) amp! Richard

At 00:26 15/06/2005, you wrote: So close, but a typo loses the point! It was the 3/4 and 1-1/8 that threw me - I was building a 5/20 variant one time and needed those cutters. I reckon the 5/8 was for a fuseholder. I've transcribed the Mullard book to CD. Richard

On Thu, 2005-06-02 at 23:36 +0100, Tony Duell wrote: One oddity with the IDC header on the keyboard PCB itself - it's not numbered in a conventional fashion; pin numbering according to the PCB run from 1-8 down one side and then 9-16 back down the other side (i.e. same as a 16 pin IC). Meanwhile the CPU board keyboard connector (and corresponding schematic) is numbered in a more conventional pattern with odd pins on one side and even on the other. Well the Nascom 2 keyboard seems to be very similar to the Nascom 1, except that (as others have noted) it has an extra scan row so uses 7 data bits rather than 6. Otherwise the signals seem to be the same, except: 1) VCC is only on pin 16, not 15 and 16 as with the Nascom 1. 2) Pin 8 on the keyboard PCB connects simply to a resistor, the other end of which is connected to VCC. I assume it's supposed to connect to DB7 on the CPU's data bus and allow detection in software of whether a keyboard's present or not... In all it uses 13 out of the 16 pins on the keyboard PCB. Pins between the keyboard and CPU board *do* match up 1-1, it's just a case of ignoring completely the pin numbering on the keyboard PCB! cheers Jules

On Thursday 09 June 2005 17:18, Witchy wrote: He said that the one connector was numbered 'across' and the other connector, like ICs, is numbered down one side and up the other... Time for new reading glasses maybe? :) /me ducks Pat -- Purdue University Research Computing --- http://www.rcac.purdue.edu/ The Computer Refuge --- http://computer-refuge.org

Eeek!!! [Oddest pin numbering I ever saw, on a 50 pin header (2 * 25). 1-24, then 49 down one side, then 50, 25-48 up the other side. That is, like a 48 pin IC with pins 49 and 50 tagged on one end. Not suprising, as that's what it was. It was called the Elmer board connector in an HP150-II. Elmer was one of the custom chips in that machine, it contained the real time clock, baud rate generator, some of the comms handshake line ports, etc. Problem was, HP didn't get it finished in time, so some early machines had a little PCB with normal chips on it. That plugged into that extra connector...] Anyone who uses different pin numbering at the 2 ends of a straight-through cable needs to be shown the clue-by-4 :-) Alas I don't have a Nascom 1 schematic... It may not even be used at all. The keyboard scan routine, as I said, just ANDs the input data with 0x7f. WHich 3 signals are not used? I would guess at Q2 and Q5 and one other. Argh! :-) -tony

I've now taken a look at that.... Initially I was puzzled. It makes no sense at all if the keys are simple switches.... Tell me (as I can't get to my Nascom at the moment), how many connections does each key have? My guess is 4. Moreover, I will guess that there are 2 pairs of pins that are dead shorts to each other, but one pair is insulated from the other. And that this doesn't change if you press the key. My guess (again) is that these keys are actually magnetic. There are 2 loops of wire trough a little magnetic torroid core. Pressing a key brings a magnet against the core, changing its magnetic properties, anf thus the coupling between the loops. Each of the lines on the matrix in the schematic is a set of these loops in series. Horizontal lines are the sense loops, vertical lines are the drive loops. That would explain the cryptic comment on the schematics that the order of keys in the matrix may not be the electrical order on the PCB. Now for an amazing coincidence (if my guess is right). As you probably know, I've had an HP9845B on my bench for the last few months [1]. Last weekend I finally finished working out the rear section with the processor and memory boards, the PSU, etc, and started on the keyboard assembly. Tge keyboard is exactly as I've just described. The electronics is somewhat different to the Nascom 1 keyboard, but the 'switches'[2] may well be the same. Hmmm... [1] My advice to anyone with a non-working 9845 is to find some other mug to fix it. If you're not that easily put off, I will try to answer hardware-related questions. [2] Most of them have 4 pins, as described. The arrow keys have 6, one drive loop and 2 sense loops. The reason is that said keys contain 2 core as they kave 3 postions -- up, pressed gently (single cursor movement) and pressed hard (repeated cursor movement). Well, it _is_ an HP... It can't be a keyboard detect line. It goes to a TTL input, which would float high anyway. It was probably easier to use an 8 bit buffer on the CPU board, and there was no good reason not to wire all 8 inputs to the keyboard connector. The first 2 could be used as general-purpose outputs... NMISQ/ is somewahat like the reset input, but it produces an NMI (Duh...) I wondered if it was used for the break key or something, Apparently not. -tony

I've spent some more time on the HP9845 keybaord. There's one correction to that dsecription, pressing the key moves the magnet away from the torroid, thus desaturating it and increasing the coupling between the loops. The 9845 electronics (and it seems to be similar to that used in the HP250, the printset for which is on bitsavers). It's quite different to the Nascom 1 schematic, but I suspect the keys may well be similar.