Crazy with Cables by DB Moore 43 !¡ 43 !¡ 43 !¡ 43 !¡ 43 !¡ 43 !¡ 43 !¡ 43 After what now seems like half of my life, [but was only in fact about 18mths], of living with my A500 with only DF0: to do all the disk-functions, this Xmas I bought myself [and my Meegee!] a present....in the form of a Scram-500 expansion/SCSI board. Buying this, for me, was stupidly belated, because I've had a Seagate ST-157 h/drive sitting in my cupboard for the last 11 of those 18mths! [I should say that at about the 10mth mark, I did hook up a 5.25" drive to Meegee, which I used to run as a 'virtual disk'. I used to keep all my Workbench on the 5.25-unit, boot from DF0:, then assign everything over to the other drive. Archaic in every sense....but at the time, it was an absolute boon!!] Anyhow, I'd ordered the Scram in 'caseless kit' form, so when it arrived in the post, I reverted back to the assembler that I am, told my children that if they bothered me for the next few hours I'd solder *them* to the clothes-line outside, and set about the task. It's only a small board, really well made and masked, and in around an hour, I had it all done and tested, and plugged it onto Meegee. The A500 complained [mysteriously!] for the first couple of power-ups. I pulled the board back off the expansion-busz, checked it out again and found nothing awry so refitted it, and powered-up again. This time, it auto- configured without a glich, so I figure it was my fault for not wiping the crud off the seriously neglected expansion busz contacts! [I was a bit excited at the time!] However, my finances being as they are, [like, usually near zero!], had excluded any notion of populating the board with any ZIPs, but I knew this in advance. The first objective I had in mind when I purchased the Scram, was the SCSI-controlling facility to get the Seagate on line. This is where I came up against the cable problem. The Scram-board has a DB-25 'industry standard' SCSI connector, where-as the Seagate has a 50-way DIL connector on it. The disk-manual that comes with the Scram, suggests for those 'on the cheap' to make their own cable up. Good and well, but nowhere on the disk, nor in the printed manual that accompanies it, is there any pin assignments for the 50-way connector....just the DB-25 is covered. Extremely dejected [and threatening the kids with the soldering-iron!!], I rang Chris Ralph [who I must here thank from the bottom of my hardware-heart, for being such a tolerant and helpful guy....he really knows his stuff!], and left a message on his answering-machine expounding my dilemma. Before he got back to me, I rang around a few other Amiga h'ware people...who didn't have the pinout info I wanted, but who funnily enough admitted to having had the same problem at sometime in their life....ie. the need for a 25->50way SCSI cable. One or two of them, even said they'd tried to construct this cable once upon a time, and they couldn't get it to work...even though it was seemingly electrically correct!? Later that same evening, Chris got back to me with the 50-way pinouts, and told me that in the book that comes with the Scram2000, these pinouts are supplied!! [damn and blast why they didn't put them in the Scram500 book?!] Anyway, to perhaps save others from this, those pinouts are here below, with the DB-25 pinouts in the right column: SCSI 50-way 25-way pin 2....DB0 pin 8....DB0 pin 4....DB1 pin 21...DB1 pin 6....DB2 pin 22...DB2 pin 8....DB3 pin 10...DB3 pin 10...DB4 pin 23...DB4 pin 12...DB5 pin 11...DB5 pin 14...DB6 pin 12...DB6 pin 16...DB7 pin 13...DB7 pin 18...DBP pin 20...DBP pin 32...ATN pin 17...ATN pin 36...BSY pin 6....BSY pin 38...ACK pin 5....ACK pin 40...RST pin 4....RST pin 42...MSG pin 2....MSG pin 44...SEL pin 19...SEL pin 46...C/D pin 15...C/D pin 48...REQ pin 1....REQ pin 50...I/O pin 3....I/O all ODD-numbered pins...GND pins 7,9,14,16,18, & 24...GND pins 34,20,22,24,26,28,30 are all not connected The interesting thing, was that Chris too had tried to patch a cable like this with no success. He postulated to me, he thought that maybe the capacitance of the cable was 'screwy' or something??? It's a bit grey, but I figured nothing ventured, nothing gained...so armed with this new info, I sat down at my bench to make one. In view of what Chris had said, I figured that if initially I kept the cable as short as possible, this might alleviate any capacitance problem....or highlight it!! With my 'de-cased' A500, I could make a cable only about 10cm long that would reach the Seagate [sitting on the desk!!] So I made the cable, 10cm long, metered it out to check, and hooked it all up. Power up, boot-up with the Scramprep-disk...all OK so far....then tried to run the prep-program....and nothin' happened!! Meegee had *tried* to load the prep-program [watching the memory count!], went looking for ?something?, and then quit it! Hmmm. I then put SnoopDos to work, and found out that what it *had* gone looking for, was something called 'Scram16.device', which was where it was supposed to be on the disk..it just wasn't interested in co-operating for some reason. Suspecting that maybe the software on the Scramprep-disk was NFG, I de-lzh'd the archives on the Megadisc PD disk Misc-147 [where I'd first discovered the Scram-board!!], because all the prep-software is on that disk as well. I tried again.....and got *exactly* the same response, !zilch! I reckoned it highly unlikely that the software was to blame now, had a few horrid thoughts about a dead SCSI-chip, or a dead Seagate, but in view of what others had said about their troubles with this cable-patch, I eyed the convoluted mass of ribbon cable hooked up to the Scram [which closely resembled the dreadlocks of and aging rastafarian!], shrugged my shoulders, and figured that I was going to have to do what the Scram book suggested, and go to a Mac-dealer to buy the cable 'pre-made'. Well, I got to the 'fruit-shop', and asked for said cable. The guy I asked burst out laughing.."I'm sorry sir, I only *sell* these machines!", [typical!], but he called hither their techie from the back room, who sorta [????] understood what I wanted. Nope, they didn't have one, but they *did* have a DB-25 to 50-way Centronics cable and a 50way Centronics to 50-way DIL adapter. How much? You don't want to know....and neither did I! I tapped on the screen of a close by Mac LC....muttered "Rubbish, you know that?!" [I used the 4 letter word that starts with 'sh__' and equates to rubbish!], and left for home again. I had to give this homemade cable another bash. I've always been dubious about IDC-connectors. Even though they meter-out OK, that in itself is no guarantee that their working integrity is assured...so I bought a couple of spare 50-way type just in case. When I got home, I cut the existing IDC connector [on the h/drive end] off, and installed a new one on my now 8cm long cable, and tried again. Nothing, no improvement what-so-ever. What ?was? going on here?? On the disk that comes with the Scram [and on Misc-147], there's a bit of a discussion about these two SCSI-cable setups...about how the 50-way ribbon has an earth wire [all the ODD number wires] interleaved between the signal-wires to stop 'crosstalk', and how the 25-way 'bundled' cables try to achieve the same end, by the cancelling interaction between the signal/ground wires in their construction. It goes on to point out, that the 'bundled' cables are not nearly as reliable as the interleaved types in eliminating 'crosstalk', especially at high transfer rates....makes you wonder a bit! In light of this, I looked at my construction technique again, this time a bit more objectively. Everything was interleaved right up to where the 50-way ribbon met the DB-25 connector [solder-type DB-plug, I might point out]. Here, the ribbon-cable sprayed out like Don King's haircut, with all the 'not connected' wires cut short, all the signal-wires soldered to their attendant DB-25 mates, and all the ground wires bundled together in 3 groups, with three ground-wires coming off the DB-25 to the 3 ground 'groups'. Could it be, that in the last couple of centimeters of cable, this ground wire grouping was letting some arcane 'crosstalk' take place, that was thwarting the signal communications?? You'd hardly think so, but..... I removed the cable yet again, to try and see if this was the case. I removed all the wires from the DB-25, clamped it in the vice, and soldered 6 , 90degree 'launcher' pins [like the ones on the A500's memory expansion slot] to the GND-pins on the DB-25 connector. The 'launcher' pins on the GND-pins numbered 14, 16, 18, and 24 I soldered with their ends pointing 'up' towards the 'top' of the DB-25, and the other 2 on GND-pins 7 and 9, pointing 'down' towards the 'bottom' of the connector. [make sense?] I then tinned the ends of the 'launcher'-pins rather liberally with solder, so as to facilitate 'tacking' the GND-wires to them later on. Next, I re-soldered all the signal-wires of the ribbon cable to their respective terminations on the DB-25. This ain't too bad a task....!hint!>>> do all the 'top' ones first [pins 1->13], then flip it all over and do the 'bottom' pins. Then, all the 'not connected' wires of the 50-way ribbon...[7 of them]...I simply 'folded back' on the cable, and held them flat against it with a piece of 'gaff tape'. Now,[and this is the fun bit!!], you'll have 25 GND wires left to solder to the end of the 'launcher' pins. The trick is [I think??], not to go *directly* to the GND terminations, but rather, pretend you're a weaver of sorts, and 'thread' the GND wires up, though, across, and around the signal-wires in any fashion that you think my look still like there's a semblance of interleaving between the signal-wires. Fine, pointy-nosed pliers were my best friend in this task. I did all of this, metered the cable to check it, and propping the Seagate up on a couple of match-boxes [so it'd reach...remember, the cable was only 7cm long at this stage!!], plugged it all in and powered up again. There was no obvious sign of success, until I tried to run the Scramprep program again....and bliss of bliss, this time, it *!did!* load-up. Some-one should have said in the documentation, that Scramprep *won't* load, unless it can see the SCSI-device being there....which I suppose is what the Scram16.device does! [you live and learn!!] Of course, an 8cm long cable with the h/drive sitting on match-boxes is not only impraticle, but it's also very disconcerting, so I figured seeing that I needed a longer one, I could try and eliminate a couple of other 'grey' areas. I made another one up using the same technique, and this is the one I'm using now, at a length of 40cm, and it too works fine. I might point out, that there's no way the threading of the GND-wires would be exactly the same as the first one I made...the pattern doesn't seem to be as important as the technique, if you get my leaning here. Just to prove a point to myself, I made another one up at 55cm [which I think is about the longest anyone should ever need?], and like, it's no doubt got a different pattern of threading with the GND-wires than the other 2, but it still worked fine...so there you go! By the way, if you take to the plastic shell of the DB-25 with a hacksaw, and cut a slot the width of it and about 6mm high, you can get it all back together again with a bit of patience! I hope this helps some-one out. I hope this works with other brands of hard-drive and SCSI-devices too! [you never can tell, cables are like that....I've a Seagate ST-157N]. Seeing as I stuck my foot in it, and told those other people who've had trouble with this cable about my success with it, I now am making a couple for them at their request. If any Megadisc subscriber would like me to do one for them, I suppose you could ring Tim or Steve at MD, who'll no doubt point you in my direction. Cheers! DBM. Note: Maybe the last paragraph should be the opening one!!? - = - = - = - = - = - = - = - = - = -  Addendum: Since I wrote the SCSI-cable text, I've now completed the construction of another 6 cables [taking the total to 9!], and I'm going absolutely bonkers! It's 'tedium plus' doing ribbon-cables, especially with the 43 solder-joints to do on the DB-25 end. The advantage of doing this many though, is learning a procedure that makes it as 'stress free' as possible. It also helps establish that what I'm doing actually works, and *ALL* of the cables made so far, [including one I did at 80cm long], work just dandy on my set-up. As I've only one SCSI-device to plug into the Scram, I've no way of checking this method of cable-construction works with multiple SCSI-devices....but I'd be interested in hearing from anyone who does have a try in this respect. I'd be equally interested in knowing if it works on other brands of h/drives or SCSI units too! As said though, one tends to figure out a way of doing it, so that you don't go 'bonkers' in the attempt, so here are my tips: I don't install the 50-way DIL header-plug on the drive end, until I have finished soldering the DB-25 end of the cable in situ. I found this made life easier, without the extra weight and cumberance of the IDC plug hanging around. [I *still* don't like IDC-connectors....I refitted the one I use, with a solder-pin version of the 50-way connector on it, mainly because these withstand being removed/replaced often....to check other cables out, in my case. I know that IDC-connectors *aren't* as resilient in this respect.] Preparing the end of the ribbon-cable for soldering is where you can save some grief. You have to 'singalize' all of the conductors for soldering to the DB-25. Peel them all apart, to a length of 30->40mm. Next, segregate all the odd-numbered wires from the even ones...[ 'oddeventhied'??? ].....by folding them back to either side of the ribbon-cable proper. Cut all the even numbered conductors about 4mm shorter than the odd-ones, and then cut all the 'not connected' wires [20, 22, 24, 26, 28, 30, and 34] about another 4mm shorter again. This gets these wires 'out of the count' for good.... it's handy. Now you have the truly admirable task of stripping and tinning the last 2mm of the remaining 43 conductors....wot fun!!!! When you've finished, keep them segregated as above. Soldering the 90-degree 'launcher-pins' to the DB-25, is as per the other text....it's the best way of doing it, just remember to give the ends a nice thick coat of solder before the next bit. Clamp the DB-25 connector somehow, with the solder-pins facing you and the connector 'the right way up'. I found it best to start at the right of this connector [pin 1], and work left across the top row of pins, towards pin 13. If you follow this in soldering the respective signal-wires on their associated terminations, you'll end up looking at about half [the last half] of the wires, crossing over top of the first ones you soldered...which looks horrific...but if you then free the DB-25 from whatever's holding it, and flip it over 180degrees, things will again look a tad more sane, and you'll have the bottom row of DB-25 pins to solder-up. This time, work from the left [pin 14] towards the right in soldering the remaining signal-wires in place...it's easier in my opinion. This part is made easier by keeping the GND-wires segregated as above....it makes for easier counting, and you're less likely to stuff-up! If you've done OK, it will again look frightening, with only [!!!!] 25 GND-wires to attach to the 'launcher-pins' previously installed. These will have some room to work with [why I suggested cutting the signal-wires 4mm shorter than the GND-wires], and you just go at it with the 'art of weaving' in mind. If they come from the left, feed them through the signal-wires, and attach them to the right, and vice-vera. You'll only need to attach 4-wires to every 'launcher-pin', and the whole convoluted mess will need to be flipped over a few times before it is all finished....but be creative! Try and get those GND-wires through and between as many signal-wires as you can. Now, I said before that you may need to take to the backing-shell with a hacksaw, to get it on the connector, and this may still be the case in some instances. However, I found that if you get the b/shells with the biggest aperture [for thick cables], and stick to the lengths mentioned here, you can give the DB-25 a twist or two relative to the ribbon-cable, and things go together with a bit more ease. The cable seems to be unaffected by this procedure. Lastly, you have to fit the IDC-connector [if you're using one], and here, I can surely say 'good luck'! I used to make a *lot* of these in relation to several assembly jobs I've had [about a couple of 100 a day!], and !EVEN! taking all the care in the world and having the correct tools for crimping them in place, it was still so that around 10% of them didn't work.....usually shorted somewhere....or elsewise open circuit at the connector. As a matter of fact, I used to cut all my ribbon-lengths about 2cm longer than needed to be, just in case of such a happening, which gave you the scope to cut the offending connector off and installing another one...and the cable still reached. IDC connectors are usually *very* hard to take apart from the cable intact, they more often than not break the little 'latching tabs' at the side off. In the bin with these! I suspect the trouble lies not with the connector itself, but with the manufacture of the ribbon....all the inner conductors need to be *perfectly centered* within the insulation surrounding them, for IDC to work. It would seem that this is often not the case. I hate advising 'how to fit IDC for the D.I.Y. person' for these reasons.....but here goes. You !*HAVE*! to get the orientation of the plug to pin 1 of the cable right....even before you think of putting the connector on the cable. All IDC-connectors [I hope!] have pin 1 designated by a little arrow either stamped into, or relieved on, the plastic of the connector body. This says 2 [get that...TWO] things... pin 1 is here, and, feed the cable into the connector from THIS side!! Do it from the other side, and pin 1 will be pin 2 of the cable, and likewise for all the rest...IDC connectors and cables go together *one way only*. Remember this!! You can use a bench-vice for the crimping, just so long as it draws closed evenly, if the jaw wobbles up and down whilst you turn the handle...not good! After you've ascertained the orientation of things, open-up the IDC connector as far as practicable, and slip the ribbon into it. It's wise to push more ribbon through the connector and trim it off after the crimping, than it is to try and do it 'flush' in one go. This is because sometimes, the inner conductors get a bit out of line when the cable gets cut of the roll, and this will cause a short upon crimping.....a corollary of Murphy's Law. Next, place this in the vice, and if you can, just hold it there with the most minimal of pressure. Double-check the connector is squarely faced within the jaws. Now, *THIS IS MOST IMPORTANT*. Make sure that the line of the cable's conductors are aligned, as nearly perfectly as possible to 90-degrees in respect of the connector's body, and that the cable is centered correctly within the connector. If all is OK, you can now close the vice onto the connector, do it up firm and the 'lid' of the IDC should 'click' home, or else stay in place when it's removed. Some don't make any sound!! Thats about it. I always meter my cables before I plug them in, by sticking 2 rows of 'launcher pins' in the IDC end, and checking for continuity between the two ends of the cable, and also for lack of same between adjacent conductors....IDC sometimes does this, as mentioned. The reason I went to the extraneous length of tapping out all this text, is three-fold. Firstly...cost. DB-25 SCSI--> 50way SCSI cables aren't your everyday item. If you're fortunate [unlikely!], you might be still able to find one in a 'fruit-shop', and failing that, you may find a company [there's a few about] that'll either have stock or be prepared to make a specific cable like this up. You'll pay!!! I did approach such a company in Brisbane with a query on this cable, and though they didn't have one, they offered to make me one....for $28!! Things were worse in the 'friut-shop', they no longer carry this cable as a stock item, but where offering the setup previously described. For those who *really* want to know, that setup was going to be the better part of $40!! These cables that I've made, weigh-in at under $10...depends on what sort of connector you choose. You could go for 50-way--> Centronics female socket, but you still have to buy the SCSI DB-25-->Centronics cable, but it's still cheaper to do as I have. Secondly, this isn't the first time that I've tried to do a cable-patch like this. Previous experience with it, and I'm not just talking SCSI here, left me scratching my head as to why I couldn't get it to work. Even though electrically correct in every way, they just never did anything but frustrate in their refusal to pass data. This has dogged my sensibilities for years....and I know I'm not alone here. I hope others so frustrated read this, so they may finally sigh in relief of mental torture!! Thirdly, DB-25 SCSI is certainly the 'entrenched' industry-standard now, and I can see that in future time, a lot of people 'on the cheap', will have opportunity to buy 2nd-hand SCSI-units....particularly H/drives.., that have the 'old' 50-way SCSI-standard on them. I think it would be horrid nonsense for folk to be able to buy an old H/drive as so, for say $200-->$300, and then need go and spend another $40 on cabling! This doc. is sorta dedicated to people like this, folk after my own heart, who need to know about cheaper alternatives. I'd also like to have them know that a venture into a 'fruit-shop' is something they need not degrade themselves with! [ha ha ha ha ha] I might point out, that this method of cable construction may be of use for other cable-patches, that either don't work, or are just plain noisey or unreliable. Who knows?? Cheers! DBM 43 !¡ 43 !¡ 43 !¡ 43 !¡ 43 !¡ 43 !¡ 43 !¡ 43