Replica SOFTY 2

[ChrisOddy]

My replica SOFTY 2 is finished !, see here https://theoddys.com/acorn/replica_boards/replica_softy_2/replica_softy_2.html.

Many thanks go to Timbucus for the loan of his machine without which it would never have happened.

Chris

[Alan Bain]

Originally Posted by ChrisOddy

My replica SOFTY 2 is finished !, see here https://theoddys.com/acorn/replica_boards/replica_softy_2/replica_softy_2.html.

Many thanks go to Timbucus for the loan of his machine without which it would never have happened.

Chris

Is this the Dataman design from the days when they were helpful (TM) unlike the Dataman S4 for which schematics seem to be top secret (although they aren't that hard to trace out)?

[SiriusHardware]

If you've actually done that Alan, consider making the diagrams available to all - although I suppose the argument could be made that Dataman still exist and may object to their IP being reconstructed and distributed even if they no longer make or repair the model themselves. (Or do they?).

I know they do still exist because I contacted them a little while ago about the internal dongle upgrade needed to update my MEMPRO to a 40PRO (A cool £175 at the time of asking, which I am still not sure I can justify) - but I did get a polite reply about it, and the upgrade is unsurprisingly still available as I don't imagine many were ever sold at that price.

[SiriusHardware]

Thank goodness the SOFTY 2 used the SC/MP II, as per the MK14. Looking around on Cricklewood's site for something entirely different earlier I happened across their entry for SC/MP 1s - the price from there is now into three figures.

https://www.cricklewoodelectronics.com/SCMP1.html

If I had paid that much for one antique chip I don't think I would ever be brave enough to turn it on, especially given the wacky nature of its power supply requirements. (Plus and minus supplies, rather than the simple +5V supply of the SC/MP II).

Other suppliers probably sell them too, of course.

[ortek_service]

Originally Posted by SiriusHardware

Thank goodness the SOFTY 2 used the SC/MP II, as per the MK14. Looking around on Cricklewood's site for something entirely different earlier I happened across their entry for SC/MP 1s - the price from there is now into three figures.

https://www.cricklewoodelectronics.com/SCMP1.html

If I had paid that much for one antique chip I don't think I would ever be brave enough to turn it on, especially given the wacky nature of its power supply requirements. (Plus and minus supplies, rather than the simple +5V supply of the SC/MP II).

Other suppliers probably sell them too, of course.

Yes, Although I recall when people were paying £600 for a 486DX2-66 CPU IC in the early 1990's, to build their own PC's. So probably relatively > 10x the cost, with inflation.
I'd found out from others on here (at an event at the Cambridge CfCH around a year ago), that Cricklewood still sold these original PMOS ISP-8A/500D (SC/MP-I) IC's (apparently at £75+VAT a few years ago, but it has been at £95+VAT for a while). And I've not found anywhere else listing these yet.
So I was tempted to make an adaptor to use the single-rail NMOS INS8060 (ISP-8A/600 SC/MP-II) in my Introkit, as I knew the one I'd bought worked OK after verifying it in an MK14.
But Timbucus kindly tested my SC/MP-I in his SCRUMPI at the RCF-2023 a couple of months ago.

So fortunately it seems I don't need to buy a rather-expensive 'new' one to try to get my Introkit going (I also recently discovered I'd actually got a good stock of the non-standard pinout obsolete 74L86, so I should hopefully have spares for virtually all the IC's on this, to try swapping all the (socketed) IC's to start with to hopefully find an easy fix without having to resort to doing some logic analysing etc on it, to see why it isn't running)

[ortek_service]

Originally Posted by Alan Bain

Is this the Dataman design from the days when they were helpful (TM) unlike the Dataman S4 for which schematics seem to be top secret (although they aren't that hard to trace out)?

Yes, Dataman supplied the schematics for the original (Barry Savage design) 3-rail SOFTY, (as it was available as a Kit), as well as the quite-similar (but single-rail only) SOFTY-2.

But when they moved to the (rather different / better in many ways as built-in LCD screen for truely portable use) much more expensive S3, they swapped to using a Toshiba CMOS Z80 Microcontroller - that was Mask-programmed (although only had one generic 2764 in its ROM, so had to load a library EPROM into its battery-backed RAM), so much more difficult to mke a replica.
Someone has asked them about schematics here, but they weren't willing to help: https://forum.dataman.com/viewtopic.php?t=59457
Although they do still make available many other resources for the S3 & S4 / sell some spares at not too bad prices. And the S3 manual did tell you about adding you own device / algorithm, and quite a bit of info on how it works (but only a fairly-detailed block diagram, rather than a full circuit): https://www.dataman.com/mwdownloads/download/link/id/38/


It could be that they didn't actually design the S3 (/ the Similar S4), like what happened with their original (Advantech designed) Dataman 48 and the later (Elnec re-badge 48Pro series)

However, from my previous post (#76) in this previous Softy thread: https://www.vintage-radio.net/forum/showthread.php?t=167741&page=4
I did mention that:
"Dataman did make as big thing about how open they would be with the S3, and its hardware and a developers / technical manual with full circuit (rather than just block diagram in the provided manual) and source-code.
But as they later advertised you had to pay £150 to get hold of the 'developers manual', I doubt too many copies were printed. The circuity isn't too complex to rev-eng (especially with quite detailed block diagram), but their use of a Toshiba Mask-programmed for them (with SOFTY marked on it) low-power Z80-core micro and custom? Texas-Instruments made Keypad etc. would have made building you own from these details a bit more involved."

[factory]

I was given an S3 last year, unfortunately it is too far gone to repair, the leaking Yuasa battery pack has pretty much destroyed everything inside, I got a bit of practice of cleaning Ni-Cd corrosion out of it, but not much else. On top of needing a new PCB designing, the costs of replacing of buying new parts (assuming they are still made) would exceed the cost of buying another untested one, plus the possibility of the corrosion having got into the micro & eaten the bond wires.
Of course the S4, with extra memory, seems more popular judging by sold prices.

In the S3, the board is branded Dataman in the copper layer, the "Softy 3" branded custom micro is an NEC in this one, the keypad is from Texas Instruments and the display looks to be a standard part.

I took a few pictures (as there didn't seem to be any others that I could find on the web), which are on the eevblog forums here;
Before cleaning part 1; https://www.eevblog.com/forum/testgear/test-equipment-anonym...erapy-thread/msg5122677/#msg5122677
Part 2; https://www.eevblog.com/forum/testgear/test-equipment-anonym...erapy-thread/msg5123985/#msg5123985

After cleaning; https://www.eevblog.com/forum/testgear/test-equipment-anonym...erapy-thread/msg5138007/#msg5138007

David

[ChrisOddy]

I have one set of black vacuum formed SOFTY 2 cases if anybody is thinking of building one ?

Free with the purchase of a PCB !

Chris

[SiriusHardware]

I'm not in the market for one myself Chris, but getting something like that remade by the original vacuum forming method seems to me to be a whole new level of achievement.

How on earth did you create a model of the original 'lid' from which to make new lids, did you make some sort of cast from Tim's original lid or did you 3D laser-scan it or did you just make meticulous measurements and draw the 'former' in a CAD package or something - and then even if you did that, where did you get them made?

Impressive, any way I look at it.

[ChrisOddy]

https://theoddys.com/acorn/replica_boards/replica_softy_2/replica_softy_2_enclosure.html

[SiriusHardware]

Brilliant! And well done Tim for the loan of the original unit to make this possible. I can imagine there might be one or two original units out there with broken casings, hopefully their owners will find their way to you.

[ortek_service]

Originally Posted by ortek_service

>>
>>
So fortunately it seems I don't need to buy a rather-expensive 'new' one to try to get my Introkit going (I also recently discovered I'd actually got a good stock of the non-standard pinout obsolete 74L86, so I should hopefully have spares for virtually all the IC's on this, to try swapping all the (socketed) IC's to start with to hopefully find an easy fix without having to resort to doing some logic analysing etc on it, to see why it isn't running)

Well re-discovering I had some 74L86's has proved rather fortuitous!
- As I was now able to assemble a complete-kit of spare logic IC's for my Intokit (Or pinout equivalents for the more-obscure 85L451 (= slower 74LS173), although I could only find a 74365(DM8095) in my MK14 replica, as could only find 74HC365's spare).
So I had planned to swap all of these in one go, and see what difference it made (maybe swapping them back one at a time if that did actually fix it).

However, Whilst swapping a temporarily-fitted 7442 to my build of a ChrisOddy Softy-2 replica (that had enabled me to get that running for the first time last week - shortly-after also having my build of his Softy-1 replica running for the first time, after finally-completing the assembly with all required parts), to the proper 7445 (that I'd just received a few of I'd recently ordered off Aliexpress), I discovered the Softy-2 now no longer seemed to work properly!

Some IC-swapping later, I discovered 3 out of 9 of the 7445's I'd bought didn't seem to work in my built Softy-2 replica!
I hadn't got much else to hand to try these in, that yet fully-worked, so I then thought of using the Dataman Programmer's IC-Test function
- That's built in to the (Elnec) Dataman 'Pro' versions software (without having t resort to old DOS? software on my old Dataman 48XP), and does most 74 TTL & CMOS versions (+ 4000-series and also an 8255 / looks like you might be able to add your own test vectors?)
And this confirmed that 3 of my 9off 7445's were actually faulty!


So I then thought it would actually be a better idea, to instead properly-test each of the logic IC's from my Introkit on the Dataman Pro, as all of these (or pinout / function equiv's) were in its library - apart from the 81LS95 (But I had bought some 74LS795 equiv's) plus the 74L86 oddity (that I now had some spares of).
All of the supported IC's tested OK, so I put them back-in and swapped the 81LS95 to a 74LS795 & the 74L86 to another.
And I found the Introkit now worked! (For first time in > 27years I'd had it).
Swapping these two IC's back, confirmed the orignal 81LS95 was OK, but the original DM74L86N had been at faulty!


I would have replied about this success to a more Introkit-titled thread, but these were about the only references to it I could really find on here:
Inc. one I started: https://www.vintage-radio.net/forum/showthread.php?t=195775
Or maybe: https://www.vintage-radio.net/forum/showthread.php?p=1572246
With details on NS re-labelled calculator Keyboard-display: https://www.vintage-radio.net/forum/showthread.php?t=159537

But unfortunately, these are now all closed, so I'd have to request to re-open / start a new one.

[SiriusHardware]

There was someone else whose IC tester rejected all of their 7445s (in their case used in an MK14 replica) - unfortunately I can't recall who or where. I remember thinking at the time that it was surely the tester and not the ICs which were the problem.

Re-opening a relevant but closed old thread to continue on the same subject isn't usually a problem, normally a polite request to the mods via the 'report' (red triangle) symbol on the last post will get it re-opened for you.

Congratulations on getting it working, anyway. Another one saved.

[ortek_service]

Well I suppose they could have been unlucky and all the 7445's they'd bought were actually faulty (maybe even fake, as they are now often > £1 each) - especially if bout from China etc. I know Chris has often had problems - especially with EPROM's.
I don't seem to have had too many problems - all 10off of the HD74LS367's (that I'd ordered at the same time and was delivered in combined outer package with these to save postage, from but different Aliexpress sellers), which I needed 4off for the Softy-2, passed OK on the Dataman Pro(+) programmer's IC test.

I was quite-impressed by the Dataman Pro programmers's IC test, as it seemed to work flawlessly on all the 74 TTL IC's I tried on it. Although it just says 'passed' if OK and only shows you expected and result test vector (usually test vector 0, so not sure how many there are for each device) when it finds a failure. The three 7445's that failed did actually show different failures between the devices, with multiple L instead of H & H instead of L shown.
I also found devices survived OK, if I accidentally put one in the ZIF socket the wrong way round! (programmer was under edge of monitor, so it was a bit dark and I resorted to using a torch to ensure IC was in the ZIF socket correctly) - giving a 'H/W error' message and quitting out of currently-selected device. So it looks like it has over-current limit / detect on each pin.

I presume the Dataman 'Mempro' versions don't give you the IC test function, and you have to pay for the extra 'upgrade' (Serial ID protection IC?) to full version.

I wonder if it was the TL866 that someone else use to check their 7445's, as IIRC this also does IC-Testing (But I haven't got one yet, to try). And maybe, with the TL866 being rather cheaper than the Dataman Pro's, the support for fixing bugs isn't as good (In the past, I'd found support from Elnec, via Dataman, was good for resolving any issues with new devices they'd added for us - after buying the new 'custom' adaptors for it from them, of course)

I could have probably made a 74L86 to standard 74xx86 pinout adapter, in order to test that as well. But probably easier to just use a logic probe, in-circuit.


Google seems to do quite a good job at find all references to the Introkit on here, so didn't want to request re-opening just for this when I'd already previously mentioned it here.
Although I do now need to make a convenient PSU for it, to give the -12V as well as the main +5V supply, as lots of potential for making expensive mistakes with using two (isolated) bench PSU outputs, having to connect '+' on one to '-' on the other. I'm thinking a 5V to (isolated) 12V DC-DC would be safer, to run it all off a single +5V supply as Replica softy-2 has (with the now on-board 5V to 30V boost DC-DC converter).
However, I noticed the -12V supply current was just over 100mA!, so need at least a 2W version (When I think that many of the ones I've got are <=1W).

[SiriusHardware]

I presume the Dataman 'Mempro' versions don't give you the IC test function

To be honest I haven't noticed, but I will check that out this evening in the interests of science. I have two DOS based programmers, my Hi-Lo systems All07A and an ICE-Tech MM1000E, both of which do have that capability. However, it would be a good idea to find out if the Mempro does it too because it would be a lot more convenient to have it on a programmer whose software runs under current versions of Windows.

If not, it would be one more incentive to pay for the outrageously expensive plug-in dongle which 'converts' the Mempro to a 40Pro.

[factory]

Having used an older Sunshine variant of the Hi-Lo with Expro software at work, I wasn't that impressed by the IC test function, it can only really tell you the pinout & basic function is OK, it can't tell the difference between output types.
If you have a pin compatible relabelled fake it will pass them too. Hopefully the Dataman will do better.

P.S. The DM74L86N is a National Semi part, old Nat Semi parts are well known for failing.

David

[SiriusHardware]

Oh yes, in most cases the IC test facility will only give you a working / not working indication. Bear in mind that when those old programmers were made the idea that some day, someone might go to a lot of trouble to 'dress up' an LS IC as some older variant would not even have entered anyone's head.

In most cases it is enough to know whether the IC is working according to its expected truth table, or not.

The DM74L86N is a National Semi part, old Nat Semi parts are well known for failing

As the 'National Introkit' was a microprocessor trainer kit from National Semiconductor, it is highly likely that most if not all of the devices supplied in it were NS. Most (but interestingly not all) of the ICs in my original MK14 are NS devices, unsurprisingly as Science Of Cambridge struck a deal with NS for them to supply all the bits.

[ortek_service]

Originally Posted by factory

Having used an older Sunshine variant of the Hi-Lo with Expro software at work, I wasn't that impressed by the IC test function, it can only really tell you the pinout & basic function is OK, it can't tell the difference between output types.
If you have a pin compatible relabelled fake it will pass them too. Hopefully the Dataman will do better.
>>
David

Originally Posted by SiriusHardware

Oh yes, in most cases the IC test facility will only give you a working / not working indication. Bear in mind that when those old programmers were made the idea that some day, someone might go to a lot of trouble to 'dress up' an LS IC as some older variant would not even have entered anyone's head.

In most cases it is enough to know whether the IC is working according to its expected truth table, or not.
>>
>>

Well co-incidentally (and after saying I hadn't had much trouble with receiving faulty/fake IC's from China) I got 10off 'SN74LS73AN' marked Dual JK Flip-Flop IC's yesterday (supplied loose in a bit of standard bubble-wrap, so a few bent pins and maybe not great for ESD even if these were 'CMOS') - after ordering some quite cheap off Aliexpress as I hadn't got any spare LS versions that the Softy-2 uses.
So I thought 'd give them a quick test in that, in place of the current DM74LS73AN that Chris had given me, and which had been working OK.
I then discovered that with any of the ten of these 'SN74LS73AN' IC's fitted, there was now no longer any display (or at least no sync to it, that old monitor's V.Hold / H.Hold would lock-to). However, moving my finger over the top corner of it, often made these work OK!

I did then try them all on a Dataman (4)48Pro+ using the Test IC by Test Vectors, and choosing the '74 TTL' (rather than '74 CMOS') library and '7473'. And they all passed repeatedly passed that test (with no finger over it!), as did the originally-used DM74LS73AN.


After some 'scoping / DVM-measuring on the Softy-2 (with each-make of the two marked-makes in turn fitted, and studying the schematic), I discovered that on the Softy-2, the pins 2 nCLR for Flip-Flop1), 3 (K for Flip-Flop1) & 10 (K for Flip-Flop2) inputs have been left unconnected - but should normally effectively be pulled-high by TTL's input circuitry (which requires sinking of a moderate current out of 'input' in order to pull it low).

However these inputs all measured 0V (into 10M DVM I/P) when using the 'SN74LS73AN' marked IC's.
And shorting these to ground via a milliammeter gave no change in meter reading (down to 10uA resolution of it on mA-range).
But shorting input pins 2 (nCLR1) & 3 (K1) to pin 4 (oddly Vcc! on the 'non-standard pinout' 74xx73 - with power & ground in the centre similar to certain 74Lxx devices), would make them all work OK. Pin 10 (K Input to 2nd Flip-Flop)'s state didn't seem to be important to make it work in the Softy-2.

Whereas, with the original genuine 'DM74LS74AN' I'd been using, fitted, the voltages (into DVM's 10M I/P-Z) on the unconnected inputs were:
Pin2: 1.74V,
Pin3: 1.6V,
Pin10: 0.85V (Strangely lower than Pins 4&5 but all a bit lower than expected if LS TTL should have a fairly-good internal pull-up)
And shorting these to ground via milliammeter gave 460uA (pin2) / 180uA (pins 3 & 10)
(With shorting pin 2 or pin 3 to ground, stopping it working - so showing these were normally high).


I also noticed that the OK 'DM74LS73' outputs were going upto 4V on this (as expected for TTL) .
But on the 'SN74LS73AN' ones, the outputs went up to around the full 5.0V positive rail supply voltage.

So I think these 'SN74LS73AN' marked IC's could actually be 74HC73's! as seem to have 'CMOS' outputs, and inputs.
(All have the same markings inc. batch-codes, laser? 'engraved' quite deep into top surface. And no obvious 'black-topping' of previous markings.)

I did also try some known originally genuine old-stock 7473's in the Softy-2.
And 7473's worked OK in the Softy-2, with inputs high and able to source a fair amount of current (>=1mA) when pulled-low.

And I also tried some known-genuine NOS, still in carriers 74HCT73's, in the Softy-2. But these 74HCT73's didn't work properly in the Softy-2, and had the same 'floating' input issue I'd been finding with the suspect ones I'd received - Although not quite as bad.
Whereas I always thought 74HCT devices were basically 74HC devices, but with lower input-threshold voltages and also pull-ups on their inputs? to make them work with 74LS outputs that only just about went to 2/3*Vcc logic-high threshold.


I did at least have rather-more joy with an INS8060 SC/MP-II I received only a day after buying it from a certain UK supplier (after having a 'best offer' several pounds below the buy it now price accepted). And this worked OK, when tested in the Softy-2.

[factory]

I'm really not surprized at the fake 74LS73, it's just about still made, brand new LS series are a few £ each from Farnell/Mouser, HC series are below £1, hence the relabelled HC ICs.

As far as I know, the HCT just has compatible voltage levels with LS, I've been using these in SMT form to construct replacements for custom HP ICs from the late 60s.

And you confirmed what I thought about the logic IC test function of the Dataman, just a basic check as the various Sunshine/Hi-Lo/Expro branded programmers, to properly check at the various voltage levels would require more complex circuitry, probably a DAC & ADC input/output to properly measure each pin. Or testing by hand.

David

[SiriusHardware]

I have to admit that's why I initially sought an SC/MP II for my MK14E from among the group here, as I had absolutely no faith that one coming from anywhere else would be 'real'.

(The one kindly supplied by Realtime is working well, I'm glad to say).