UK Vintage Radio Repair and Restoration Powered By Google Custom Search Vintage Radio and TV Service Data

Go Back   UK Vintage Radio Repair and Restoration Discussion Forum > Specific Vintage Equipment > Vintage Computers

Notices

Vintage Computers Any vintage computer systems, calculators, video games etc., but with an emphasis on 1980s and earlier equipment.

Closed Thread
 
Thread Tools
Old 9th Jan 2021, 3:48 pm   #81
SiriusHardware
Dekatron
 
Join Date: Aug 2011
Location: Newcastle, Tyne and Wear, UK.
Posts: 11,483
Default Re: Softy 1

Hi Owen ('Graham' here by the way, it's no secret but I have been 'Sirius' by default here for a long time now.)

I think even a 16-bit additive checksum is better than nothing. Granted, you could have a case where byte 0 has a bit stuck low when it should be high and byte 4 has the same bit stuck high which should be low and you'd then still get the same additive checksum so it isn't foolproof, but generally speaking if there is a 16 bit checksum written on a label and you read it out and don't get the same checksum, that is enough to tell you there is a probably a problem with the device content.

Most good hex editors hedge their bets by giving you the checksum calculated in several different ways so if the checksum on my device matched just one of them, I would be inclined to believe the device content was OK unless I was eventually forced back to the view that it might not be.
SiriusHardware is online now  
Old 9th Jan 2021, 3:58 pm   #82
Slothie
Octode
 
Join Date: Apr 2018
Location: Newbury, Berkshire, UK.
Posts: 1,287
Default Re: Softy 1

Quote:
Originally Posted by ortek_service View Post
And is something this is trying to resolve: https://www.baldengineer.com/bit-pre...ith-kicad.html
I never knew this guy was doing this. I've watched some of his videos and they're good. I wonder if he needs an MK14 schematic.... The problem is the schematics I have I'm not entirely sure what "issue" they are (well, aside from my own Issue VI ) But I think I could go back through my Git repository and find the probably Issue IV schematic I based mine on and correct it so the component ids are correct.
Slothie is offline  
Old 9th Jan 2021, 4:41 pm   #83
ortek_service
Octode
 
ortek_service's Avatar
 
Join Date: May 2018
Location: Northampton, Northamptonshire, UK.
Posts: 1,394
Default Re: Softy 1

Quote:
Originally Posted by SiriusHardware View Post
Hi Owen ('Graham' here by the way, it's no secret but I have been 'Sirius' by default here for a long time now.)

I think even a 16-bit additive checksum is better than nothing. Granted, you could have a case where byte 0 has a bit stuck low when it should be high and byte 4 has the same bit stuck high which should be low and you'd then still get the same additive checksum so it isn't foolproof, but generally speaking if there is a 16 bit checksum written on a label and you read it out and don't get the same checksum, that is enough to tell you there is a probably a problem with the device content.

Most good hex editors hedge their bets by giving you the checksum calculated in several different ways so if the checksum on my device matched just one of them, I would be inclined to believe the device content was OK unless I was eventually forced back to the view that it might not be.

Hi Graham
- Yes quite a few Hex-Editors do this / give you option of which one to display. Just have to make sure you avoid an old version of HexWorkshop (v3.1?) that had a bug in the calculation of one of these despite it being one you had to buy a licence code online for.

But there's lots of free ones out there that do as good a job
- But can never remember which one has what features like Raw disk / memory card access, with so many similar names: eg. HexWorks(32?), HxD, HHD, Hexplorer, WinHex etc.
Plus I've also used some Dataman etc. Programmer software (often in demo mode etc without programmer) to convert formats, swap Nibbles, load images to certain offsets for DIY 'Multi-slot' cartridge EPROM's etc as not that many Hex Editors did all of this.
Although most are probably better than the first PSEDit? DOS? one I used on PC's, after previously using BBC BASIC to to work on EPROM Images.

Owen
ortek_service is offline  
Old 9th Jan 2021, 4:48 pm   #84
ortek_service
Octode
 
ortek_service's Avatar
 
Join Date: May 2018
Location: Northampton, Northamptonshire, UK.
Posts: 1,394
Default Re: Softy 1

Quote:
Originally Posted by Slothie View Post
Quote:
Originally Posted by ortek_service View Post
And is something this is trying to resolve: https://www.baldengineer.com/bit-pre...ith-kicad.html
I never knew this guy was doing this. I've watched some of his videos and they're good. I wonder if he needs an MK14 schematic.... The problem is the schematics I have I'm not entirely sure what "issue" they are (well, aside from my own Issue VI ) But I think I could go back through my Git repository and find the probably Issue IV schematic I based mine on and correct it so the component ids are correct.
Yes, I discovered by accident (Although there's also a post on Hackaday about it). It seems it's mostly Apple ones so far, that he was mainly interested in - He did have a 'wants list' on there / another website he had / maybe the github area.
But it was currently rather-lacking that many popular UK ones - So may be an incentive to learn to use KiCAD (that's been rapidly gaining popularity as now one of the few completely-unlimited free versions, plus plugins are available to convert to/from other formats - Although probably still want to create parts library in it / use a common one for these)
ortek_service is offline  
Old 9th Jan 2021, 5:30 pm   #85
ortek_service
Octode
 
ortek_service's Avatar
 
Join Date: May 2018
Location: Northampton, Northamptonshire, UK.
Posts: 1,394
Default Re: Softy 1

I've now attached an OCR'd version of the original 'Lower-Res' scan of the manual (Inc. non-reduced file-size one, in case doing this has made quality worse)

And this is now a link to an OCR'd version of the later 'Hi-Res' re-scanned version: https://drive.google.com/file/d/1NM2...ew?usp=sharing

- Without reducing file size, as this made some text at the bottom that didn't OCR, unreadable. (But hadn't been a problem on 'Lower-Res' original, due to it not being as close to edge of scanner on these pages).

All 3 versions should also be available here: https://drive.google.com/drive/folde...ky?usp=sharing


If anyone actually has / has-seen mention of the 'Conversion Card' for it, that would be interesting to see details of this, as I've never found a reference to it in this manual / anywhere else. But Chris thought he seen it mentioned on here.

Owen
ortek_service is offline  
Old 9th Jan 2021, 8:33 pm   #86
circuitryboy
Pentode
 
circuitryboy's Avatar
 
Join Date: Jun 2012
Location: Bristol, UK.
Posts: 115
Default Re: Softy 1

Over the years I have accumulated a hoard of SC/MP related data/pics from the Web.
These items may be of interest.
I include the Softy 1 as later PROM version and serial scratched on underside. The distressed specimen we saw recently is 492.
Attached Files
File Type: zip Softy 381.zip (325.4 KB, 69 views)
File Type: zip Softy 2.zip (3.25 MB, 133 views)
circuitryboy is offline  
Old 10th Jan 2021, 3:59 pm   #87
TonyDuell
Dekatron
 
Join Date: Jun 2015
Location: Biggin Hill, London, UK.
Posts: 5,190
Default Re: Softy 1

I have now dumped the EPROM from my Softy. It looks 'sane' (not all 0's or FF's for instance). The programmer (An odd thing called a 'Ferret' claims the checksum is D06D. The EPROM starts with 08 C4 06 36 C4 70 32.... The progammer saved the data in Intel Hex, I think the later version with longer addresses.

I'd rather not make it public until somebody can check this at least could be a valid dump of the firmware (in case I pressed the wrong buttons on the programmer). After that I am of course happy for it to be public. So could one of the people collecting the firmware please contact me by PM giving their e-mail address and I'll send it to them.
TonyDuell is online now  
Old 10th Jan 2021, 6:34 pm   #88
SiriusHardware
Dekatron
 
Join Date: Aug 2011
Location: Newcastle, Tyne and Wear, UK.
Posts: 11,483
Default Re: Softy 1

Those first few bytes are certainly valid SC/MP code.

You could perhaps programme another EPROM (if you have one) with the code you've read out and see if it works in your own Softy? If the copy works OK then you can at least be assured you read it correctly.

There might still be a 'bad' bit lurking in a little used subroutine somewhere but you can only really find something like that by exercising every single function of the unit, or by meticulous disassembly and tracing of the code, as Owen's friend Chris seems to like to do.
SiriusHardware is online now  
Old 11th Jan 2021, 9:48 am   #89
ortek_service
Octode
 
ortek_service's Avatar
 
Join Date: May 2018
Location: Northampton, Northamptonshire, UK.
Posts: 1,394
Default Re: Softy 1

Quote:
Originally Posted by circuitryboy View Post
Over the years I have accumulated a hoard of SC/MP related data/pics from the Web.
These items may be of interest.
I include the Softy 1 as later PROM version and serial scratched on underside. The distressed specimen we saw recently is 492.
Thanks for these - It's nice to see what's inside an S2 (or at least 2, by the looks of the different pictures) at last. And it looks like the basic design was the same as the original - with many of the same IC's, but with a couple more 2114's added for another 1KB.

So the main changes from the original are ptobably a change of PCB form-factor and ditching the 3-rail support for both the programming (and also its own EPROM by the looks of it, as there's a 2716 fitted). And this could have simplified things a bit (although maybe more for the external PSU rails)

The Original Softy board is also interesting, with the 'DIY' add-on 2516 / 2716 switch (Also with '08' next to it, but switch only looks 2 position and not enough wires to switch between 3-rail and single rail - That all 2516's are)

And 381 (v3.81?) label on EPROM also indicates there were several versions and that these were original labels (When I first saw the Bygonebytes one, I though it may have been added much later with a thermal labeller machine)

I presume you never had access to any of these, to copy the EPROM?
- S2 info seems even harder to come by, with maybe no official circuits released originally. Although, I've not found a copy of the manual so far, only a photo of cover on this record of an online auction listing:

https://www.worthpoint.com/worthoped...mmer-307527523

(Strangely with a manual for the original Softy, as well, from the photo on that one. And appears to be a bit more professionally=produced than the one made available via bygonebytes - Which it seems does have an earlier-version of the firmware, compared to one in your photo).

The only other Softy-2 I'd found that someone actually had, was in this picture (But no other details, than they got it in 2017 from an auction)

http://www.computermuseum.org.uk/mac...n_softy_2.html
ortek_service is offline  
Old 11th Jan 2021, 10:51 am   #90
ortek_service
Octode
 
ortek_service's Avatar
 
Join Date: May 2018
Location: Northampton, Northamptonshire, UK.
Posts: 1,394
Default Re: Softy 1

Quote:
Originally Posted by TonyDuell View Post
I have now dumped the EPROM from my Softy. It looks 'sane' (not all 0's or FF's for instance). The programmer (An odd thing called a 'Ferret' claims the checksum is D06D. The EPROM starts with 08 C4 06 36 C4 70 32.... The progammer saved the data in Intel Hex, I think the later version with longer addresses.

I'd rather not make it public until somebody can check this at least could be a valid dump of the firmware (in case I pressed the wrong buttons on the programmer). After that I am of course happy for it to be public. So could one of the people collecting the firmware please contact me by PM giving their e-mail address and I'll send it to them.
Thanks for doing this. I've loaded the .bin from Chris's Softy webpage, into a hex editor and the first 8 bytes are 08 C4 06 36 C4 70 32 CE - As they are, if you have a look in the softy1_304.lst listing file, that Chris has on the webpage / direct-link here: http://81.174.146.201/acorn/PROMs%20...softy1_304.lst
So this matches the first few bytes you supplied here.

I'm not sure if you were viewing the contents directly, or you were looking at the Intel Hex file's content it saved (In which case there maybe some extra bytes before / after the data, for the structure of the file).

The first Hex-Editor I tried (ICY Hexplorer), shows lots of different Checksums - but all in decimal, with no option to change to Hex!
And converting manually to hex, none of the Checksum 32 / CRC16 / CRC16CCITT match your D06Dh checksum.

However, loading the file into my old Dataman-48 software (with 2716 device, as it doesn't support 2708/2758 etc. but with top half-zero'd / ignored), it gives a (32bit) Checksum of 0001D06Dh.
So assuming this is just an simple additive one, then it would just wrap-round for a 16bit version and the top 16bits can be ignored. And the bottom 16bits do match what you've got.

Therefore it does appear yours is also version '304' - Did you have a label on it? And it seems there aren't any corruptions in yours / Chris's readout one.

But if you'd like us to double-check your readout, then can send it to Chris, as he saw your reply on here that you'd now readout your and asked me to see if I can get the readout for him as he's not registered on here yet
- He gives his email contact on his home page: http://81.174.146.201/acorn/index.html
I'll also try & work out how to PM you with my email address, as I've not done this on here before.


It seems Bygonebytes has now received the EPROM & PROM back again, as Chris has told me that he has now powered up his SOFTY he had from new, but its not fully working (which I think he may have found, before removing the EPROM & PROM) And has video but garbage on the display, so sounds like the CPU clock is running as the video needs that but its not executing anything.

- I'm not sure if you've managed to power your up recently, to see if it still works.

Chris's replica boards are almost finished at JLCPCB, so he'll hopefully have a working one himself fairly-soon.
ortek_service is offline  
Old 11th Jan 2021, 11:43 am   #91
ortek_service
Octode
 
ortek_service's Avatar
 
Join Date: May 2018
Location: Northampton, Northamptonshire, UK.
Posts: 1,394
Default Re: Softy 1

And I've just tried loading Chris's readout into HxD Hexeditor, and that also gives Checksum-16 as D06Dh (With Checksum-32 as 0001D06Dh)
So it does look like yours exactly matches Chris's readout, but I've PM'd you my email if you want to send me the full-file to compare.
ortek_service is offline  
Old 11th Jan 2021, 3:14 pm   #92
SiriusHardware
Dekatron
 
Join Date: Aug 2011
Location: Newcastle, Tyne and Wear, UK.
Posts: 11,483
Default Re: Softy 1

That's probably the best that you could have hoped for really, for both of these rare survivors to have the same firmware version so they could be successfully verified against each other. It also means that whatever the problem with Chris's friend's Softy is, it probably isn't the firmware.

How about the CG PROMs? Would be interesting to know if they are identical as well.
SiriusHardware is online now  
Old 11th Jan 2021, 4:47 pm   #93
TonyDuell
Dekatron
 
Join Date: Jun 2015
Location: Biggin Hill, London, UK.
Posts: 5,190
Default Re: Softy 1

I've hopefully sent my IntelHex dump of the Softy firmware to ortek_service so he can verify it.

At least if it agrees with the image from another Softy then it's likely both are good.
TonyDuell is online now  
Old 13th Jan 2021, 12:01 am   #94
ortek_service
Octode
 
ortek_service's Avatar
 
Join Date: May 2018
Location: Northampton, Northamptonshire, UK.
Posts: 1,394
Default Re: Softy 1

Quote:
Originally Posted by TonyDuell View Post
I've hopefully sent my IntelHex dump of the Softy firmware to ortek_service so he can verify it.

At least if it agrees with the image from another Softy then it's likely both are good.

Thanks for sending this - I've just found it went to my old email address (so now updated my forum account details)

And I've sent an email reply, saying I confirm it is identical - based on using 3 different file compares (including Windows command line fc /b binary-compare)
(Once I'd converted Tony Duell's SOFTY.HEX programmer saved-file, to Raw Binary - as Chris's is in that format, and this makes comparison / checksumming much easier. I also checksummed it it in HxD, WinHex + Dataman-48's software and I got the same Checksum-16bit of D06Dh).

I did find an odd 50spaces at start of the first line of Tony Duell's SOFTY.HEX programmer saved-file, which upset many Hex editor etc. converters.
But manually removing these / loading into original Dataman-48's software and forcing to Intel-Hex instead of auto-detect resolved this.

And I also found that the original Dataman-48's software added a short special header to its saved Intel-Hex file, that wasn't in the originally-supplied one. But this didn't cause any problems with loading into various Hex Editors
(But would probably upset file-compares on .hex format)
ortek_service is offline  
Old 13th Jan 2021, 12:09 am   #95
ortek_service
Octode
 
ortek_service's Avatar
 
Join Date: May 2018
Location: Northampton, Northamptonshire, UK.
Posts: 1,394
Default Re: Softy 1

Quote:
Originally Posted by SiriusHardware View Post
That's probably the best that you could have hoped for really, for both of these rare survivors to have the same firmware version so they could be successfully verified against each other. It also means that whatever the problem with Chris's friend's Softy is, it probably isn't the firmware.

How about the CG PROMs? Would be interesting to know if they are identical as well.
Well the CG PROM, being fusible-link, is unlikely to have any odd bit / nibble failures with age, as once links are fused then shouldn't be able to re-connect and don't have EPROM's voltage-charge to leak away.

But Chris did write a Python script to process his CG PROM readout into actual character pixel patterns, to ensure this had worked OK.
With results visible as ASCII-character pictures, in the results file he put on his webpage: http://81.174.146.201/acorn/PROMs%20...racter_rom.txt

- And this certainly looks OK.
ortek_service is offline  
Old 13th Jan 2021, 12:25 am   #96
SiriusHardware
Dekatron
 
Join Date: Aug 2011
Location: Newcastle, Tyne and Wear, UK.
Posts: 11,483
Default Re: Softy 1

There are stories about the links in fusible-link PROMs occasionally 'regrowing' in the same fashion as 'tin whiskers' grow inside old germanium transistors - whether it really happens to PROMs is obviously difficult to know but they do also just fail, in fact Buzby123 here had an original SOC DM74S571 fail on him just about a month ago. Working one minute, dead the next.

As you may have noticed, Bipolar PROMs run at disconcertingly high operating temperatures even when working normally. It can't do them much good in the long run.

As long as someone somewhere has the PROM contents held independently outside of an actual PROM, that's all good.
SiriusHardware is online now  
Old 13th Jan 2021, 1:10 am   #97
ortek_service
Octode
 
ortek_service's Avatar
 
Join Date: May 2018
Location: Northampton, Northamptonshire, UK.
Posts: 1,394
Default Re: Softy 1

Quote:
Originally Posted by SiriusHardware View Post
There are stories about the links in fusible-link PROMs occasionally 'regrowing' in the same fashion as 'tin whiskers' grow inside old germanium transistors - whether it really happens to PROMs is obviously difficult to know but they do also just fail, in fact Buzby123 here had an original SOC DM74S571 fail on him just about a month ago. Working one minute, dead the next.

As you may have noticed, Bipolar PROMs run at disconcertingly high operating temperatures even when working normally. It can't do them much good in the long run.

As long as someone somewhere has the PROM contents held independently outside of an actual PROM, that's all good.

I suppose that although it may appear that a fuse has re-connected, it's difficult to know whether that actually happened, without a detailed post-mortem on it.
I presume that once a fuse has 'blown', then that's it and you don't need to provide any more 'write' time to it like you would for EPROM's to ensure full-enough charge. So it might not be quite like tin-migration whiskers (one reason they added lead to solder, to prevent this) / germanium transistor short to case-shield problem.
And which also re-occurred in more recent years with a Maxim Satellite-tuner IC, that had a manufacturing fault - but ironically you could temporarily fix by heating it up with hot-air tool for a short while.


I did read about that short-lived after programming PROM, but not sure if it was found if link had appeared to grow-back - In which case, maybe reprogramming might have fixed it again.

I'd expect heat problems to affect the whole IC, rather than just the odd location. And I don't think they ever produced NMOS (or better still CMOS) PROM's with fusible-links (Maybe not fast enough originally , for some applications)
Although it seems Commodore / MOS had issues even with Mask-programmed ROM's (and added-Mask copy of a Philips Programmable-PLA)
which ran quite hot and failed rather sooner than expected.

Luckily, I imagine these Softy's haven't had too much use over the last 40years, so the PROM hasn't got cooked that much.
ortek_service is offline  
Old 13th Jan 2021, 1:55 am   #98
SiriusHardware
Dekatron
 
Join Date: Aug 2011
Location: Newcastle, Tyne and Wear, UK.
Posts: 11,483
Default Re: Softy 1

Quote:
Originally Posted by ortek_service View Post
I did read about that short-lived after programming PROM, but not sure if it was found if link had appeared to grow-back - In which case, maybe reprogramming might have fixed it again.
Not exactly short lived . It was an original Science of Cambridge programmed and supplied PROM. Buzby sent the pair to me for testing as he didn't have a known working machine to test them in, they worked fine for me but one of them died after he had run it for about one evening.

It was my hope that one of the fuses had just re-made in the mythical fashion already described and if so, it might be possible to blow it again. But when I got it back again, although the lower half of the device was still reading normally the whole upper half of the device was reading 'F', so I would guess possibly a failure of the internal addressing logic.

Last edited by SiriusHardware; 13th Jan 2021 at 2:10 am.
SiriusHardware is online now  
Old 13th Jan 2021, 4:23 pm   #99
TonyDuell
Dekatron
 
Join Date: Jun 2015
Location: Biggin Hill, London, UK.
Posts: 5,190
Default Re: Softy 1

As regards the hex file of my Softy EPROM :

I read out the 2708 using something called a Ferret. This is a combined RS232 breakout box, interface converter (RS232, current loop, parallel), strip printer and EPROM programmer (!). It can handle the 3-rail EPROMs and claims to be able to output IntelHex and many other formats.

I had to read the EPROM on that, send the IntelHex file over the RS232 port to an HP 95LX, then kermit it onto this PC. At any point spaces could have been added, but I suspect the Ferret outputs them anyway. The original IntelHex format (which dates back to the early 1970s, my MCS8i can handle it for example) was used with paper tape and is specifically specified to skip over 'extra' characters. A hex converter program that doesn't skip until it finds a colon is broken!

But you now have the data and it agrees with one of the other Softies.

Incdentally, there is no serial number scratched on my Softy PCB.
TonyDuell is online now  
Old 14th Jan 2021, 8:34 pm   #100
ortek_service
Octode
 
ortek_service's Avatar
 
Join Date: May 2018
Location: Northampton, Northamptonshire, UK.
Posts: 1,394
Default Re: Softy 1

Quote:
Originally Posted by SiriusHardware View Post
Quote:
Originally Posted by ortek_service View Post
I did read about that short-lived after programming PROM, but not sure if it was found if link had appeared to grow-back - In which case, maybe reprogramming might have fixed it again.
Not exactly short lived . It was an original Science of Cambridge programmed and supplied PROM. Buzby sent the pair to me for testing as he didn't have a known working machine to test them in, they worked fine for me but one of them died after he had run it for about one evening.

It was my hope that one of the fuses had just re-made in the mythical fashion already described and if so, it might be possible to blow it again. But when I got it back again, although the lower half of the device was still reading normally the whole upper half of the device was reading 'F', so I would guess possibly a failure of the internal addressing logic.

I hadn't realised it was an originally supplied-programmed IC - thinking it was a unused old-stock one that had recently been programmed.
And from PROM's I've seen, I'm sure 0 was blank-state (the opposite of EPROM's), but it might be manufacturer-dependent if TI/Tesla ones are programmed with Higher voltage applied to opposite Vcc - Datalines polarity compared to NS ones.
So it does look like it may have failed due to age / helped by a bit of heat.
- Probably not a Static-zap at least, with Bipolar type IC's...
ortek_service is offline  
Closed Thread

Thread Tools



All times are GMT +1. The time now is 8:06 pm.


All information and advice on this forum is subject to the WARNING AND DISCLAIMER located at https://www.vintage-radio.net/rules.html.
Failure to heed this warning may result in death or serious injury to yourself and/or others.


Powered by vBulletin®
Copyright ©2000 - 2024, vBulletin Solutions, Inc.
Copyright ©2002 - 2023, Paul Stenning.