Tesla Programmer

[ChrisOddy]

Last year I started looking at the problem of programming Tesla PROMs and seeing if I could use the existing Acorn PROM Programmer (intended for National Semiconductor PROMs).

Well it turned out not as the programming method is effectively reversed, data lines are pulled low rather than high. It proved difficult to adapt the design to cope with both makes of PROM due to a lack of spare control lines. The easy solution was to build a second programmer card. After a 6 months delay (caused by other project diversions !) I've finally gotten around to finishing it off. I've updated my PROMER2 software to PROMER3 which works with either or both Programmers.

I've put a full description and write up here including the source code: http://www.theoddys.com/acorn/replic...mer_board.html

Chris

[Slothie]

Great work! Funnily enough last week I dug out the PCBs I had made and components for my Tesla PROM programmer that was discussed on this forum a couple of years ago and got shelved due to personal issues. I have a translated data sheet for these Tesla PROMs that I made, if anyones interested, but the details are essentially the same as in Chtis's documentation. It's good to see the algorithm I translated from Czech using online tools was correct, as confirmed by Chris's document. I am going to put together the programmer and see if I can get it working. I have a pair of Acorn 1 PCBs I bought off eBay 2 years ago and I think I need to at least start thinking about getting the parts and tools I need to put it together.

[SiriusHardware]

Thanks Chris, (and Ian, nice to see you around again).

Chris, I note your observation (which has been noted before, I think, by circuitryboy and others) about the similarity between the programming regime for the Tesla devices and the equivalent Texas SN... series devices. That originally gave me some hope that MH74S571s could be programmed by any programmer which supported the SN74S571... but ... Texas seem to have stopped one device lower than that, at the SN74S287. There does not ever seem to have been an SN74S571, so there is no direct Texas equivalent for the 571 which is a pity.

However it does make me wonder if a device programmer which supports the SN74S287 could programme a Tesla MH74S571 in two passes if the MH74S571 was placed in an adaptor which allowed the high address bit to be manually pulled high or low.

I take it the PROMER software runs on an Acorn System 1, which not many people are likely to have. If so, Ian, we still need your alternative based on a more generic, readily available and cheap controller like an Arduino.

[joebog1]

It is great work!!. A beautifully laid out article. Every part of it is explained!!.
Superb effort.

Joe

[SiriusHardware]

So Chris, do you plan to sell these PCBs? I know you do sell your MK14 replica PCBs, in fact, one of the regular MK14 contributors here uses one. I didn't see any indication anywhere in your link that these Acorn System 1 Tesla programming PCBs are for sale.

[ChrisOddy]

Yes I do sell the PCB's but you'll need to go to the For Sale page, I don't link from the individual board pages : http://www.theoddys.com/acorn/replic...mer_board.html

The PROMER program will not run on a System 1, it needs a VDU, Keyboard etc. so you need at least a System 2 and really a disk based System 3 as I wouldn't want to load it from tape !

[TonyDuell]

I must look into this, I suspect it would run on one of my System 4 machines (twin disk drives). The one that it won't run on has the 6809 CPU board, you see...

[SiriusHardware]

I was thinking possibly of removing (or not fitting) everything east of the 8255 and instead controlling the 8255 with a current microcontroller PCB like an Arduino or a Raspberry Pi. (level shifting would be required for the latter, yes, but its in-built mass storage / file system would make the creation, management and selection of programming files so much easier).

The availability of a PCB does make this a more appealing prospect.

[ortek_service]

Yes, Chris, it was good to hear you'd now got this all working OK.

I recall initially 'scoping the Dataman-Pro (Elnec re-badge, and one of very few that supported the Tesla PROM's), to verify its programming method against the somewhat vague info on programming these.
And having to a bit cunning with shunt-diode + pull-up, to have level low-enough to fool it into attempting to program a location, but high-enough to be able to see how long it then held it hard to ground for
- As had initially discovered that it was being 'intelligent', and reading the bits first, to decide whether they needed to be programmed (unlike most EPROM programmers, that just always program, then verify after each pulse).

It was also surprising that Elnec had opted to just have one long 45ms pulse, rather than trying multiple shorter ones, for quicker speed / less heaing, and only one attempt at a location (before completely aborting prog. sequence. - with retrying option going right back to start - although fortunately it then skipped all those locations that had previously programmed OK and read correct)
It seems they'd done something similar for NS-Proms, as I recall previously having one that kept failing at many locations, so had to repeat the sequence many ten's of time, till it eventually-programmed. So maybe their programmability declines with age / that was from a rogue batch.

I never found a common programmer that supported the TI SN74Sxxx 4bit PROM's (Only 8bit ones) - Neither original (Advantech-design) Dataman 48 or the (Elnec-design) Dataman 48-Pro ones (which both do NS ones), to see what they did. So even if you could still get hold of TI PROM's, programming them could be difficult (although it seems Tesla algorithm is so similar, then can use one for that).
It doesn't seem that Tesla just copied TI devices, as there never was a SN74S571, only the '287. Although they did use virtually the same programming recommendations / odd inverted-programming arrangement (With the NS ones seemingly a bit more conventional, even if higher 10V etc. voltages were still needed to be applied).

As NS ones are now getting very hard to obtain, with previous NOS's being exhausted, then Tesla ones seem to be only alternative - especially as 8bit PROM's / parallel (E)EPROM's all tend to be in packages that are too large to make neat adapters for.

[SiriusHardware]

Hi OS, nice to see you are still around. I think one of my two programmers here may support Texas SN-series BPROMS but I will have to look to see if that includes the SN74S287.

I have to say that we all appreciate the investigative work by you and Chris, also circuitryboy who worked on the problem independently - I suppose no matter how careful you are, when you are working with one-shot devices you must have to waste a few while fine-tuning things so there would have been a cost in terms of parts as well as time, Although I suppose you could minimise that a little by only trying to programme individual bits at a time in the early stages of testing, rather than the whole device.

[ChrisOddy]

I only wasted one Tesla chip ! My test software only programmed one byte or even bit at a time. Once conifdent I then took the plunge and programmed Mk14 and Acorn System 1 chips and they were all fine.

[SiriusHardware]

That just proves you are the Better Man. I no doubt would have gone through about a dozen before I got it right.

Fear Of Getting It Wrong would have made me keep getting it wrong.

[SiriusHardware]

Just giving this thread a nudge as I have just ordered a PCB for one of these from Chris with a view to mating it with a simple / cheap modern micro board. Some parts I already have, some I will have to scrape together but I hope eventually to interface it to an Arduino or Pi Zero to allow programming of the 'problem' Tesla MH74S571s by that method.

[Mark1960]

When Vcc is raised to 10.5 volts does the output voltage of unprogrammed data pins also increase to 10.5v? This might be clamped to 5v by the esd protection diodes internal to the 8255, so just wondered if current limit resistors should be added on the 8255 inputs.

[SiriusHardware]

I'm actually not even sure if I will use the 8255, I might go straight from the (as yet undecided) microcontroller to the power control and data in / out lines of the programmer.

That issue of possible high voltage out from pins which are not being programmed is something to look out for though.

[Slothie]

I must get back to mine, I had got the pinouts of my mosfets wrong and blew them up but have since received the replacements. Need to get my mojo going again!

[ortek_service]

Originally Posted by Mark1960

When Vcc is raised to 10.5 volts does the output voltage of unprogrammed data pins also increase to 10.5v? This might be clamped to 5v by the esd protection diodes internal to the 8255, so just wondered if current limit resistors should be added on the 8255 inputs.

IIRC, Most PROM's default to logic-0 when unprogrammed / fuse not 'blown' - The opposite to EPROM's that default to all 1's when erased.

However, I'm not sure what happens once programmed, and Vcc is still at 10.5V.
- Especially as programming on these Tesla ones seems to be inverted, with having to pull data-lines to 0V to blow the fuse to make output '1'. So if the current-sink transistor is switched-off, whilst Vcc is still at 10.5V, then maybe the data-line will go to that for a while with standard '3-state' output (Rather than Open-collector) types.
However, if these PROM's are like other 74(S) TTL, then output source current may be rather low, and so 8255 Input protection diodes (If present on non CMOS types?) should clamp it without lifting it's supply rail too much.

From bitter experience of having two pairs of 8255's fail, when designing my own Acorn BBC computer-controlled universal EPROM programmer > 30yrs ago, I discovered that having too much capacitance on the EPROM Vpp pin and simply relay-switching this back to the +5V rail, resulted in that being lifted (as 7805's etc can only-source) to a high-voltage for a short period, which killed the 8255's!
- So I re-designed it, to have Vpp supply totally-independent of the main +5V supply to the control IC's, and have Vpp settable to +5V by itself.

I'll mention it to Chris, to see if he's looked into the possibility of the above issue / may be able to do some 'scope checks whilst programming a test IC.

[Mark1960]

I think I was reading the wrong programming algorithm, but still worth checking what happens to the pins for bits that are not being programmed when Vcc is raised to 10.5v. The transistors pulling the pins to ground need to sink 150mA, so if the outputs are pulled up to 10.5v when Vcc is increased this could be quite risky for the 8255.

I’ve heard that 8255s are in short supply recently, not sure if that is another issue.

[ortek_service]

Originally Posted by Mark1960

>>
>>
I’ve heard that 8255s are in short supply recently, not sure if that is another issue.

I'd not heard of any problems with getting 8255's, but they are getting on a bit and Intel (Maybe one of the few manufacturers of it) may well have discontinued it by now, and so only stocked by obsolete parts specialists.
- However, the 82C55 CMOS version is still shown as active by Renesas (Who took over Intersil one of previous manufacturers for it)
https://www.renesas.com/us/en/produc...eral-interface

But I'm not sure if 82C55 would always be a direct-fit substitute.

Looking on Octopart, it seems that Rochester (who specialise in buying up old stocks / may also buy up wafer-masks and get re-made with a different prefix like Lansdale did with old Motorola IC's) are now often listed as the manufacturer. But the MR8255 that Rochester list / are also available via Digi-key marketplace are now a rather high £80!
https://www.digikey.co.uk/en/product...aIBxgKyJAXQL5A

But in the UK at least, Cricklewood Electronics are still listing these at a bit more reasonable £12.50+VAT. And a little cheaper (£10) at Silicon-Ark:
https://www.silicon-ark.co.uk/index....ch&search=8255
- Still a bit more than what they use to cost a few decades ago, and rather cheaper direct from the Far East etc, (As little as 50p for (Renesas / ex-NEC) uPD8255 and also 82C55's on Aliexpress) if you take a gamble on whether these have been faked yet.
Although probably a few other smaller components companies still with NOS's that have their own website / are on various online marketplace / auction sites.

Luckily, I think I've still got a few NOS spares / pulls around.
IIRC, they were used on some old PC ISA I/O cards, that I think I've got, but these are probably now worth keeping intact rather than stripping for spares.

[SiriusHardware]

I happen to have an 8255 or three, but all of the above is convincing me that I should probably go directly to / from the programmer's control hardware and 4-bit data buses and miss out the 8255.

Chris's design seems to have stayed (as far as possible) quite close to that of the original DM74S571 programmer, for example in its use of BFY50 medium power transistors - the design of that programmer has been messed with as little as possible to achieve an MH74S571 programmer which still looks appropriately retro.

In practice, it may be easier to find some T0126 medium power NPN transistors which will work just as well.

I did have a tube of around 10 74C906 somewhere, of course I temporarily don't know where that is but now I have the incentive to find them...