Not CCS at all, but a Legacy PIC17C756(A) plea!

[aopg64]

Hi folks,

Sorry, this isn't a CCS item at all, but a plea for some vintage PICs to help me support a legacy product! So this is for all those guys who were using PICs in the 90's!

We have a legacy product that uses the PIC 17C756 or 17C756A 68-pin PLCC. This device isn't going to be redesigned, but the firmware keeps needing minor modifications.

For the Production devices we can still get the OTP part which is fine. I don't need any of them! Unfortunately this is what all the component searches turn up, despite alleged UV-erasability. The vendors just haven't correctly amended their descriptions of the part they are selling.

Up until a few weeks back I had 2 UV-erasable parts which I used for development. Unfortunately they both died a few weeks back and now I am unable to develop any more. My boss isn't keen on me wasting ~50 or so OTP parts to fix a serious bug but perversely is happy for me to spend several days so far already and maybe a few weeks in total trying to get an early 90's Picmaster emulator to work having dug out a mid-90's PC with an ISA bus to drive it. Yes really!

So I'm now using a 100MHz ISA bus PC to try and emulate a bug which I reckon won't emulate correctly at all!

I've survived developing with only 2 UV-erasable devices for the past 4-5 years, so all I need really is about 2 of these devices, though more would be preferable, and I can continue supporting this product for what is expected to be the last 3yrs of its life!

NB When I joined the company 10yrs ago they were 'just about to replace it', but never did and so we are in this sorry state.

So, does anybody have one or more of the following in the back of their component drawers? :

PIC17C756(A) UV-erasable, 68-pin PLCC, not code-protected devices that we could buy?

Preferably in the UK, for speed of receipt, but I guess the US is a more likely source...

NB Note that unlike almost all other devices the code protection is one-way, and once used the device is trashed and cannot be re-used despite the UV erase feature. So no pre-used, code-protected devices can be accepted (we have many of those!) But pre-used, non-code-protected devices would be just fine.

Thanks for reading and any eventual help!

Nigel
_________________
No comment! (Can't think of anything interesting to say!)

[Ttelmah]

You should be able to get these from the US.
The problem is RoHs. They never made an erasable version that is RoHs compliant, so the UK suppliers can't sell them, even if they have stock. Similarly US suppliers shipping to a UK address, won't supply these unless you can show an exemption (medical equipment for example). However if you have a friend in the US, they can order these from a US supplier, and then 'accidentally' ship them on to you....
Several of the US wholesalers do list them, as 'available US only'.

Best Wishes

[aopg64]

Hi Ttelmah,

That's an interesting angle. That may be why some searches list no stock online but ask that you call for a quote. Our company does have US offices, but I can imagine this being a pain to get through our purchasing people at either end as working in the same building together is already hard enough, let alone across the pond!
:-)

Thanks,

Nige
_________________
No comment! (Can't think of anything interesting to say!)

[asmboy]

rohs exemptions:
i have done design jobs( circuits and code) for a US biomedical research instrument company - AND ALSO A job for a company with a unique RF product.
i recall that the US and EU BOTH have a ROHS exemption for products in those two categories.

if YOUR product can reasonably meet that definition -0 you might find relief

[Ttelmah]

It is basically anything 'life critical'. So, automotive, aviation, medical etc..
Other thing to comment, is that Microchip themselves might well be able to help.

Best Wishes

[Mike Walne]

As I understand it, IF you can claim that your product was designed, and marketed, BEFORE the requirement to meet ROHS, you SHOULD be able to continue to SELL and SERVICE that product here in the UK.

Mike

I may of course be wrong, but it's worth giving it a try. It does not make sense to have to throw away a servicable item simply because it contains non-ROHS components. If I had any of the parts you want I, for one, would certainly let you have them.

EDIT Is there not a form, fit, and functional replacement?

[Ttelmah]

Yes, you are slightly missing the point here. The chips themselves, in OTP versions are readily available. It is the development version (erasable), that you can't get.
I had exactly the same problem on a different chip. On an 'obsolete' part it was not worth their while tooling up to make a development chip that meets RoHs. In fact the problem is with the suppliers not understanding the regs, rather than with the regs themselves. 'Equipment for research and development', is one of the exemptions. So you should be able to order a development chip without problems. However try persuading UK and US suppliers of this....
In my case, Microchip were aware of this exemption, and happily supplied chips (free). Which is why I said contact them.

Best Wishes

[asmboy]

and there is certainly no pin compatible 18F part either

wow - really sucks to be in your shoes. ouch

was this an .ASM coded project ???

[aopg64]

Hi All,

Thanks for your comments and suggestions. In no particular order:

1. I have contacted Microchip; they gave me a few chip search links, all of which seemed to lead to OTP parts. I have since added to my support ticket quoting the RoHS angle, to see if they could confirm it.

2. We can get RoHS exceptions here due to the nature of some of our products. We are having to update quite a lot of kit from the mid 90's.

3. Luckily, it's almost all C, just some .asm for a few library files, and I don't touch that...

4. Yes, it's annoying there was never a pin-for-pin replacement for the 17C756. Another bit of kit I had to upgrade luckily did have a flash equivalent as it was so old we didn't have any UV versions of it anywhere any more so that really would have caused major problems! Phew!

Thanks again to all for your advice/help.

Nige
_________________
No comment! (Can't think of anything interesting to say!)

[temtronic]

Another option might be to purchase a close replacement PIC and design an 'adapter' board to allow the new PIC to plug into the old socket.
There are several mfrs of PCB adapters, hopefully one has what you could use, providing space isn't tight.

I did that with a 68HC11 product just before I found out about PICs 20+ years ago.

With all the new variations of PICs now out, something has to be 'close', probably better !

hth
jay

[aopg64]

Hi Jay,

I'm afraid there is no political will to design an adaptor or redesign the main PCB. This is a product that was meant to be retired ~10yrs ago! Plus porting from the C17 compiler to CCS which we've used for as long as I've been here - other than for this product - would be a pain. We looked at an available adaptor socket to which we could put our own PCB, but the adaptor was >UK£60, so about US$100 each!!

A completely different replacement unit is being designed by another area of the company and that has had a back-burner low priority too as that has lingered for the last 2-3 years with the engineers doing a bit more on it in their downtime....and by the time that is ready we know the main customer is going to want to change to a completely new system anyway!

Ah, the joys of working on government/corporate long-term legacy systems!

It is really, really annoying that Microchip didn't make a pin for pin compatible flash device....

Anyway, one of the links Microchip Support suggested gave the impression it did have the correct device, though the description said it was OTP - hopefully a cut and paste error and it _is_ the UV erasable windowed device. I've sent an RFQ and for them to confirm what device it is.

Fingers crossed!

Nige
_________________
No comment! (Can't think of anything interesting to say!)

[RF_Developer]

This is one of the consequences of Microchip's policy of never (or almost) obsoleting any part.

It's great in that there is long term security of supply. Buyers love that. But it makes it quite difficult to justify changing to newer parts as unavailabilty of PICs is rarely a driver for redesign.

The commercial pros and cons of this are complex. There is certainly a strong argument for sticking with a tried and tested (and quirks known) PIC for as long as possible and only moving to newer parts if they offer some definite advantage, such as a requirement for better performance, greater IO count, lower package size for more compact designs etc.

On the other hand some PICs are effectively obsolete, or at least obsolescent. All PIC17s fall into this category, as the entire family has been sidetracked, but are still available for production. As they are strong discouraged for new designs, the availability of design/development orientated versions is very limited, and may even dry up altogether (which I suspect you'd like as it would make your task simple: can't modify it). The PIC17C756 is decribed by Microchip as "End of Life", and its replacement, the PIC17F756A as "Not Recommended for new designs". Any modification is a "new design" in this context.

It may be that the company you work for won't do anything to move toward a replacement design while the old devices are available. The trouble is that with Microchip they'll always be available, until Microchip stop trading that is. As you've seen, replacement projects can, and are shelved indefinitely as there is no pressing requirement to redesign, simply a desire to, which can often easily be countered by commercial needs, i.e. it cost money to redesign, and "nothing" to carry on with the old. The trouble is that all the effort you are putting in to finding these parts is a cost. It costs to keep on searching for old, difficult to find parts. Maintaining the old design is not zero cost. A redesign, or a replacement product should come when the costs of maintaing the old are greater than a new design. One big problem with that is that new designs are seen as single large cost, whereas maintaining old is seen as lots of small ongoing costs, where no one, two, or evne more costs are very large. If the true cost of maintaining such old products was known in advance, and predicting such costs is very difficult, and as the costs are not properly and fully recorded it often difficult to get a true lifetime cost looking back, then maybe redesigns would come a lot earlier in many product's life.

There comes a point, of course, where it really isn't justifiable to keep on throwing money at old designs. Also most companies have a policy of continuous improvements to all products, and so for they, if they put their policy into practice, and many companies don't - they just like to talk aboout it rather than do it - they will bring new products to the market long before there's any supply issues with the previous products. The sort of supply issues you are having should surely be seen as massive big red flags that something is seriously wrong somewhere.

I have been saddled with a somewhat lesser problem: before I arrived someone standardised on a PIC for most new designs. Great, except that it was the PIC18F8585, which, it turns out, has some serious errata issues. Despite this PICs serious shortcomings (as in any given build of code may, or may not work under some, or many different situations which cannot be predicted in advance: translation: its down to chance whether this PIC will work reliably, or not..., with my code!)

I can't really help in finding the part you need. I have to say that your situation sounds really bad, and one of the worst examples of its type I've heard of. I have seen it a few times, more times than I'd prefer :-( There are times when the bullet has to be bitten. Microchip are unlikely to force your employer's hand in this respect. The last time I was in anything like this situation was when we had a product that used a NXP ARM micro, the LPC2214. NXP updated it with revised, higher speed IO and other tweaks, but due to our reliance on the undocumented IO port start-up behaviour of the early parts, we couldn't use the new parts in our product. In the end it came down to buying the last 80 examples of hte old parts in the UK, from RS of all paces, who had a couple of waffles, one part used, left. When we tried to order these parts that had been specially reserved for us, we were told there was only 50! We were able to tell them that the other 30 were probably with one of their sales engineers; the one who had found the parts for us. Sure enough he had them. Then they were gone.... Me too, that was it, I was out of that company. The MD had decided I, and a collegue who had turned over every possible stone worldwide to find the part, had not done enough, and not quickly enough, to find them. Stuff that..!

Good luck.

RF Developer.

[Ttelmah]

As a completely lateral thought on this, if the parts currently ordered turn out not to be eraseable, it might be worth talking to one of the companies who used to sell ICE units for these. They might have one of these still on the shelf, and be prepared to do a good price for this. Back when these were the current type of chip, I used/had one of these, and since ICD didn't exist on these chips, this was the 'way to go' at the time.

Best Wishes

[aopg64]

Hi ttelmah,

About Para 4 or 5 of the original post - I am actually trying to get the Picmaster ICE that we have working. Last Tuesday I had it running on a 90's vintage PC. Not with the latest code and whatever I did then I can't seem to repeat now!

It was amazing when it did run though! Just like the real thing! A real blast from the past.

Anyway guys, the bosses have decided we remove the little-used feature that was causing the trouble (imaginary serial input!) which was one line of code to change and do a complete re-test of the 99% of the other features with the new compiled code.

So the only worry now is they find a bigger bug that we missed in the last testing round!
;-)

Still, it keeps me in a job! Though my firmware skills are gradually going back in time rather than forward.....Scary that the lad who did the previous round of testing at 20yrs old is younger than the ICE (1991 - vintage!) I am using!

Thanks again,

Nige
_________________
No comment! (Can't think of anything interesting to say!)