Read 2716 UV EPROM

[circa77]

What I refer to as a 'chip personality profile' for want of a better expression doesn't only consist of the programming algorithm, it also defines the number of pins the device has, which pins are the address pins, which are the data pins, which pins to apply main power to, which pin to apply VPP to, which pins are the /CE pins - it is a 'bundle' of ALL of the information that the device programmer needs to have in order to read or program the selected chip, of which the programming algorithm is only one part.

You're splitting hairs. All of that is simply rolled up in the algorithm, especially when you're using a driver-per-pin programmer. You may refer to it that way, but it's not industry nomenclature.

 

[SiriusHardware]

I'll offer a counterargument, which is that if the chipmakers freely published their algorithms so that any idiot could build their version of a programmer, a chipmaker might well see a decline in their sales due to users having bad experiences (read: low yields) from unqualified, crap programmers.

I agree, I think that's exactly why device manufacturers often did restrict distribution of their programming algorithms and tried to have a pool of 'certified' programmers which were approved for use with their programmable products. Shonky device programmers used by people who assume the programmer knows what it's doing certainly could give a chip product line an undeserved bad reputation.

 

[circa77]

I don't have any legacy 2716s to try, but it sounds like it depends on whether the part requires special voltages that the programmer can't supply. I've read legacy 2764s without issue. I bought it mainly to program EEPROMs, which I use in place of legacy EPROMs. Or the occasional replacement/expanded firmware for RC-bus or MinZ computers (e.g. SST39VF040).

And about that "legacy" business: Do you mean those to be somehow distinct from modern 2716s and 2764s?

 

[durgadas311]

What I refer to as a 'chip personality profile' for want of a better expression doesn't only consist of the programming algorithm, it also defines the number of pins the device has, which pins are the address pins, which are the data pins, which pins to apply main power to, which pin to apply VPP to, which pins are the /CE pins - it is a 'bundle' of ALL of the information that the device programmer needs to have in order to read or program the selected chip, of which the programming algorithm is only one part.

Exactly. What "algorithm" means to most people does not suffice for what we're talking about.

 

[durgadas311]

I've been designing systems for close to 50 years. I've been designing with programmable logic for 40.
I have thousands of data sheets, data books, and texts, know them well, hold patents in this stuff, and have never heard anything so goofy.
If you think that's a well-established term, provide citations.

I don't need to. These terms are obviously what people on this thread understand and naturally use. Ultimately, that's what matters. There are plenty of google hits that prove this term is in wide-enough usage that we don't need some word-police nagging us. I remember programmers with "personality modules", only you can explain why you don't.

 

[circa77]

Two terrible answers. What something means to "most people" is immaterial. The terms that actual professional practitioners use matter. And yes, actual citations matter too, and "just google it" is lazy, sloppy, and a guarantee of finding the stupidest answers that have been widely propagated by amateurs who should listen to pros more than they post nonsense.

"Personality modules", however, is an accepted term in this and other contexts. But that doesn't mean you're entitled to mangle and shoehorn a variation on that term into any place you feel like without sounding silly. This is not "you do you, sugar" unless you're doing truly original work, and from what I've seen so far, you're not. I mean, I'm having this conversation with someone who thinks an EEPROM is an NVRAM. It ain't, honey.

 

[gslick]

Why should anyone take professional terminology advice from someone who sounds silly and immature using "windoze" instead of Windows?

 

[circa77]

If you're capable of distinguishing statements of fact from expressions of opinion, the answer to your question is obvious.

 

[durgadas311]

Back to the original problem, if this thread is even salvageable. I'm wondering if there might be something about the chips VPP pin (21) such that setting the T48 to "28C16" presents a solid logic "1" signal while "2716" does not (since in that case it needs to optionally go to 25V to program i.e. it is not TTL-compatible). You also mentioned that you had tried "27C16", which perhaps is a typo, but a "27C16" is different than a "2716" and the programmer might not be able to drive a 2716 correctly using the 27C16 personality. It depends on just what it does for each. The (Linux) XML file for the TL866ii+ has a lot of variations on the 2716 and I've not dug into the specifics to see how they might differ.

 

[SiriusHardware]

OK, so we need the help of someone who has:-
-One of the programmers which won't read Intel 2716s
-An Intel 2716 which the programmer refuses to read properly
-A scope

If anyone with this combination of assets can help out, we may be able to suggest measurements which will help us to get to the bottom of this particular problem - for example on the pin mentioned by durgadas311 - VPP, pin 21, which I understand needs to be at +5V in read mode.

One thing, though, I have also come across one case of a respected old legacy programmer (Stag PP39?) which could not read these particular devices 'as themselves' but could read them 'as' a 28C16.

 

[daver2]

Back to engineering...

And pin 21 is a power pin not a logic pin - the current consumption is potentially much higher for a power pin than a logic pin.

Dave

 

[fdiskitup]

My chips have great personalities.
I’d be willing to help troubleshoot.

 

[fdiskitup]

im updating the firmware in my crappy t48 eprom programmer

 

[fdiskitup]

The T48 is updated
I have 2 x D2716 Intel chips on the desk.
I'm running Windows 10 on a tower pc. (i.e. fully powered USB)
I've got a cheap scope I can hook up, or I could use an arduino mega (cheap version) to monitor the data/address chip legs.

 

[fdiskitup]

...you know in the programmer software it does actually say ALgorithm..... (but that is quite hard to spell).



From the supported list in the XGecu site neither the T48 nor T56 directly list intel "D2716" as supported... but D is the grey ceramic package. All the data sheets point to M2716 as equivalent - this ALgorithm doesn't work for reading or writing.
you'd think reading would be ok with any 2716 profile.

right back in the thread before all the debate we'd figured out reading with a different profile was fine. i.e. AM28C16A profile. And I verified the data by mounting the chips back in the machine and booting from them, and also dumping the rom contents to the terminal as text. 40 year old chips with no window covers...

This served me fine for archiving the contents of potentially one of a kind homebrew chips, but it was a little unsatisfactory.

I've put in a support request with the XGecu website, lets see if I get a response.

 

[durgadas311]

Back to engineering...

And pin 21 is a power pin not a logic pin - the current consumption is potentially much higher for a power pin than a logic pin.

Dave

While that is technically true, it is all about what the programmer asserts on that pin. For programming, the current draw is certainly higher during programming, but for READ mode (pin 21 = Vcc), the current draw is minimal. The datasheets don't define what happens if Vpp drops below Vcc (or goes to 0V). All I'm suggesting is that it is a place to look. Comparing the difference in reading between 2716 and 28C16, that appears to be the main difference. There may be something about the programmer and/or personalities that causes Vpp to fail to reach Vcc in the cases that reading fails. It would be nice if we had a schematic for the programmer so we could see just how pin 21 gets connected in each case. It just seems that pin 21 is different when using 28C16 than when using 2716. Of course, one must not try to program the 2716 using the 28C16 personality, as pin 21 (/WE) will go to 0V (logic 0) and then enters that undefined state.

 

[durgadas311]

It may not be possible to use the T48 (at al.) programmer for devices that require 25V (or 21V). The ones that use 12V *might* work. But, if you need to burn new 2716s, you could substitute 28C16 chips and burn them as EEPROMs. They should drop-in to most circuits as a replacement.

 

[SiriusHardware]

Yes, as stated earlier these chip programmers are legendary for not being able to supply a high enough VPP for 25V and 21V devices and I think the 'supported devices' list for at least one of these types of programmer admits that it is compatible with certain devices in read mode only. It may actually be impossible for the hardware as designed to produce a high enough VPP, so maybe for the moment we should concentrate on what -should- work, namely the reading of Intel 2716s. Laser focus on just one thing at a time.

 

[SiriusHardware]

OK, so, could anyone who can muster the requirements outlined in post #110 please try scoping what happens on PIn 21 of the 2716 during the read process?

 

[fargo]

I received my GQ-4x4 programmer yesterday. The MM2716, which TM48 failed to read, is listed in the GQ-4x4 program. The programmer could read the MM2716, TMS2516 and other intel 2716 compatibles without any hassle. When choosing a 2716, this programmer even gives a message advising to use an external power supply to program the chip.

I'm keeping the T48 because it's good for newer ROMs, but for older ones, the GQ-4x4 is the one I'm going to use.