Post by thread:[PIC] Best way to program a OTP EPROM PIC

This is kind of picking up where I left off some time ago, when I was asking if the Wisp628 would program a one-time-programmable EPROM chip, and Wouter responded that it would not, because it couldn't supply the amount of current necessary. Someone else chimed in, Olin I believe, that C chips were evil in any case and that I should stick with flash chips. OK, I accept both of those things. However, there is still a nagging question: If I do, in spite of good advice, want to play around with a C chip, what is the best way to program it? Will an ICD2 do that as a matter of course? Will Olin's EasyProg do it? Would I have to build some sort of parallel programmer? What is the best way? John

John Nall wrote: > OK, I accept both of those things. However, there is still a nagging > question: If I do, in spite of good advice, want to play around > with a C chip, what is the best way to program it? Will an ICD2 do > that as a matter of course? Will Olin's EasyProg do it? It might be capable electrically, but only no C chips are supported by the firmware/software. ****************************************************************** Embed Inc, Littleton Massachusetts, (978) 742-9014. #1 PIC consultant in 2004 program year. http://www.embedinc.com/products

Pigstart+ Does all c and a lot of the f chips Steve -----Original Message----- From: spam_OUTpiclist-bouncesTakeThisOuTmit.edu [.....piclist-bouncesKILLspam@spam@mit.edu] On Behalf Of John Nall Sent: 29 January 2006 19:18 To: Microcontroller discussion list - Public. Subject: [PIC] Best way to program a OTP EPROM PIC This is kind of picking up where I left off some time ago, when I was asking if the Wisp628 would program a one-time-programmable EPROM chip, and Wouter responded that it would not, because it couldn't supply the amount of current necessary. Someone else chimed in, Olin I believe, that C chips were evil in any case and that I should stick with flash chips. OK, I accept both of those things. However, there is still a nagging question: If I do, in spite of good advice, want to play around with a C chip, what is the best way to program it? Will an ICD2 do that as a matter of course? Will Olin's EasyProg do it? Would I have to build some sort of parallel programmer? What is the best way? John

> OK, I accept both of those things. However, there is still > a nagging > question: If I do, in spite of good advice, want to play > around with a > C chip, what is the best way to program it? Dunno, I use a PS+ because I have one. But do you realise that besides the progger you will need an UV-eraser, and you will need (a bunch of) windowed versions of the chip you want to use? The alternative is use once and throw away. Maybe you are a much better programmer than I am? Oh, and your target circuit will probably need a ZIF socket. And don't code protect your windowed chip! And do write down the osc calibration word! And don't rely on the RAM being all zero's! Wouter van Ooijen -- ------------------------------------------- Van Ooijen Technische Informatica: http://www.voti.nl consultancy, development, PICmicro products docent Hogeschool van Utrecht: http://www.voti.nl/hvu

On January 29, 2006 11:18 am, John Nall wrote: > This is kind of picking up where I left off some time ago, when I was > asking if the Wisp628 would program a one-time-programmable EPROM > chip, and Wouter responded that it would not, because it couldn't > supply the amount of current necessary. Someone else chimed in, Olin > I believe, that C chips were evil in any case and that I should > stick with flash chips. Evil C chips, now that is interesting to hear. Probably a couple of errata that got corrected later with the F series chips. In other words, if you don't have chips, it is recommended you get the latest chips, which are F series instead of messing around with older stock which are eventually going to get phased-out of production. > OK, I accept both of those things. First, expect less help, because you ignored good advice. ;-P > However, there is still a nagging question: If I do, in spite of > good advice, want to play around with a C chip, what is the best way > to program it? Will an ICD2 do that as a matter of course? Some C chips need to be programmed in parallel fashion, therefore have to be physically out of the circuit to program so they cannot be programed in-circuit. You will need to read the microchip website documentation for programming those chips you have in question. Some can be programmed serially, others cannot. Plus the programming voltages/currents/etc are different. > Will Olin's EasyProg do it? If the programmer you have does not specify it can handle a C, then it is likely it won't, so don't expect it to magically work. This means you have to read the documentation for what the programmer can and cannot program. > Would I have to build some sort of parallel programmer? What is the > best way? Sounds like just a personal project, and you probably got to build your own programmer or buy a picprogrammer that can handle them. You might want to go to google and enter these 3 search words: pic16c programmer schematic >From that search, this random page appeared interesting: http://www.jdm.homepage.dk/newpics.htm ...but it was the 1st page I looked at.

On Jan 29, 2006, at 1:09 PM, Wouter van Ooijen wrote: > But do you realise that besides > the progger you will need an UV-eraser, and you will need (a bunch of) > windowed versions of the chip you want to use? The alternative is use > once and throw away. Maybe you are a much better programmer than I am? > There are pin-compatible and usually code-compatible flash versions of most of the the -C chips that can be used for development, aren't there? Especially now that microchip has made some 12-bit core 16F chips (16F54, 16F55, etc) BillW (who also has a fair set of Evil C chips sitting around.)

Wouter van Ooijen wrote: > > > The alternative is use >once and throw away. > . Well, I didn't give the whole story. I have a chip in my possession (a 14000) and I have the source of a program which I would very much like to program into it. It was written specifically for the 14000 and is supposed to work. (No, I'm not really that naive, but it might work). The alternative is to find a flash chip that has the necessary peripherals and re-do the code for that. Not an impossible task, or at least I don't think so, but if just burning the code into the 14000 is possible, then that seems easier, and I already have the chip. >> Maybe you are a much better programmer than I am? > > . No -- with the greatest humility, I will yield any day to someone who can write a compiler! :-) . >> Oh, and your target circuit will probably need a ZIF socket. And don't >code protect your windowed chip! And do write down the osc calibration >word! And don't rely on the RAM being all zero's! > > . OK, I used the wrong phrase when I said "play" with them. I have explained the motivation, above. I am now thoroughly convinced that I do not want to play with them. :-)

> OK, I accept both of those things. However, there is still > a nagging question: If I do, in spite of good advice, want to play > around with a C chip, what is the best way to program it? Depends which one it is. I started with a simple parallel programmer but all it does is 16C84. Just a 74LSsomething and a 555 booster Just talking about "playing around with" OTPs gets my sphincter tightening. 'orrible things to develop on. And spare me the fuffing about with UV and pricey windowed parts. Yuck Wouter said - > The alternative is use once and throw away OTP can be used more than once, if the code is less than half the space available (Microchip thoughtfully made FF NOP), and you are limited by how many bits are in the start org > Maybe you are a much better programmer than I am? Even when I developed 68HC products using an emulator, I never felt comfortable burning an OTP. Money down the drain if it needed changing

On Sun, Jan 29, 2006 at 02:18:14PM -0500, John Nall wrote: > This is kind of picking up where I left off some time ago, when I was > asking if the Wisp628 would program a one-time-programmable EPROM chip, > and Wouter responded that it would not, because it couldn't supply the > amount of current necessary. That's correct. Most EPROM based chips need 50 mA of current for Vpp. > Someone else chimed in, Olin I believe, > that C chips were evil in any case and that I should stick with flash > chips. I note in your other post that it's a 14000 part. No flash equivalent for it. > OK, I accept both of those things. However, there is still a nagging > question: If I do, in spite of good advice, want to play around with a > C chip, what is the best way to program it? Will an ICD2 do that as a > matter of course? Will Olin's EasyProg do it? Would I have to build > some sort of parallel programmer? What is the best way? Probably the last one. A high current switchable Vpp is pretty easy to put together using a LM317 and a couple of transistors. Use one transistor to ground the ADJ pin and a second to ground OUT after passing through a current limiting resistor. When both transistors are switched on, the 1.25V output of the LM317 is shunted via the out transistor. When both are off the ADJ pin rises to normal voltage divider setting of 13V. If the current limiting resistor is 100 ohms, then 130mA of current is available to drive Vpp. This is an old Steve Ciarcia design that he used for an intelligent EPROM programmer. BAJ

Jinx wrote: > OTP can be used more than once, if the code is less than half > the space available (Microchip thoughtfully made FF NOP), and > you are limited by how many bits are in the start org NOP is 00. You can program it as many times as you want by putting GOTOs in 0x00 and 0x04. Next time you make those into a NOP and program GOTOs in 0x01 and 0x05. If you want more, make a GOTO into a new table somewhere. Add bank switching if needed. -- Hector Martin (hectorKILLspammarcansoft.com) Public Key: http://www.marcansoft.com/hector.asc

Jose Da Silva wrote: >On January 29, 2006 11:18 am, John Nall wrote: > > >>This is kind of picking up where I left off some time ago, when I was >>asking if the Wisp628 would program a one-time-programmable EPROM >>chip, and Wouter responded that it would not, because it couldn't >>supply the amount of current necessary. Someone else chimed in, Olin >>I believe, that C chips were evil in any case and that I should >>stick with flash chips. >> >> > Evil C chips, now that is interesting to hear. >Probably a couple of errata that got corrected later with the F series >chips. In other words, if you don't have chips, it is recommended you >get the latest chips, which are F series instead of messing around with >older stock which are eventually going to get phased-out of production. > > > > OK, I accept both of those things. I am not sure you should. The "C" chips are more rugged, the EEPROM section actually works without dropping bits, and the 'C's are less susceptible to EMI. Production cut? It would kill a LOT of Microchip's income, so I doubt it will happen anytime soon. Don't get me wrong. I love the convenience of the "F" parts, but Microchip's flash implementation is not quite ready for prime time yet. I use a DIY K150 to program PIC16C devices. They are assembled and sold in the USA through amazon electronics, hobby engineering, and Carl's electronics. About $50 with a ZIF socket. --Bob -- Note: To protect our network, attachments must be sent to .....attachKILLspam.....engineer.cotse.net . 1-520-850-1673 USA/Canada http://beam.to/azengineer

William Chops Westfield wrote: >> For ONE, find someone locally with a fancy "device programmer" >and talk them into programming it for you. > > . Yup. I have thought about that. Although finding someone locally would pretty much be impossible. In these parts a "pic" is something that you carry to Saturday night dances, stuck up your sleeve, in case someone makes a move on your woman! :-) But I might try to find some kind soul who would do it for me, yes. It is complicated, however, by the fact that the 14000 seems to be kind of a bastard child insofar as Microchip is concerned. MPLAB has never heard of such a device (even though I have one sitting in front of me!). My experience with kind souls is that you want to ask the very minimum of them. (And I have been a kind soul myself a few times, so have complete sympathy.)

> NOP is 00 Ooops. I couldn't remember which and looked in an MPLAB disassembly for an 18F and saw FFFF as the alternate NOP For the 16F all 1s is addlw 0xff. Which might be mostly harmless (compared to CALL) but it's not NOP The 00 0000 0xx0 0000 bit pattern for 14-bit NOP means that 0000, 0020, 0040 and 0060 will execute as true NOP, although MPASM compiles the NOP op-code to only 0000 > You can program it as many times as you want by putting > GOTOs in 0x00 and 0x04 I believe OTP come from the factory full of 00 (I would have said FF before but now I think that's how F parts come - stupid brain fart). So with an OTP you can only make 0 into 1, and need to move the program code up in memory (now it's all coming back - same as the 68, except they had 9D ? as NOP - not particularly friendly) and alter, bits permitting, GOTO START Or not change START, but write 00 over the lower program Another option is that you put the program at the top of memory in the first place with a low GOTO START value and let the PC work through the NOPs until it hits the code. Subsequent versions move down through memory > If you want more, make a GOTO into a new table somewhere That'd work. I'd guess all this fuss might be worth it to correct a production run or something like that. Otherwise just use F

On Sun, Jan 29, 2006 at 03:38:54PM -0700, Bob Axtell wrote: > I am not sure you should. The "C" chips are more rugged, the EEPROM > section actually > works without dropping bits, and the 'C's are less susceptible to EMI. > Production cut? > It would kill a LOT of Microchip's income, so I doubt it will happen > anytime soon. For what it's worth I pull apart pretty much every random peice of hardware I get my hands on and read the chip numbers off it. (amoung other things) I've seen dozens of "C" chips in hardware, the only "F" chips I've seen are either 18F's, in short production run hardware like PIC programmers, or stuff I built! -- EraseMEpetespam_OUTTakeThisOuTpetertodd.ca http://www.petertodd.ca

On 1/30/06, Bob Axtell <engineerspam_OUTcotse.net> wrote: > > I am not sure you should. The "C" chips are more rugged, the EEPROM > section actually works without dropping bits, and the 'C's are less > susceptible to EMI. I totally agree that the "C" chips are more rugged. We used 16C72A and 16C621A/662A and other "C" chips in previous designs and they have better EMC performance than the newer chips like PIC16F819. > Production cut? > It would kill a LOT of Microchip's income, so I doubt it will happen > anytime soon. A lot of the automotive companies are using the 16C parts and Microchip will not cut the production for those parts. My 16C72A based design has been running for 5 years and I will count on Microchip to supply it for another 5 years or more. Of course, the flash parts are recommended in the new designs. PS+ clones will be a good choice to program C based chips. Simple programmers also supports many C parts using software like WinPIC 800. But PIC14000 is a rare specie so that I am not sure about any software which supports it. Regards, Xiaofan

Xiaofan Chen wrote: >> But PIC14000 is a rare specie so that I am not sure about >any software which supports it. > > Yes, that certainly appears to be the case! Oh well. Nothing that anyone has said has given me much hope on this particular project, so guess I will just trashcan the 14000 chip. (I won't REALLY trashcan it -- I will put it in a drawer, along with a little note. Wonder why Microchip disowned it? Did it cause lung cancer or something?). Some ideas are good, some ideas are bad, and some ideas are in between the two. So will quit multi-tasking and go back to working on the translator (I needed a break, anyway). The first cut on that is just about done, by the way. All of the 18F452 instructions have now been mapped to the appropriate 30F instruction (or instructions, since some 452 instructions produce 2, or even 3, 30F instructions). That part was tedious -- the rest should be more fun. :-) And by the way, to make sort of a strained joke: Some time ago, talking about all the extra registers on the 30F family, someone asked: "Of what earthly use are all those extra registers??" OK, here is the answer: When you are writing a translator, they come in handy! W1 = temporary storage. W2, W3 and W4 used for indirect addressing (take the place of FSR0, FSR1 and FSR2), W5 and W6 receive the results of multiplication (take the place of PROD), W7 takes the place of TBLPTR and W8 takes the place of TABLAT. :-)

> > OK, I accept both of those things. However, there is still > > a nagging question: If I do, in spite of good advice, > want to play > > around with a C chip, what is the best way to program it? > > Depends which one it is. I started with a simple parallel programmer > but all it does is 16C84. Just a 74LSsomething and a 555 booster The 16C84 is a naming error. It is, for all practical purposes, an F chip. Wouter van Ooijen -- ------------------------------------------- Van Ooijen Technische Informatica: http://www.voti.nl consultancy, development, PICmicro products docent Hogeschool van Utrecht: http://www.voti.nl/hvu

> Well, I didn't give the whole story. I have a chip in my > possession (a > 14000) and I have the source of a program which I would very > much like > to program into it. It was written specifically for the 14000 and is > supposed to work. (No, I'm not really that naive, but it > might work). > The alternative is to find a flash chip that has the necessary > peripherals and re-do the code for that. Not an impossible > task, or at > least I don't think so, but if just burning the code into the > 14000 is > possible, then that seems easier, and I already have the chip. OK, that's the only way I would use an OTP chip: using a already debugged program. > No -- with the greatest humility, I will yield any day to someone who > can write a compiler! :-) Don't be awe'd too much, it is not that difficult. It just takes a lot of time. Wouter van Ooijen -- ------------------------------------------- Van Ooijen Technische Informatica: http://www.voti.nl consultancy, development, PICmicro products docent Hogeschool van Utrecht: http://www.voti.nl/hvu

>> The 16C84 is a naming error. It is, for all practical purposes, an F >> chip. > > In what respect ? It can be electrically re-programmed, using algorithms and hardware that programs 16F chips. All other C chips are EPROM chips: either program-once (OTP), or windowed (for development, use an UV-eraser to erase). Wouter van Ooijen -- ------------------------------------------- Van Ooijen Technische Informatica: http://www.voti.nl consultancy, development, PICmicro products docent Hogeschool van Utrecht: http://www.voti.nl/hvu

Yes, the PIC16C84 was an experiment by Microchip. --Bob Wouter van Ooijen wrote: {Quote hidden}>>>The 16C84 is a naming error. It is, for all practical purposes, an F >>>chip. >>> >>> >>In what respect ? >> >> > >It can be electrically re-programmed, using algorithms and hardware that >programs 16F chips. All other C chips are EPROM chips: either >program-once (OTP), or windowed (for development, use an UV-eraser to >erase). > >Wouter van Ooijen > >-- ------------------------------------------- >Van Ooijen Technische Informatica: http://www.voti.nl >consultancy, development, PICmicro products >docent Hogeschool van Utrecht: http://www.voti.nl/hvu > > > > -- Note: To protect our network, attachments must be sent to @spam@attachKILLspamengineer.cotse.net . 1-520-850-1673 USA/Canada http://beam.to/azengineer

Jinx wrote: > I believe OTP come from the factory full of 00 (I would have > said FF before but now I think that's how F parts come - stupid > brain fart). So with an OTP you can only make 0 into 1, and > need to move the program code up in memory (now it's all > coming back - same as the 68, except they had 9D ? as NOP > - not particularly friendly) and alter, bits permitting, GOTO > START Hmm.. I thought both flash and OTP use 0x3fff unprogrammed (that's for 14-bit). I don't have any C parts here to check though. -- Hector Martin (KILLspamhectorKILLspammarcansoft.com) Public Key: http://www.marcansoft.com/hector.asc

> Hmm.. I thought both flash and OTP use 0x3fff unprogrammed > (that's for 14-bit). I don't have any C parts here to check though Me neither. I just recall (I think correctly) that from a discussion ages ago OTP are supplied full of 00. No idea whether that's an inherent part of the process of making an OTP (as I mentioned before, 68HC OTPs are, or at least used to be, 00 too) or whether MChip chose by design to have the C versions full of 00 (NOP) that can be written over

> >> The 16C84 is a naming error. It is, for all practical purposes, > >> an F chip > > > > In what respect ? > > It can be electrically re-programmed Wow, I did not know that. A couple of C84s came with the kit and being a PIC newbie at the time, I knew no better than what was in the documentation, and it sounds like the author of that based what he said on his experience with the other Cs. I barely spent a few weeks with it and moved on to the PS+ and Fs shortly after, none the wiser

Wouter van Ooijen wrote: >> > > No -- with the greatest humility, I will yield any day to someone who >> can write a compiler! :-) >> > > Don't be awe'd too much, it is not that difficult. It just takes a lot > of time. > . "Awe" is not the word I would use -- "impressed" is a better word. I've had compiler theory, and know how it is done. I'm impressed by anyone who actually does it, though. I know the theory of how to climb a mountain, too, but am greatly impressed by people who actually do it. :-)

Reminds me when I was doing development using the 16C54....boy tell ya, when the F84 came out, that was great!! Not that it was ICP, but the erase time. Always having to keep two or three in the eraser....eliminated that problem. Other than the fact the reset vector was at the end of the device...not at 0x00 So I even have one of the old Picstarts, bare board with the zif socket. Now of course, to program an windowed part, most prom programmers support all the old chips, not sure about the F series. --------------------------------- What are the most popular cars? Find out at Yahoo! Autos

> Nothing that >anyone has said has given me much hope on this particular project, so >guess I will just trashcan the 14000 chip. (I won't REALLY trashcan it >-- I will put it in a drawer, along with a little note. Lounging in the drawer means nobody is in a hurry to see it work. Since you claim to live in the sticks, you could mail it off to someone and have them program it and send back. Worth a shot. Meanwhile, where I work we still have our first programmer, and we still burn 16C65's from time to time. The 14000 is listed... Barry

Barry Gershenfeld wrote: > > Lounging in the drawer means nobody is in a hurry to see it work. Since > you claim to live in the sticks . . . . CLAIM to live in the sticks? Why would anyone lie about something like that? It is hardly a status symbol. :-) I might claim to live in Beverly Hills, or claim to live in the Waldorf Astoria, and then perhaps there might be reason for being incredulous. But not the sticks of Northern Florida. No one claims to live there unless they have to. Hot in the summer, cold in the winter, bugs, alligators, water moccasins . . . . > > . . . you could mail it off to someone and > have them program it and send back. Worth a shot. Meanwhile, > where I work we still have our first programmer, and we still burn > 16C65's from time to time. The 14000 is listed... > Am I to infer that you might be willing to volunteer? John