Post by thread:[PIC:] Programmer oddities/sanity check

At 08:20 PM 4/24/04 -0400, Anthony Toft wrote: >I am trying to program a chip, using a Tait style programmer. I have >recently switched to a 18v/800mA wallwart (measures 24v on the DVM) >instead of 2 9v batteries. I think this is enough power, I am seeing >4.5v on Vdd and 13.2v on Vpp. Questions: a) Is the 24V reading with no programmer attached? If so, it's the peak voltage from the rectifier. "Usually", warts such as that will produce the rated numbers - if you load it, you are likely to see 18V at 800mA or so. This ought to be plenty to program with I think. b) Only 4.5V and 13.2V??? With the programmer attached? With no chip? With a chip? ??? With only the programmer attached to the wart, it should be closer to 18 out of the wart and the programmers 5V line should be 5V. If your pic-chip sample causes the drop to 4.5V, it looks like it may indeed be fried. If so do yourself a favor and toss it. c) Before continuing further, since you have some doubt about your wall wart, test it with some load resistors. If testing only with a DVM, you should realize that there may be significant amounts of ripple. To test that theory either measure the output under load with the meter set to show AC Volts -or- add 1000 or 10,000 uFds or so. The extra capacitance will average out the ripple. You can make your DVM tell the truth if you ask it nicely! d) Once you are sure of the output capabililty of your power supply, then you can proceed. e) Other people will have to add to this as far as the Tait programmer goes. I no nothing about it. As far as the programming algorithms are concerned, you will need to do your homework and look up the specs for each one and compare them. Where you say "the datasheet suggests that otherwise they are identical and meant as drop in replacements" that may refer to functionality, not necessarily how they program. If no one else can add to this, you will need to Read The Fine Manual more thoroughly. But what a fun way to spend the weekend, no? Good luck, Tom -- http://www.piclist.com#nomail Going offline? Don't AutoReply us! email listservKILLspammitvma.mit.edu with SET PICList DIGEST in the body

> I am trying to program a chip, using a Tait style programmer. I have > recently switched to a 18v/800mA wallwart (measures 24v on the DVM) > instead of 2 9v batteries. I think this is enough power, I am seeing > 4.5v on Vdd and 13.2v on Vpp. The problem is the only PIC I have is one > I possibly cooked some time ago. The programming is _very_ erratic, > predominantly failing with reading 0x3FFF for the device ID. Am I > correct in assuming the PIC _is_ dead and the programmer is working? I > don't want to risk cooking off another PIC (even a sample) and > making/buying another programmer is prohibitive for my (very) meager > project budget. Well, it's possible it still is the programmer. How long are your cables between the programmer and the PC? What OS you running? Is there a setting for slowing down how fast the PC is doing things? How long are the cables between the programmer and the PIC? > Also the only thing I have found different between the 16f627 (original > chip) and the 16f627a (sample) is the programming algorithm, my > understanding of the datasheet suggests that otherwise they are > identical and meant as drop in replacements. Is this correct? Mostly yes. There are a few bug fixes and an additional minor feature on the a part, however the different is zero from a firmware/plug in point of view. TTYL ---------------------------------- Herbert's PIC Stuff: http://repatch.dyndns.org:8383/pic_stuff/ -- http://www.piclist.com#nomail Going offline? Don't AutoReply us! email .....listservKILLspam.....mitvma.mit.edu with SET PICList DIGEST in the body

On Sun, Apr 25, 2004 at 12:22:20PM +0200, Jan-Erik Soderholm wrote: > Tom wrote : > > > As far as the programming algorithms are > > concerned, you will need to do your homework and look up the > > specs for each one and compare them. > > Or, if you'd rather would like to get on with some *real* > PIC development, just use a programmer (HW/SW combo) > that specificaly supports whatever chip you are using. > > I have never understood why everone should have to > read the programming specifications, and not just those > that are actualy designing *programmers*... Jan-Erik, That's what Anthony is doing. He specifically had a programmer question. For some reasons, some of which I understand and frankly some that I don't, Mchip feels compelled to change the programming interface every time they bring out a new chip. It's disheartning as a programmer developer because every time you think you're finished, here comes another wrinkle. 1) The 16F84 comes out with a serial programming interface. Great! Except that the chip erase sequence is not clearly specified. 2) Then the 16F87X parts come out. Great! Even mostly has the same programming commands as the 16F84 and a clearly layed out erase sequence (different than the 16F84 magic sequence). 3) Then the 18F series chips come out. It's a whole new ball game where the interface is essentially a subset of the 18F instruction sequence. So anyone who wants to design for 18F chips now essentially has to start from scratch. Also because of the memory arry size, block writing becomes a requirement. 4) Then the nanowatt chips comes along that requires a potentially different Vdd/Vpp/PGM sequence because the MCLR pin can now be an input instead of just a reset pin. 5) Then the A versions of some of the above parts come along with the block writing algorithm. It turns out to be the same as the 1 word algorithm but you must write in blocks. And oh BTW when doing the config sequence there are a bunch of variations from 1 word to 4 word writes. Sigh. That's why I'm a bootloader man. It can create a simple consistent interface apart from the crazy world of programming algorithms. But there are two major problems: bootstrapping and lack of universal support. As I have one on occasion, I'd like to discuss the issue. First a disclaimer: This discussion from henceforth is related to the hobbyist/student. Anyone who is doing developement on a professional basis should invest in professional programmers and tools. 'Nuff said. OK. What attributes does a hobbyist developer want? My take: 1) Fast. They want to get started today. 2) Cheap. No sense in spending upwards of $100 for a one off project with a part that they got as a sample, or purchased from Wouter (http://www.voti.nl) or Randy (http://www.glitchbuster.com) for a really good price. 3) Reliable. Debugging really sucks and is frustrating. Now the Tait/NOPPP/JDM/TLVP style programmers have been the staple. But I really think, and have consistently stated, that the hobby focus should not be on programmers or programming chips. It's a tool to develop projects. In addition I believe that it's an unnecessary tool and that for the vast majority of development that a bootloader will serve the three needs above better than any external programmer: 1) Fast. You can get started on a bootloader instantly. In addition it's always ICP. 2) Cheap. Nothing but the chip and the associated interface electronics. A perfect example is Wouter's ZPL bootloader, which requires nothing more than an ordinary NPN transistor, a resistor, and an optional zener diode, all which can be integrated directly into the programming cable. 3) Reliable. Since the bootloader is written exactly for the chip that it's embedded into, there less complexity in the system. Since there are less interface parts, there's less to fail or be flaky. There are a few issues. Not every chip supports it, especially the 8 pin 12F chips (a real bummer BTW). It does occupy program memory. It may also require dedicated chip resources, though ZPL uses the MCLR pin for its interface. But it seems to me that it'll get a hobby developer started faster and provide a clearer path to finishing. Frankly all of my finished projects were bootloader developed. But back to the problems. A hobby developer needs a reliable universal PIC programmer not for development, but for project setup and finalization: 1) Setup the bootloader to prepare for development and. 2) Finalization of the true target chip for the project if the bootloaded chip isn't the final target. For example developing with a 16F88 but the final target is a 12F629. That programmer exists and fills every hobby (and most pro) needs. Wouter's WISP628 programmer. Simple to build, reliable, inexpensive, universal (or nearly so), and high voltage. But has the same problem: needs a bootstrap. So here's the challenge: A simple serial, bootstrap, high voltage programmer that can by built to bootstrap the WISP628. It needs to serve no other purpose and only needs to program one chip, the 16F628. I've been wrestling with the problem as a replacement for the TLVP. However I've yet to figure out how to reliably capture a bit from the TX serial line without resorting to one shots, which fail when mistuned. I figure if we can cheaply and reliabily capture one bit per character, as ZPL operates, it should be possible to build a cheap, reliable, albeit slow bootstrap programmer for the 628. Any ideas? BAJ -- http://www.piclist.com hint: To leave the PICList piclist-unsubscribe-requestspam_OUTmitvma.mit.edu

Byron A Jeff wrote : > On Sun, Apr 25, 2004 at 12:22:20PM +0200, Jan-Erik Soderholm wrote: > > I have never understood why everone should have to > > read the programming specifications, and not just those > > that are actualy designing *programmers*... > > Jan-Erik, > > That's what Anthony is doing. OK, Sorry. I missunderstod. I thought he was trying to get a programmer designed by someone *else* to work. Anyway... [Proramming method history snipped] Nice history summary! Still, mostly of interrest to those *designing* programmers, not anything a "normal" PIC developer should have to know about. (Note, not *have* to know about..) > That's why I'm a bootloader man. It can create a simple > consistent interface apart from the crazy world of programmin > algorithms. I couldn't care less. I payed the designer of my programmer to take care about that "crazy world". Which I think anyone *not* designing his own programmer should do. > OK. What attributes does a hobbyist developer want? My take: > > 3) Reliable. Debugging really sucks and is frustrating. And should have been done by the programmer designer... > Now the Tait/NOPPP/JDM/TLVP style programmers have been the > staple. But I really think, and have consistently stated, that the > hobby focus should not be on programmers or programming chips. Agree ! :-) > It's a tool to develop projects. Absolutly ! > 1) Fast. You can get started on a bootloader instantly. You have to have the bootloaded loaded into your target first. > There are a few issues. Not every chip supports it, > especially the 8 pin 12F chips (a real bummer BTW). > It does occupy program memory. It may also require > dedicated chip resources, though ZPL uses the > MCLR pin for its interface. And another one : - You can not start with a naked fresh chip. > That programmer exists and fills every hobby (and most pro) > needs. Wouter's WISP628 programmer. Simple to build, reliable, > inexpensive, universal (or nearly so), and high voltage. But has > the same problem: needs a bootstrap. Yes but, this isn't realy "the same" problem, is it ? You can buy your Wisp628 complete with a programmed F628 (just look at it as any other "component" in the kit). Then you can use your Wisp628 to "bootstrap" any new firmware on your own. So the "bootstrap" of the Wisp, is a one time effort. The "bootstrap" of the bootloader, has to be done on *every* individual target chip you use. Or just simply forget about the fact that the Wisp628 contains a PIC at all, and just regard it as a "tool" to get your real work done. > So here's the challenge: A simple serial, bootstrap, high > voltage programmer that can by built to bootstrap the > WISP628. It needs to serve no other purpose > and only needs to program one chip, the 16F628. Since this should/could be a once-in-your-lifetime issue, I realy don't see what problem this would solve. {Quote hidden}> I've been wrestling with the problem as a replacement for the > TLVP. However I've yet to figure out how to reliably capture > a bit from the TX serial line without resorting to one shots, > which fail when mistuned. > > I figure if we can cheaply and reliabily capture one bit per > character, as ZPL operates, it should be possible to build > a cheap, reliable, albeit slow bootstrap programmer for the 628. > > Any ideas? Hm, maybe buy a pre-programmed F628 from Wouter ? Or ask someone nearby to flash one for you ? As a side note, in my current project to get four project groups on our local technical gymnasium going, we selected between ZPL and Wisp628. We decided to include a complete Wisp628 in each kit. That way, the project groups are self-running, and can easily re-start using a fresh/new 18F252 without having to ask someone else to "bootstrap" it for them. A little higher vcost, but less potential problems... Best Regards, Jan-Erik. -- http://www.piclist.com hint: To leave the PICList @spam@piclist-unsubscribe-requestKILLspammitvma.mit.edu

part 1 1959 bytes content-type:text/plain; charset=us-ascii; format=flowed (decoded 7bit) Anthony Toft wrote: >>I have been trough this problem of programming 628A vs 628 before with >>Tait's like programmer. > > > I haven't yet tried to program the 'A part, I originally designed in a > '627, but Microchip samples only has 'A parts. I haven't found anywhere > that definately says yes, they are drop in replacements in the > application. I have found there are programming differences, but that's > it. > > >>What programming software are you using? > > > I am using pp06 for Linux (pplin) and the Oshonsoft picprog for windows. > If you do not have to use pp06, here is what I do: I am using David Tait free software FPP (GIF file attached) (Windows only). It programs 628 without a hitch. It says it can also program the 627. http://people.man.ac.uk/~mbhstdj/piclinks.html Although the A part are not in the list of parts it can handle, a slight tweak in the programming procedure allows to program the 628A part (I have not tried OTHER A parts). Here is my finding on the difference in programming algorithm: While comparing the Microchip programming paper many time I noticed that both chips could be "bulk erased" but that same command would also erase the config word in the 628A which is a separate command on the 628 part. Now the tweak to program the 628 and 628A parts: To program the 628 part with FPP: - Press the Program button and that is all. FPP erase before programming automatically. For the 628A part: - Uncheck Config; - Press Erase; - Check Config; - Press Erase again - Then press Program I am also using Bojan Dobaj's PICALLW and also have to erase the 628A before it can program the 628A part. Hope this help. Gaston P.S. If you decide to go the FPP way and you need help, don't hesitate to ask. -- http://www.piclist.com hint: To leave the PICList piclist-unsubscribe-requestEraseME.....mitvma.mit.edu part 2 9199 bytes content-type:image/gif; name=fpp.gif (decode)