I love a good argument Facts remain: 1. ICE is NOT a production chip. They are modified versions of production chips. They may be very close... and I'm willing to give that they will probably work just as production chip in 99.999999 percent of cases. But they are NOT production chips. Can't ICE actual production device. You must remove the production CPU and insert ICE. 2. Expense does matter and does make ICE not as good as ICD. Especially to "the rest of us." The "if you can't afford..." thing really gets me going. 3. Extra features of ICE are (IMH(N)O) minor compared to extra cost. I can see most all the things you mentioned happening from ICD as "after the fact" type of "oh! that's what must have happened" events. Not as good, but still ok. Points conceded: 1. ICE doesn't use processor resources (like RB6&7, registers etc...) but these are details of MCHip implementation. Scenix "did it right" and used the OSC pins so no processor resources are involved. 2. Not all target processors have ICD and almost all have ICE of some sort or another. As I said, if you can't get ICD, ICE is next choice. 3. Some extra features of ICE are nice to have. To be completely honest, the intensity of my opinions on this point are "penis envy" to a large degree. I can't afford to have ICE for every processor I'm interested in and it galls me. I think a lot of hobbyists (and maybe a few pros?) feel the same way. I can ask for and maybe get ICE from some of my clients, but that will never satisfy my desire to be able to "look inside" all the chips I want to work with. I want to make cheap ICD (or ICE if that were possible) available for a little more than the cost of the processor. Just to clear up a point: I have no intention of releasing an ICD device as a "I developed it" for profit device. I want to see ICD abilities either A) so low in cost that you might as well buy it (and the $80 price with the upcoming MChip Seminars gets that close) or B) open source, you breadboard it with a RS232 level converter and some cables, like Tony Nixons ASCII programmer or ROMZAP bootloader. Throwing in test bench and programming abilities for other devices like Dan Michael's WS(C)T: http://www.sni.net/~oricom/projects.htm or just ICD and programming like the "rumor" for the WARP-13 programmer from: http://www.new-elect.com makes for an even nicer way to get started or add some inexpensive extra ability to a pro bench. So, when is TECH-TOOLS coming out with a combined Test-bench, programmer, ICD for $50? --- James Newton mailto:jamesnewton@geocities.com 1-619-652-0593 http://techref.massmind.org NEW! FINALLY A REAL NAME! Members can add private/public comments/pages ($0 TANSTAAFL web hosting) -----Original Message----- From: pic microcontroller discussion list [mailto:PICLIST@MITVMA.MIT.EDU]On Behalf Of Jerry Merrill Sent: Monday, March 13, 2000 10:14 To: PICLIST@MITVMA.MIT.EDU Subject: Re: Magenta ICD Importance: Low James: I agree that real hardware is far superior to simulation. Although simulations can be quite accurate, it is very difficult to specify stimuli that mimics real-world and even more difficult to make that virtual world interact with the simulated PIC. Of course, it is also very non-real-time. HOWEVER, I have to disagree on one point you have stressed several times now: that an ICD is SUPERIOR to an ICE. I am obviously biased since I produce an ICE, but then again....you are about to produce an ICD so I think its fair.:) You may have had some bad experiences with old ICEs in the past but those were probably due to inferior products that did not use real bondouts. Those experiences should not extrapolate to the idea that ICEs are inferior to monitor based debugging by definition. The slave bondouts provided by Microchip for their devices use REAL PRODUCTION SILICON. This means that a properly designed ICE is not 'trying' to emulate the 'REAL' device; it **IS** the real device. -Same peripherals, Same I/O characteristics, same timing, same personality (quirks and all), SAME SILICON. The difference is that the bondout makes more of the internals of the device visible so that an ICE manufacturer can add external hardware. This extra visibility and control allows us to add things like execution tracing, timing, unlimited CODE breakpoints, data monitoring and breaking, and detecting and breaking on STACK over/under flow. Of course, it also allows us to load the user programs into SRAM instead of reprogramming FLASH. Finally, it allows us to do all of this without using ANY of the processor's resources; no interrupts, no stack space, no code space, no RAM space(file registers), no I/O pins. If you have a small budget then an ICD is the best you can afford, making it the best choice for your budget. But that does not make it BETTER, just the BEST you can afford Don't get me wrong. I think an ICD is a great compromise if you can't afford a real ICE. ICE rules! If you can't get an ICE, second best is an ICD! :) > >ICD rules! If you can't get an ICD, second best is an EMU. > >--- >James Newton mailto:jamesnewton@geocities.com 1-619-652-0593 >http://techref.massmind.org NEW! FINALLY A REAL NAME! >Members can add private/public comments/pages ($0 TANSTAAFL web hosting) > Jerry Merrill jerrym@tech-tools.com http://www.tech-tools.com FAX: (972) 494-5814 VOICE:(972) 272-9392 TechTools PO Box 462101 Garland, TX 75046-2101