On Sat, 18 May 2002, wouter van ooijen & floortje hanneman wrote:

>> I believe that to reliably program parts with internal oscillators, your
>> programmer design has to be able to control Vcc as well as Vpp (for high
>> voltage programming) or RBn/PGM (for low voltage programming, where n=3
>> for the 16F87x, 4 for the 16F62x, or 5 for the 18Fxxx). The idea is that
>> you avoid powering up the part until you're ready to program it; you
>> have to raise both simultaneously or nearly so, so that the program on
>> the chip doesn't start running before you get it into programming mode.
>> (Unless you have disabled the power-on reset timer, you have 72
>> milliseconds before the code starts to run.)
>
>Am I missing a point here? My ICSP programmer talkes its power from the
>target cicruit, so it has no means whatsoever to control it. It just pulls
>/MCLR low, and then *quickly* rises it to +13V. No problem whatsoever.

Until corrected I'll state that the programming mode is entered by pulling
/MCLR from GND (or < 0.4V) to Vpp within the specified short time. Of
course Vdd needs to be supplied first, but it does not matter whether it
was turned on just now or hours ago. I always add a 1N4148 diode with A to
OSC1 and K to /MCLR. It stops the clock when you or the ICSP reset the PIC
and it helps to keep the /MCLR high if you use a strange pullup resistor.
This is only used for development and testing but it has never failed me.
The stopped clock together with the PWRT buy enough time for slower ICSP
programmers to drive the /MCLR. Or so my theory goes (I use a heavily
modified Microchip datasheet (AN577?) type programmer for all
development).

Peter

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics