On Tue, May 27, 2003 at 08:34:14PM -0500, Picdude wrote:
> On Tuesday 27 May 2003 15:16, Byron A Jeff scribbled:
> > > I have never been able to use LVP for one reason ... the LVP pin is very
> > > inconveniently located on RB3, since I use ICSP almost exclusively
> > > nowadays.
> >
> > The LVP wasn't the issue. In fact I have a THVP design here:
> >
> > http://www.finitesite.com/d3jsys/proghvp.html
>
> I've had one question w.r.t. most HVP ICSP's, and perhaps you might know ...
> with the target PIC in-circuit, it's usually directly connected to it's own
> on-board PS, usually a 7805.  So how is the programmer connected to the Vdd
> line?

Three possible answers:

1) They don't have to be connected. You can just share a ground. This one is
   only a problem if you have a part where MCLR can become an input and a
   power cycle is the only way to reset. This is possible for me because I
   always power the programmer from the PC, and never from the target.

2) You can route board power through the ICSP connector so that the programmer
   can control Vdd when connected and a jumper connects straight through when
   the programmer isn't. My ICSP version of the TLVP here:

   http://www.finitesite.com/d3jsys/proglvp-icsp.gif

   can be wired that way. And BTW you can rewire the THVP same way.

3) Do what Wouter does: short the power on the target briefly

>  A diode on the programmer Vdd line only protects the programmer when
> the app circuit switches on.  I won't want to connect a diode permanently in
> the app-circuit's Vdd line for other reasons.


Note that this particular HVP circuit was designed for ICSP, so none of these
issues exist. And instead of putting the diode on the target, it would be
better placed on the programmer, since that's what would need the protection.

> Is the programmer Vdd line really piggybacked/paralleled with the app power?

No! That would be a disaster.

> I assume the app power will be off right?

Not necessarily. That can cause parasitic power issues where the programmer is
in fact trying to power the whole target with unknown power requirements.

>  Most articles/docs on ICSP address issues with MCLR connections, but not Vdd.
>
> I notice however, that your design uses app-ckt power during programming, and
> does not control the Vdd line.  But how is the PIC reset during programming?

The MCLR line resets the PIC. You can do that under power. The only exception
is when MCLR is turned into an input. Of course there's still the issue of
resetting everything else on the target.

> I've tried with the Tait, and it does not work if I leave the app-ckt power
> on only.

So the right solution is pretty clear to me: Route app-ckt power through the
ICSP connector and run it through a P-channel MOSFET that's on the programmer
which can then be controlled by the programmer. You need a P channel so that
you can get the it to turn on by dropping the gate low.

Hope this helps,

BAJ

--
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