Lpc2000

At 06:47 AM 2/27/04 -0600, you wrote:
>On Fri, 27 Feb 2004 05:45:27 -0000, tsvetanusunov wrote:
>
> >It may not be that simple. The fact that Philips has been claiming
> >there are no sequencing requirements was bothering me so I just
> >did a simple experiment. I disconnected the 1.8V supply from the
> >micro and powered it up. After the other supplies had stabilized I
> >reconnected the 1.8V supply by hand. The micro started w/o a
> >problem. I've got to refine the setup a little more but
> >straightforward sequencing is not at fault.
>
>indeed it may be not that simple, I believe in our case Vreg rise
>time difference is matter of milliseconds.
>we didn't measure rise times on +1.8 +3.3V with 10uF/47uF but this
>just solved the problem
>we'll do some more experiments
>Tsvetan
>=================================================================
>Yes.  dv/dt is a known cause for false triggering of scr's.
>Another problem we had with some of the Atmel AT91's is a poor
>reset of the nTRST line which caused the JTAG chain to go active
>and halt the processor core.  It showed up as 3 memory accesses
>to fill the pipeline, then no bus activity at all.  I have no idea
>if any of this applies, just passing on observations that might
>apply.
>
>-Bill

Since if there is a sequencing problem it will have an effect on all of us 
I figured this was worth a little effort to gather some information if it 
was possible.  So I did in fact refine my setup here.  I've bodgered a 
prototype board and it now has a couple of switches taped to it.

The bad news is I can't duplicate the problem.  I do, however, have some 
data that Tsvetvan may be able to use together with what he has to narrow 
down the search.  Maybe.

I put switches on the 1.8V line and the power monitor reset line from the 
1.8V monitor (monitors were left on the 3.3V and 5V lines).  I ran the 
following tests (a minimum 100 cycles each) w/o seeing any lockups.

1- Disconnect/connect main power
2- Disconnect/connect main power w 1.8V monitor out disabled.
3- Use switch to disconnect/connect 1.8V (3.3V and 5v on continuously)
4- Use switch to disconnect/connect 1.8V (3.3V and 5v on continuously), 
1.8V monitor out disabled.
5- Repeat test 1 and 2 with a weaker pullup on the reset line.

I also ran test 1 on a Nohau evaluation board I have.

My test used the serial port rather than a blinky LED and sent an audible 
BEL so I could do the switching with all of my visual attention and just 
listen for the tone.

Both boards (mine and Nohau's) are 4 layer boards with linear supplies.  I 
have power monitor circuitry monitoring all the power lines(5v, 3.3V and 
1.8v) with 4K7 pullup to 5V (I don't trust RC reset circuits to be 
reliable).  Nohau appears to have a simple pull-up (3+ K effective) to 
3.3V.  Strangely I can't find a cap on the Nohau circuit, I suspect I've 
just overlooked it but the CD they sent containing the schematic is no 
longer readable (it appears to have slowly deteriorated). Maybe someone 
else can check for it?

Some rise times

Nohau eval board (2106)
reset - ~320uS
1.8V - ~1200uS (to stable) ~110uS rise
3.3V - ~350uS (to stable) ~230uS rise

My board (2104)

reset - ~1.1uS (That's with the weaker 15K pullup, the stronger pullup is 
probably a little faster).  It's also well after the monitor supplies have 
stabilized since the reset is help for mS.

1.8V - ~620uS (to stable) ~390 rise
3.3V - ~560uS (to stable) ~470 rise

With switch on the 1.8V line
1.8V - ~27uS (to stable) ~12uS rise

Tsvetvan, if your power rise times are in fact in the multiple mS range 
that would suggest they are quite slow in comparison.  It might be 
interesting to know how clean the reset line was on startup.

Of course I may just be (un)lucky here but I also had to replace the micro 
a couple of time during this (my errors while modifying the board) and none 
of the ones I used showed a problem.

I hope this is of some help Tsvetvan.  I'm eager to hear what you find.

Robert

" 'Freedom' has no meaning of itself.  There are always restrictions,
be they legal, genetic, or physical.  If you don't believe me, try to
chew a radio signal. "

                         Kelvin Throop, III