[sdiy] Strange uController behaviour (slightly OT)

[Alex Stettler]

Hello!


Many thanks to everybody giving me his advice/experiences on strange
uController behavior!

I finally was able the get the whole circuit running as perfect as our Swiss
watches ;-))

I added pull-down resistors to the I/O-lines with the long unconnected
wires, as I can't simply remove them (they are later connected to another
module).
Now the uC works absolutely correct even with black tape on the window, as
this normally should be the case.

It seems, that the light falling on the chip in the earlier setup was
cancelling the effect of noise picked up through the long unconnected wires.

So you see: The more complex a digital circuit gets and the more modules/pcb
are involved in the circuit the more you have to look at analog problems
like induction from one data-line to another, 50/60 Hz mains noise, noise
picked up through lines even if they are configured as output, etc.

Regards,
Alex

-----Original Message-----
From: owner-synth-diy at dropmix.xs4all.nl
[mailto:owner-synth-diy at dropmix.xs4all.nl]On Behalf Of Roman
Sent: Friday, May 16, 2003 5:01 PM
To: Alex Stettler
Cc: synth-diy at dropmix.xs4all.nl
Subject: Re: [sdiy] Strange uController behaviour (slightly OT)


I had once an unexplained, but repeatable strange behaviour
of PIC16F84. It had MIDI output constantly LOW after power-up,
and it came back to normal after first note sent.
There was nothing to justify it in the firmware.

It turns out the micro had long wire connected to open drain
pin, so I guess interferences came into the device completely
fooling it. It happens before even power is on, so it doesn't
matter if the pin is configured afterwards as output or input.
It has enought time to go crazy.

The fix was to add pull-up resistor at this open-drain pin.
So yes, maybe pull-down in your situation might also help.

I say - get rid of those long wires, or use buffers.
This thing with covering EE window might be coincidence,
or the black tape you use generates so much static, that
together with those long wires it's possible to change
normal behaviour of your micro.

Roman

---- Wiadomość Oryginalna ----
Od: Alex Stettler <alex at xatomic.ch>
Do: DIY Synth List <synth-diy at dropmix.xs4all.nl>
Data: Fri, 16 May 2003 01:05:11 +0200
Temat: [sdiy] Strange uController behaviour (slightly OT)

>Hello!
>
>
>I'm just programming an older MC68705P3 uController in a self developped
>circuit. The MC68705 is an 8-bit EPROM uController.
>The uController has an LCD, some keys and two ADCs around it.
>The strange behaviour of the uController is, that if the EPROM-window is
>covered with opaque tape as this is usually done with EPROM-devices, the uC
>works more or less correct with only the LCD having nonsense characters on
>the screen. As soon as I remove the black tape, the uC works perfect!
>So I did some further experiments to decrease the number of possible
>error-sources and experienced the following behaviour:
>
>1. If there is opaque tape on the ERPOM-window only plays a role at
>power-up. As soon power is applied and the uC started working, it works
>correct, with or without a covered EPROM-window.
>
>2. There are four 20cm long I/O lines on a socket, which aren't connected
to
>anything yet. Just empty wires coming out of the uC PCB. Now, they have a
>really strange effect on the whole behaviour: With the leads the uC has
>often problems with proper starting, although the software sets them as
>outputs right at the beginning of the program. If I deconnect the unused
>leads from the uC PCB, the uC works absolut perfectly. If the EPROM-windows
>is covered or not doesn't play any role.
>
>So my questions are:
>-Why does a uC work better with light instead without?
>-Why do unconnected leads affect the behaviour of a uC, even if these lines
>are configured as OUTPUTS??
>-Could a pulldown-resistor help the uC to get back to normal operation?
>-Has anyone else made strange experiences with MC68705 or similar
>microcontrollers?
>
>Probably it would make a trendy looking device: A plexiglas casing and some
>blue or white coloured LEDs lighting the EPROM-window of the MCU ;-))
>
>But I'd rather prefer normal operation of the MCU.
>
>Many thanks in advance for any help!
>
>Kind regards,
>Alex
>
>
>
>**************************************
>
>Alex Stettler
>Student Elektrotechnik / Elektronik vz
>
>Hochschule für Technik und Architektur
>Burgdorf
>
>Mobile: +41 79 434 87 03
>Fax   : +41 61 901 64 04
>
>