my damn mail provider says causes trouble,
yahoo says mail undeliverable - bouncing mails (or something similar).

i missed several messages and have to read them online..

hope it works now..

sorry for messing up the list but i must try.

ST

Posts are getting randomly delayed. If you post something and it
doesn't show up within a few minutes, please give it a while longer
(couple of hours). Its getting confusing seeing multiple variants of
the same posting.

--- In Electronics_101@..., "ghidera2000"
<ghidera2000@y...> wrote:

I'm trying to figure out at what voltage (reverse) would a 1n4148
shut off current flowing through it. e.g. 5 volt supply being
opposed by a 3.3V reverse voltage.

On the datasheet, I see Vfr - (Forward voltage recovery) at 2.5V.
Would that mean that it needs 2.5 or greater volt difference
between anode and cathode before current can flow?

Zero volts and any reverse voltage up to the maximum for a 1N4148 wil
stop current flowing though it (except for the leakage current). A
small positive voltage will produce a small current flow. For
example, at room temperature you will get 1uA of forward current at
300mV of forward voltage. At 0.675V it will be around 1mA.

Keith.

--- In Electronics_101@..., "sd7008us" <sd7008us@y...> wrote:

Hi All,

It turns out, I only need 600ns delay on the interrupt pulse. What
is the best method to delay the second input signal to the XOR gate?
There is PLD gates avilable in the MCU. Would a series of "AND" does
the job?

You only need a delay that is long enough to produce the interrupt
pulse width required to trigger the microprocessor. If that is say
250ns, you need to delay by at least that. A resistor to one input of
the XOR gate and a capacitor to ground on that input would do it. A
few AND gates will probably be a bit too quick and produce a rather
short interrupt pulse.

Can you select in software which edge the interrupt triggers on?
Could you debounce the switch in software?

Keith.

At the risk of being branded a microcontroller bigot (I'm not,
really!), why not do this solely in the mcu? Debounce is pretty easy
(most bouncing occurs for about 1-2 mS, according to Horowitz/Hill)
in software. Of course if your MCU is busy doing other things, you
may not be able to watch the switch.

Your dual interrupt idea works though you would need an external
inverter. Can you trigger on leading or training edges? You might
be able to get away with bringing the switch into a port and watching
the transitions on it. If the responsiveness needs only to be in the
100s of mS range, you could poll the pin every 50-100 uS and use a
little state machine to determine switch state (i.e. debounced).

External debounce can also be done with a resistor and a cap. H/H
has a discussion of this. Sorry, I'm too lazy to look up the page.

As usual, there are at least 100 ways to skin this cat.

Phil

You might look at the 4528 CMOS multivibrator - it can
be triggered on either a positive or negative edge.
Alternatively, you could make two half monostables and
OR them together. Don Lancaster, in hid book "CMOS
Cookbook" has a pretty good explanation of half
monostables. (4528 would be easier.)

Jim
--- sd7008us <sd7008us@...> wrote:

Hi All,

I need to generate an interrupt pulse on every edge
(rising or
falling) of an input signal to a MCU. Any idea how
to do this?

Here is the waveform:

foo:
--------______--------___________--------

interrupt:
-------__-----__------__---------__------


I do have the clock avilable. The interrup pulse has
to be at least
200ms wide.

Regards,

SD

__________________________________
Do you Yahoo!?
Yahoo! Small Business $15K Web Design Giveaway

开云体育

开云体育

开云体育

I tried to respond to the post, but has not apeared on the messages
board yet.

Anyway, it turns out I only need 600ns interrupts pluse. Any idea
what is the best way to delay the second input of the XOR gate? The
MCU we are using has PLD gates that we can program. Would a series
of ANDs do the job?

Any ideas??

Regards,

SD

Hi All,

It turns out, I only need 600ns delay on the interrupt pulse. What
is the best method to delay the second input signal to the XOR gate?
There is PLD gates avilable in the MCU. Would a series of "AND" does
the job?

Regards,

SD

--- In Electronics_101@..., "Keith" <keith@k...> wrote:

--- In Electronics_101@..., "sd7008us" <sd7008us@y...>

wrote:

Hi All,

I need to generate an interrupt pulse on every edge (rising or
falling) of an input signal to a MCU. Any idea how to do this?

Here is the waveform:

foo: --------______--------___________--------

interrupt: -------__-----__------__---------__------


I do have the clock avilable. The interrup pulse has to be at

least

200ms wide.

Regards,

SD

SD

An exclusive OR gate with the input signal to one side delayed. Do
you really want a 200ms interrupt pulse? That is one seriously

slow

processor.

Keith.

Thanks all for the responds. How about we doing the following way?
Would it consider a workaround (not a clean way)?

The foo signal is coming from a push button and we need to interrupt
the MCU every time it changes value. There are two interrupts pins
to the MCU. We also need to debounce the signal after the first
interrupt that is why we need 200ms on the pulse. Here is what we
think will work: we feed the "foo" signal directly to the first
interrupt (Int0) and trigger it on the Negative edge. Then, we
inverted the "foo" and feed the output to the second interrupt
(Int1). The fireware will have to do the extra worrk to track the
states of "foo".

Here is how the waveform would look:

====================================================================

foo: -----------________----------_________-----------
Int0 Int0

/foo: ________----------________------------________
Int1 Int1
=====================================================================

This method only need one interal "PLD" inverter and no external
parts needed.

Any thoughts on this?

Regards,

SD



--- In Electronics_101@..., Stefan Trethan
<stefan_trethan@g...> wrote:

An exclusive OR gate would do this by delaying the signal to one

input

with a resistor and capacitor (and schmitt trigger if the delay

is

long). However, can you not switch the interrupt trigger edge

over

each time you detect an edge? Also, a 200ms interrupt seems very

long.

Keith.

yes that is a way better idea than the monoflop...

ST

Though I am a proponent of using microcontrollers in appropriate
problem solutions, this problem is not an appropriate use of them.

A quad 2-1 mux will solve the problem. It is much faster than a
processor, cheaper, and does not require programming. Use i1 to select
between the two four bit outputs. EN (chip enable) should be on all
the time (held low) since there was no mention in the problem of
tri-stating or busses.


Click on datasheet.

Also, if you are curious about how it would be done in discrete logic,
please refer to the logic diagram in the data sheet. (fig 5)

No, it does not turn one circuit off, that is not the way logic
circuits work. All the circuits are on all the time and certain
signals are selected or manipulated.

A processor is really useful when the problem calls for counting, any
math, state machines, key inputs; all this is easy in a processor but
not in discrete logic.

I dont disagree that the learning curve is a big barrier (as I did
point out earlier) but the benefits of soft logic and a single
package make it a totally worthwile and appropriate solution. I
would not recommend it to a beginner (especially some one who hasn't
written any code) who is trying to solve a single problem.

I would, however, recommend it to some one who is learning
electronics and wants to have a modern "tool kit" at their disposal.
An electronic techie that doesn't understand microcontrollers is at a
disadvantage. Of course, they should understand how to solve a
problem with discretes as well but I doubt they teach karnaugh maps
and all that stuff these days. Frankly, they should be versant in
boolean algebra as well - the Latin of digital electronics.

I'm an old f*rt - learned my logic with 7400 TTL (not even 74S00
series!) but LOVE the benefits of these microcontrollers to whisk
away acres (hectares? lol) of logic with a single package.

The joy of electronics is that there are always a multitude of ways
to solve the problem - one man's beauty is an other's UGLY. For
every guy that pushes a micro, there is some guy solving it with a
cap and transistor. LOL.

--- In Electronics_101@..., "Dave Mucha"
<dave_mucha@y...> wrote:

I would argue that for a novice, the micro, software, programmer
board are a barrier to getting the job done.

The person would have to learn C or ASM or use one of the free demo
(can you spell LIMITED) software packages.

THEN, would have to design a circuit for some specific chip and get
it right.

The descrete component side also requires designing the circuit and
getting it right, but it could shave days off of the project by NOT
getting involved with software.

However, if one has a simple grasp of the software and already has
the hardware (programmer) it becomes another tool.

The micro offers simple changes as it is in software.

The descrete component side offers much faster response

The Micro and descrete chips both offer expansion as rarely does

one

use all the available pins/circuits.

Dave

--- In Electronics_101@..., "Scott Thompson"
<electronguy@e...> wrote:

Just got a chance to browse the schematic.

Looks pretty comprehensive and will require further study, but

your input power connector is wired so that it shorts out the
battery.

Thanks for sharing,
Scotty

Doh! I put the battery+ to ground instead of the supply... Good
spotting ;)

I've actually done a lot of thinking about this circuit and I've
come to the conclusion that I have no idea how to effectively catch
the average peak vibration and non-average peak spikes.

I was thinking of putting a precision rectifier in the sensor box,
adding gain in the monitor box then splitting the signal into two.
Once would go straight to the ADC for measuring the highest spikes
while the other would go to a capacitor for electrical averaging of
the peaks. That way I can do the fast sampling for spikes without
the overhead of peak value discovery and averaging (20 samples per
wave thing) and just a single sample to find the average peak.

Sounds good in theory out now I have to figure out how to do it 8).

Probably use an OPA2340 to split the (rectified) signal into two and
add gain, then experiment from there.

Oh and, I did some experimenting with PWM signals through a long
cable. Without a resistor added, the signal was distorted. I added a
1K resistor and the signal came out a lot cleaner. Think I'm getting
a grasp of this cable capacitance problem. Its not so much added
noise from the enviroment as it is the cable capacitance distorting
the signal thats already there. Square waves turn into rounded
corner waves and the amplitude bounces around. Less current means
less capacitance effect, cleaner signal.

I'm trying to figure out at what voltage (reverse) would a 1n4148
shut off current flowing through it. e.g. 5 volt supply being
opposed by a 3.3V reverse voltage.

On the datasheet, I see Vfr - (Forward voltage recovery) at 2.5V.
Would that mean that it needs 2.5 or greater volt difference
between anode and cathode before current can flow?

I would argue that for a novice, the micro, software, programmer
board are a barrier to getting the job done.

The person would have to learn C or ASM or use one of the free demo
(can you spell LIMITED) software packages.

THEN, would have to design a circuit for some specific chip and get
it right.

The descrete component side also requires designing the circuit and
getting it right, but it could shave days off of the project by NOT
getting involved with software.

However, if one has a simple grasp of the software and already has
the hardware (programmer) it becomes another tool.

The micro offers simple changes as it is in software.

The descrete component side offers much faster response

The Micro and descrete chips both offer expansion as rarely does one
use all the available pins/circuits.

Dave

PICs are an emotive subject. For a small IC, they seem to stir up a
great amount of feeling There is not doubt that Micropchip have done
a great marketing job to get them so well recognised and into
applications that they should never be in.

As far as many 'old school' designers are concerned, they are
overused and a cumbersome way of replacing simple logic at times.
Some uses are much like using a wrench to hammer a nail - it gets the
job done, but it's not nice. Stefan suggested a variety of
programmable devices that are far better suited to the application.

However, given the problem, I would choose to use a PIC for a
solution - not because it's the most elegant way or the cheapest or
the the most reliable for that matter. I would use one because I
have the tools to produce one quickly, I have the experience to
programme it and I can comlete the programme in the same time I would
spend looking up order codes for the logic devices to do the same
job. (I would also do it without the dumb logic mistake I announced
to the group a day or two back on this same subject) While that makes
it the best solution for me to use, I would agree with Stefan's
comment that PICs are not the holy grail. As far as the end user is
concerned, he probably cares very little that the logic takes 30uS to
process with a PIC rather than a 100nS or so using logic gates. From
his point of view, both solutions are equally good.

Engineers the world over have passionate opinions about their ideas
and solutions and we need to forgive each other for them sometimes.
It is this same passion that gives us all the chance to see other
ideas and new solutions. It's also worth remembering that many here
are using English as a second language. Sometimes, native English
speakers can see an abrupt comment that was never intended. I am
sure this won't be the last time feather sget a little ruffled on
both sides of the debate


--- In Electronics_101@..., Stefan Trethan
<stefan_trethan@g...> wrote:

On Wed, 07 Apr 2004 22:59:04 -0000, Phil <phil1960us@y...> wrote:

I see no need to be so snide.

Sorry, sometimes i get easily agitated about this topic.


of course you can use a pic when you want .....

st

An exclusive OR gate would do this by delaying the signal to one input
with a resistor and capacitor (and schmitt trigger if the delay is
long). However, can you not switch the interrupt trigger edge over
each time you detect an edge? Also, a 200ms interrupt seems very long.

Keith.

yes that is a way better idea than the monoflop...

ST

--- In Electronics_101@..., "sd7008us" <sd7008us@y...> wrote:

Hi All,

I need to generate an interrupt pulse on every edge (rising or
falling) of an input signal to a MCU. Any idea how to do this?

Here is the waveform:

foo: --------______--------___________--------

interrupt: -------__-----__------__---------__------


I do have the clock avilable. The interrup pulse has to be at least
200ms wide.

Regards,

SD

SD,

An exclusive OR gate would do this by delaying the signal to one input
with a resistor and capacitor (and schmitt trigger if the delay is
long). However, can you not switch the interrupt trigger edge over
each time you detect an edge? Also, a 200ms interrupt seems very long.

Keith.