Dedicated debouncer IC...

[Phil Hobbs]

On 2020-08-13 05:12, Tom Gardner wrote:

On 13/08/20 08:59, Phil Hobbs wrote:
(Dave is a lot easier to take when run at 1.75x speed on Youtube.)

2x, closed captions, speakers off

But then I dislike 99.9% of the electronics videos since
I can read articles much faster than they can speak.

I agree, except once in awhile Mike Harrison has something fun about his
installation art, or Big Clive tears down some horrifically dangerous
Chinese power supply or shower heater. All are better at 1.75x.

These ones walk you through using an unfamiliar toolchain with all sorts
of gotchas along the way. And c\'mon, a micro that costs 3 cents in
onesies and has a standards-compliant C compiler! You\'ve got to love
that. I skipped through them, but there were several places where one
could easily have wasted some hours before figuring it out. (Which we\'ve
all done, at least those of us who use micros.) The Flying Dutchman
will deny this, of course, but he\'s full of BS.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC / Hobbs ElectroOptics
Optics, Electro-optics, Photonics, Analog Electronics
Briarcliff Manor NY 10510

http://electrooptical.net
http://hobbs-eo.com

 

[Tom Gardner]

On 13/08/20 16:00, Phil Hobbs wrote:

On 2020-08-13 05:12, Tom Gardner wrote:
On 13/08/20 08:59, Phil Hobbs wrote:
(Dave is a lot easier to take when run at 1.75x speed on Youtube.)

2x, closed captions, speakers off

But then I dislike 99.9% of the electronics videos since
I can read articles much faster than they can speak.


I agree, except once in awhile Mike Harrison has something fun about his
installation art, or Big Clive tears down some horrifically dangerous Chinese
power supply or shower heater.  All are better at 1.75x.

These ones walk you through using an unfamiliar toolchain with all sorts of
gotchas along the way.  And c\'mon, a micro that costs 3 cents in onesies and has
a standards-compliant C compiler!  You\'ve got to love that.  I skipped through
them, but there were several places where one could easily have wasted some
hours before figuring it out. (Which we\'ve all done, at least those of us who
use micros.)  The Flying Dutchman will deny this, of course, but he\'s full of
BS.

They are the 0.1%, of course!

I\'m sure you haven\'t forgotten Shahriar\'s The Signal Path,
which is more than a talking head and *2 doesn\'t work for
the audio nor for the captions

 

[Phil Hobbs]

On 2020-08-13 06:24, Don Y wrote:

On 8/13/2020 2:43 AM, Pimpom wrote:
On 8/13/2020 12:18 PM, Don Y wrote:
On 8/12/2020 11:22 PM, Pimpom wrote:
On 8/12/2020 9:13 PM, jlarkin@highlandsniptechnology.com wrote:
Why do you need to debounce?

Nothing specific at the moment. I just want to have a stock
standalone solution
for future applications.

So, you\'re going to FIX the cost and capabilities of that \"solution\"
without concern for the needs of those \"future applications\"???

shakes head

I don\'t understand the head shaking. Anyone who designs circuits
regularly will have their favourite solution for a common task. I\'ve
used various debouncing techniques but they all require passive
components, a particular type of switch, and in some cases another
active device to complement the main active devices.

If there\'s an inexpensive standalone part that needs no external parts
and can replace those techniques in most applications, what\'s wrong
with using that as a stock solution? A 2N3904/BC547 transistor can be
used at the heart of a wide range of low frequency low power
applications. Why hunt for a suitable type every time?

What do you do when you need a greater Icc?  Do you have your \"favorite
FET\"?

I sure do. Besides the late lamented BF862 (of which I have several
reels), there\'s the CPH3910 and my new fave, the Mini Circuits SAV-551+,
which (with a bit of circuit magic) makes the world\'s finest bootstraps.

\"Diode\"?  \"Switch\"?

Do you use them regardless of what the *particular* requirements and
constraints of the design happen to be?

You should be reusing ideas, not implementations.  So, for example, a
particular amplifier topology -- not a particular set of components to
implement that particular amplifier instance.

That\'s needlessly dogmatic. Engineering time is very valuable, and
there are lots of noncritical jobs where cut-n-paste makes perfect
sense. For instance, a bunch of my small instruments run off our
standard 24V wall wart and use two buck regulators to make +12-14V and
-16-20V. That\'s generally an LMR23630 running at 2 MHz and an AOZ1282CI
respectively. (The AOZ runs off the 23630\'s output so we can have more
negative headroom.) They\'re reasonably cheap, work well, and we know
how to keep the EMI low enough for use in small form factor
ultrasensitive instruments. So power and EMI are solved problems for
the most part.

We have two other tiles for a very quiet Class H TEC driver, one just
for cooling detectors and one that can heat as well, for controlling
diode lasers. Another tile is a bias generator for MPPCs and APDs.
That has to be extremely quiet as well, because it\'s connected directly
to the summing junction of the TIA via the photodiode capacitance.

You develop a preference for a particular component when you find yourself
using it in many SIMILAR applications.  But, that doesn\'t stop you from
reevaluating whether it is APPROPRIATE for a specific NEW application!
You risk limiting your solutions to embrace past experiences instead of
current needs.

And lots of folks code by cut-and-paste from Stack Overflow. That
doesn\'t work well either, so don\'t do either of those things.

(What do you do when you have to debounce 200 pushbuttons -- like on a
video switcher?  Do you use the same components -- and techniques! -- that
you did when debouncing ONE?  If so, was the \"one\" solution inefficient
as you couldn\'t gain any design economies by scaling up??  Or, if you\'re
tasked with debouncing a membrane keypad?  Or, told to delay the switch
action until after a specific period?)

I\'m pretty sure the OP isn\'t going to be needing to do any of those
things any time soon. He\'s a pretty good guy, but AFAICT he\'s retired,
and he certainly lives in the back of beyond.

There are countless ways to debounce switches.  Thinking that you can pick
one and be anywhere close to \"ideal\" is delusional.

(Hey, why not use mercury wetted contacts on all of your switches -- for
all eternity -- and not sweat the bounce?)

Of course one might ask why you\'re wasting your own valuable time
tearing a strip off the OP for presuming to know his own likely
requirements better than you do.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC / Hobbs ElectroOptics
Optics, Electro-optics, Photonics, Analog Electronics
Briarcliff Manor NY 10510

http://electrooptical.net
http://hobbs-eo.com

 

[Lasse Langwadt Christensen]

torsdag den 13. august 2020 kl. 17.00.19 UTC+2 skrev Phil Hobbs:

On 2020-08-13 05:12, Tom Gardner wrote:
On 13/08/20 08:59, Phil Hobbs wrote:
(Dave is a lot easier to take when run at 1.75x speed on Youtube.)

2x, closed captions, speakers off

But then I dislike 99.9% of the electronics videos since
I can read articles much faster than they can speak.


I agree, except once in awhile Mike Harrison has something fun about his
installation art, or Big Clive tears down some horrifically dangerous
Chinese power supply or shower heater. All are better at 1.75x.

isn\'t Mike already at 1.75x ?

 

[Phil Hobbs]

On 2020-08-13 11:18, Tom Gardner wrote:

On 13/08/20 16:00, Phil Hobbs wrote:
On 2020-08-13 05:12, Tom Gardner wrote:
On 13/08/20 08:59, Phil Hobbs wrote:
(Dave is a lot easier to take when run at 1.75x speed on Youtube.)

2x, closed captions, speakers off

But then I dislike 99.9% of the electronics videos since
I can read articles much faster than they can speak.


I agree, except once in awhile Mike Harrison has something fun about
his installation art, or Big Clive tears down some horrifically
dangerous Chinese power supply or shower heater.  All are better at
1.75x.

These ones walk you through using an unfamiliar toolchain with all
sorts of gotchas along the way.  And c\'mon, a micro that costs 3 cents
in onesies and has a standards-compliant C compiler!  You\'ve got to
love that.  I skipped through them, but there were several places
where one could easily have wasted some hours before figuring it out.
(Which we\'ve all done, at least those of us who use micros.)  The
Flying Dutchman will deny this, of course, but he\'s full of BS.

They are the 0.1%, of course!

I\'m sure you haven\'t forgotten Shahriar\'s The Signal Path,
which is more than a talking head and *2 doesn\'t work for
the audio nor for the captions

I\'m not that big a Signal Path fan. Besides Shahriar having a classical
Arabian Nights name (Scheherezade\'s husband), he\'s a smart guy who
speaks well and talks about interesting stuff sometimes. He often seems
to spend a lot of time on topics that I find old hat, though.

Once in a great while I\'ll watch one of Paul Carlson\'s repair videos,
also at 1.75x. He finds some of the craziest old gear and restores it
beautifully, for example an all-in-one bench gizmo from 1939, the
\'Supreme Vedolyzer\'. With a name like that it\'s gotta be good, right?

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC / Hobbs ElectroOptics
Optics, Electro-optics, Photonics, Analog Electronics
Briarcliff Manor NY 10510

http://electrooptical.net
http://hobbs-eo.com

 

[Phil Hobbs]

On 2020-08-13 11:34, Lasse Langwadt Christensen wrote:

torsdag den 13. august 2020 kl. 17.00.19 UTC+2 skrev Phil Hobbs:
On 2020-08-13 05:12, Tom Gardner wrote:
On 13/08/20 08:59, Phil Hobbs wrote:
(Dave is a lot easier to take when run at 1.75x speed on Youtube.)

2x, closed captions, speakers off

But then I dislike 99.9% of the electronics videos since
I can read articles much faster than they can speak.


I agree, except once in awhile Mike Harrison has something fun about his
installation art, or Big Clive tears down some horrifically dangerous
Chinese power supply or shower heater. All are better at 1.75x.

isn\'t Mike already at 1.75x ?

In bursts, but not on average. Okay, maybe 1.5x for Mike, 1.75 for
everyone else. I remember when he was a regular on SED, back in the
Eeyore era.

Speaking of disappearances, DLUNU hasn\'t been seen here for months.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC / Hobbs ElectroOptics
Optics, Electro-optics, Photonics, Analog Electronics
Briarcliff Manor NY 10510

http://electrooptical.net
http://hobbs-eo.com

 

[Lasse Langwadt Christensen]

torsdag den 13. august 2020 kl. 17.35.19 UTC+2 skrev Phil Hobbs:

On 2020-08-13 11:18, Tom Gardner wrote:
On 13/08/20 16:00, Phil Hobbs wrote:
On 2020-08-13 05:12, Tom Gardner wrote:
On 13/08/20 08:59, Phil Hobbs wrote:
(Dave is a lot easier to take when run at 1.75x speed on Youtube.)

2x, closed captions, speakers off

But then I dislike 99.9% of the electronics videos since
I can read articles much faster than they can speak.


I agree, except once in awhile Mike Harrison has something fun about
his installation art, or Big Clive tears down some horrifically
dangerous Chinese power supply or shower heater.  All are better at
1.75x.

These ones walk you through using an unfamiliar toolchain with all
sorts of gotchas along the way.  And c\'mon, a micro that costs 3 cents
in onesies and has a standards-compliant C compiler!  You\'ve got to
love that.  I skipped through them, but there were several places
where one could easily have wasted some hours before figuring it out.
(Which we\'ve all done, at least those of us who use micros.)  The
Flying Dutchman will deny this, of course, but he\'s full of BS.

They are the 0.1%, of course!

I\'m sure you haven\'t forgotten Shahriar\'s The Signal Path,
which is more than a talking head and *2 doesn\'t work for
the audio nor for the captions

I\'m not that big a Signal Path fan. Besides Shahriar having a classical
Arabian Nights name (Scheherezade\'s husband), he\'s a smart guy who
speaks well and talks about interesting stuff sometimes. He often seems
to spend a lot of time on topics that I find old hat, though.

Once in a great while I\'ll watch one of Paul Carlson\'s repair videos,
also at 1.75x. He finds some of the craziest old gear and restores it
beautifully, for example an all-in-one bench gizmo from 1939, the
\'Supreme Vedolyzer\'. With a name like that it\'s gotta be good, right?

if you just want a quick look inside all kinds of weird old gear, this
guy has plenty, https://youtu.be/uuLP8-XnBxE

 

[Don Y]

On 8/13/2020 4:34 AM, Pimpom wrote:

On 8/13/2020 3:54 PM, Don Y wrote:
There are countless ways to debounce switches. Thinking that you can pick
one and be anywhere close to \"ideal\" is delusional.

(Hey, why not use mercury wetted contacts on all of your switches -- for
all eternity -- and not sweat the bounce?)

I don\'t know if you\'re deliberately twisting my words or you really don\'t
understand. Having a favourite solution for a class of applications does not
mean that it will invariably be used in *all* designs. But it\'s good to be able
to cover a wide range of applications with an inexpensive easy-to-use part - a
jellybean. Is that so hard to understand?

Then it should be a cheap MCU -- by YOUR criteria. Write the code once
and have the disti program the devices for you. Put them in a little
bin and leave them there until needed.

Same code (algorithm) will run in other MCUs so you\'re not dependant on an
oddball chip being available into the future (like a \"hex switch debouncer\")

There\'s one particular size and pitch of screws that I use a lot. It doesn\'t
mean that I never use screws of another size - or a different type of fastener
- when they\'re called for.

 

[server]

On Thu, 13 Aug 2020 11:24:05 -0400, Phil Hobbs
<pcdhSpamMeSenseless@electrooptical.net> wrote:

On 2020-08-13 06:24, Don Y wrote:
On 8/13/2020 2:43 AM, Pimpom wrote:
On 8/13/2020 12:18 PM, Don Y wrote:
On 8/12/2020 11:22 PM, Pimpom wrote:
On 8/12/2020 9:13 PM, jlarkin@highlandsniptechnology.com wrote:
Why do you need to debounce?

Nothing specific at the moment. I just want to have a stock
standalone solution
for future applications.

So, you\'re going to FIX the cost and capabilities of that \"solution\"
without concern for the needs of those \"future applications\"???

shakes head

I don\'t understand the head shaking. Anyone who designs circuits
regularly will have their favourite solution for a common task. I\'ve
used various debouncing techniques but they all require passive
components, a particular type of switch, and in some cases another
active device to complement the main active devices.

If there\'s an inexpensive standalone part that needs no external parts
and can replace those techniques in most applications, what\'s wrong
with using that as a stock solution? A 2N3904/BC547 transistor can be
used at the heart of a wide range of low frequency low power
applications. Why hunt for a suitable type every time?

What do you do when you need a greater Icc?  Do you have your \"favorite
FET\"?

I sure do. Besides the late lamented BF862 (of which I have several
reels), there\'s the CPH3910 and my new fave, the Mini Circuits SAV-551+,
which (with a bit of circuit magic) makes the world\'s finest bootstraps.

SAV-551 is a great switch to ground too. 2 ohms, half a pF. Good for
steering current, discharging ramps, changing gain, or switching caps
in and out. It only needs a half volt or so of gate swing, but you can
drive it from bigger stuff with just a series gate resistor.

The data sheets are pretty good, for an RF part. I have supplemental
measurements, like Rds-on.

\"Diode\"?  \"Switch\"?

Do you use them regardless of what the *particular* requirements and
constraints of the design happen to be?

You should be reusing ideas, not implementations.  So, for example, a
particular amplifier topology -- not a particular set of components to
implement that particular amplifier instance.

That\'s needlessly dogmatic. Engineering time is very valuable, and
there are lots of noncritical jobs where cut-n-paste makes perfect
sense. For instance, a bunch of my small instruments run off our
standard 24V wall wart and use two buck regulators to make +12-14V and
-16-20V. That\'s generally an LMR23630 running at 2 MHz and an AOZ1282CI
respectively. (The AOZ runs off the 23630\'s output so we can have more
negative headroom.) They\'re reasonably cheap, work well, and we know
how to keep the EMI low enough for use in small form factor
ultrasensitive instruments. So power and EMI are solved problems for
the most part.

Nothing wrong with cutting and pasting chunks of a schematic, but
you\'ve got to be careful about things like mixing 0603s and 0805s
accidentally, or net names, or part library changes. Just check.



--

John Larkin Highland Technology, Inc

Science teaches us to doubt.

Claude Bernard

 

[Phil Hobbs]

On 2020-08-13 12:06, jlarkin@highlandsniptechnology.com wrote:

On Thu, 13 Aug 2020 11:24:05 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

On 2020-08-13 06:24, Don Y wrote:
On 8/13/2020 2:43 AM, Pimpom wrote:
On 8/13/2020 12:18 PM, Don Y wrote:
On 8/12/2020 11:22 PM, Pimpom wrote:
On 8/12/2020 9:13 PM, jlarkin@highlandsniptechnology.com
wrote:
Why do you need to debounce?

Nothing specific at the moment. I just want to have a stock
standalone solution for future applications.

So, you\'re going to FIX the cost and capabilities of that
\"solution\" without concern for the needs of those \"future
applications\"???

shakes head

I don\'t understand the head shaking. Anyone who designs
circuits regularly will have their favourite solution for a
common task. I\'ve used various debouncing techniques but they
all require passive components, a particular type of switch,
and in some cases another active device to complement the main
active devices.

If there\'s an inexpensive standalone part that needs no
external parts and can replace those techniques in most
applications, what\'s wrong with using that as a stock
solution? A 2N3904/BC547 transistor can be used at the heart of
a wide range of low frequency low power applications. Why hunt
for a suitable type every time?

What do you do when you need a greater Icc? Do you have your
\"favorite FET\"?

I sure do. Besides the late lamented BF862 (of which I have
several reels), there\'s the CPH3910 and my new fave, the Mini
Circuits SAV-551+, which (with a bit of circuit magic) makes the
world\'s finest bootstraps.


SAV-551 is a great switch to ground too. 2 ohms, half a pF. Good for
steering current, discharging ramps, changing gain, or switching
caps in and out. It only needs a half volt or so of gate swing, but
you can drive it from bigger stuff with just a series gate resistor.

Have you tried the new 18-GHz Renesas pHEMT parts as switches? They
wouldn\'t have much chance to oscillate in that service.

The data sheets are pretty good, for an RF part. I have supplemental
measurements, like Rds-on.

The e_N*C noise goes to zero at DC faster than the 1/f noise rises, so
even though the pHEMT has a lot more than the JFETs, its lower flatband
noise wins at bandwidths above around 5-10 MHz. Plus it has twice the
transconductance.

\"Diode\"? \"Switch\"?

Do you use them regardless of what the *particular* requirements
and constraints of the design happen to be?

You should be reusing ideas, not implementations. So, for
example, a particular amplifier topology -- not a particular set
of components to implement that particular amplifier instance.

That\'s needlessly dogmatic. Engineering time is very valuable, and
there are lots of noncritical jobs where cut-n-paste makes perfect
sense. For instance, a bunch of my small instruments run off our
standard 24V wall wart and use two buck regulators to make +12-14V
and -16-20V. That\'s generally an LMR23630 running at 2 MHz and an
AOZ1282CI respectively. (The AOZ runs off the 23630\'s output so we
can have more negative headroom.) They\'re reasonably cheap, work
well, and we know how to keep the EMI low enough for use in small
form factor ultrasensitive instruments. So power and EMI are
solved problems for the most part.

Nothing wrong with cutting and pasting chunks of a schematic, but
you\'ve got to be careful about things like mixing 0603s and 0805s
accidentally, or net names, or part library changes. Just check.

My cut-and-paste is done with scissors and rubber cement. The ref
des get redone at BOM compilation time--I number the parts as I enter
them into the BOM, which reduces the number of blunders that have to be
caught later.

The net names are a genuine issue, true.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC / Hobbs ElectroOptics
Optics, Electro-optics, Photonics, Analog Electronics
Briarcliff Manor NY 10510

http://electrooptical.net
http://hobbs-eo.com

 

[Pimpom]

On 8/13/2020 9:34 PM, Don Y wrote:

On 8/13/2020 4:34 AM, Pimpom wrote:
On 8/13/2020 3:54 PM, Don Y wrote:
There are countless ways to debounce switches. Thinking that you can pick
one and be anywhere close to \"ideal\" is delusional.

(Hey, why not use mercury wetted contacts on all of your switches -- for
all eternity -- and not sweat the bounce?)

I don\'t know if you\'re deliberately twisting my words or you really don\'t
understand. Having a favourite solution for a class of applications does not
mean that it will invariably be used in *all* designs. But it\'s good to be able
to cover a wide range of applications with an inexpensive easy-to-use part - a
jellybean. Is that so hard to understand?

Then it should be a cheap MCU -- by YOUR criteria. Write the code once
and have the disti program the devices for you. Put them in a little
bin and leave them there until needed.

\"should\"? There ya go. You have your own favourite solution and
you\'re doing more than using it yourself: you\'re saying that
others *should* use it.

Same code (algorithm) will run in other MCUs so you\'re not dependant on an
oddball chip being available into the future (like a \"hex switch debouncer\")

I\'ve used MCUs, FFs, schmitt gates, discrete parts and other
techniques, some of them with a dual purpose, to debounce
switches and I\'ll keep using them where appropriate. I completely
fail to understand why you\'re so dead set against adding one more
technique to the collection.

 

[Don Y]

On 8/13/2020 10:46 AM, Pimpom wrote:

On 8/13/2020 9:34 PM, Don Y wrote:
On 8/13/2020 4:34 AM, Pimpom wrote:
On 8/13/2020 3:54 PM, Don Y wrote:
There are countless ways to debounce switches. Thinking that you can pick
one and be anywhere close to \"ideal\" is delusional.

(Hey, why not use mercury wetted contacts on all of your switches -- for
all eternity -- and not sweat the bounce?)

I don\'t know if you\'re deliberately twisting my words or you really don\'t
understand. Having a favourite solution for a class of applications does not
mean that it will invariably be used in *all* designs. But it\'s good to be able
to cover a wide range of applications with an inexpensive easy-to-use part - a
jellybean. Is that so hard to understand?

Then it should be a cheap MCU -- by YOUR criteria. Write the code once
and have the disti program the devices for you. Put them in a little
bin and leave them there until needed.

\"should\"? There ya go. You have your own favourite solution and you\'re doing
more than using it yourself: you\'re saying that others *should* use it.

No, I gave you a solution that will address the widest variety of conditions
based on YOUR criteria.
- Single device (*and* handles multiple switches).
- *No* external components (internal RC oscillator).
- Can be reprogrammed to drive a variety of different switch configurations
(e.g., a switch *matrix*) and different usage features (e.g., 2KRO/NKRO).
- Can invert the \"sense\" of the actual switches (NC can act as NO).
- Can convert switch-presses to pulse events (YOU define the pulse width
and whether it is coincident with the activating or releasing edge of
the switch actuation!).
- Can be replaced by some equivalent MCU to provide more outputs or safeguard
against obsolescence or leverage newer/cheaper technologies.
- Can draw (effectively) ZERO power.
- Can cost pennies.

Of course, NONE of these may be important to YOUR selection criteria...

Same code (algorithm) will run in other MCUs so you\'re not dependant on an
oddball chip being available into the future (like a \"hex switch debouncer\")

I\'ve used MCUs, FFs, schmitt gates, discrete parts and other techniques, some
of them with a dual purpose, to debounce switches and I\'ll keep using them
where appropriate. I completely fail to understand why you\'re so dead set
against adding one more technique to the collection.

*YOU* are the one who imposed the selection criteria. *YOU* didn\'t like the
cost of the MAX part -- or the old Moto part. Or, the additional external
components.

How do you know that there isn\'t an MCU in <your-favorite-switch-debouncer>?

So, what EXACTLY are the selection criteria, *now*? Is a POR circuit per
switch is acceptable -- or isn\'t it? Will you ever need to support more
than one switch/button? etc.

I advocated for learning/exploring/perfecting different *ideas* to solving
the problem and selecting the most appropriate for the task at hand.

I\'ve a colleague who thought that the \"jelly bean\" way of interfacing to
a DIP switch was via an octal buffer. He blithely went about fabricating
*80* square inches of circuit board -- and hundreds of dollars of parts -- to
provide all of the \"configuration information\" to his circuit. He was so
accustomed to this \"jellybean solution\" that he never stepped back to
consider how ridiculously inappropriate it was for his current design!

\"Gee, why don\'t you let the processor that is sitting a few inches below
your board push the values into your ASIC instead of having it read a
sh*tload of switches?!\"

 

[Pimpom]

On 8/13/2020 3:54 PM, Don Y wrote:

On 8/13/2020 2:43 AM, Pimpom wrote:
On 8/13/2020 12:18 PM, Don Y wrote:
On 8/12/2020 11:22 PM, Pimpom wrote:
On 8/12/2020 9:13 PM, jlarkin@highlandsniptechnology.com wrote:
Why do you need to debounce?

Nothing specific at the moment. I just want to have a stock standalone solution
for future applications.

So, you\'re going to FIX the cost and capabilities of that \"solution\"
without concern for the needs of those \"future applications\"???

shakes head

I don\'t understand the head shaking. Anyone who designs circuits regularly will
have their favourite solution for a common task. I\'ve used various debouncing
techniques but they all require passive components, a particular type of
switch, and in some cases another active device to complement the main active
devices.

If there\'s an inexpensive standalone part that needs no external parts and can
replace those techniques in most applications, what\'s wrong with using that as
a stock solution? A 2N3904/BC547 transistor can be used at the heart of a wide
range of low frequency low power applications. Why hunt for a suitable type
every time?

What do you do when you need a greater Icc?

Then I choose a transistor with a higher rating. That doesn\'t
mean that one should choose a different type for every mA
difference within a range covered by a single type. Do you use a
different transistor each for 1mA, 5mA, 10mA? Or a different MCU
for each design when one type can do the job efficiently and
cost-effectively?

Do you have your \"favorite FET\"?
\"Diode\"? \"Switch\"?

The 1N4148 and 2N7002 are global favourites. That doesn\'t
preclude the use of other types when they are more appropriate,
and that\'s a fact you deliberately continue to ignore.

Do you use them regardless of what the *particular* requirements and
constraints of the design happen to be?

I will not dignify that with a detailed answer.

You should be reusing ideas, not implementations. So, for example, a
particular amplifier topology -- not a particular set of components to
implement that particular amplifier instance.

You develop a preference for a particular component when you find yourself
using it in many SIMILAR applications. But, that doesn\'t stop you from
reevaluating whether it is APPROPRIATE for a specific NEW application!
You risk limiting your solutions to embrace past experiences instead of
current needs.

(What do you do when you have to debounce 200 pushbuttons -- like on a
video switcher? Do you use the same components -- and techniques! -- that
you did when debouncing ONE? If so, was the \"one\" solution inefficient
as you couldn\'t gain any design economies by scaling up?? Or, if you\'re
tasked with debouncing a membrane keypad? Or, told to delay the switch
action until after a specific period?)

There are countless ways to debounce switches. Thinking that you can pick
one and be anywhere close to \"ideal\" is delusional.

You\'re the one who\'s delusional. You pick out of thin air very
specific applications that were never said or implied to be the
target implementations for a MAX681*. That\'s what makes your
arguments so silly.

(Hey, why not use mercury wetted contacts on all of your switches -- for
all eternity -- and not sweat the bounce?)

I\'ve seldom seen such a load of extrapolated rubbish even at
s.e.d. And that\'s saying plenty.

 

[Don Y]

On 8/13/2020 11:42 AM, Pimpom wrote:

There are countless ways to debounce switches. Thinking that you can pick
one and be anywhere close to \"ideal\" is delusional.

You\'re the one who\'s delusional. You pick out of thin air very specific
applications that were never said or implied to be the target implementations
for a MAX681*. That\'s what makes your arguments so silly.

YOU DIDN\'T SPECIFY AN APPLICATION OR A SET OF SELECTION CRITERIA.
Yet, are objecting to my qualifying MY solution with same.
How is that silly?

Is your goal just to have a single chip switch debouncer that doesn\'t have
external components is considerably less than $1/switch AND IS NOT AN MCU?
Sure would have been a lot simpler if you\'d SAID that!

 

[Pimpom]

On 8/14/2020 12:23 AM, Don Y wrote:

On 8/13/2020 11:42 AM, Pimpom wrote:
There are countless ways to debounce switches. Thinking that you can pick
one and be anywhere close to \"ideal\" is delusional.

You\'re the one who\'s delusional. You pick out of thin air very specific
applications that were never said or implied to be the target implementations
for a MAX681*. That\'s what makes your arguments so silly.

YOU DIDN\'T SPECIFY AN APPLICATION OR A SET OF SELECTION CRITERIA.
Yet, are objecting to my qualifying MY solution with same.
How is that silly?

It\'s silly because you invented a host of criteria where none was
asked for. I asked for a cheaper alternative to the MAX chip. Period.

Is your goal just to have a single chip switch debouncer that doesn\'t have
external components is considerably less than $1/switch AND IS NOT AN MCU?
Sure would have been a lot simpler if you\'d SAID that!

Go through the thread again and see if you can pick out a point
where I rejected using an MCU.

A dedicated chip does have one advantage over an MCU: it does not
need programming. But that\'s not the issue. If you\'d presented
your replies, starting with the first one, in a less pejorative
and less presumptuous manner, I would have been more receptive.

 

[Don Y]

On 8/13/2020 12:25 PM, Pimpom wrote:

On 8/14/2020 12:23 AM, Don Y wrote:
On 8/13/2020 11:42 AM, Pimpom wrote:
There are countless ways to debounce switches. Thinking that you can pick
one and be anywhere close to \"ideal\" is delusional.

You\'re the one who\'s delusional. You pick out of thin air very specific
applications that were never said or implied to be the target implementations
for a MAX681*. That\'s what makes your arguments so silly.

YOU DIDN\'T SPECIFY AN APPLICATION OR A SET OF SELECTION CRITERIA.
Yet, are objecting to my qualifying MY solution with same.
How is that silly?

It\'s silly because you invented a host of criteria where none was asked for. I
asked for a cheaper alternative to the MAX chip. Period.

The ATTINY9-TS8R-DEBOUNCE* series of switch debouncers look useful and
they\'re quite AFFORDABLE at << $1 for either the *DUAL* ATTINY9-TSR8-DEBOUNCE2
or the strap-configurable *SINGLE* ATTINY9-TSR8-DEBOUNCE1. They are readily
available from distribution with no minimum order.

For those who are not familiar with the series, these are externally
simple devices that have pins for each mechanical switch and for the
debounced output. Nothing else apart from Vcc and ground. The 40ms
debouncing (with options in halfmillisecond increments from 1ms to
250ms) is done internally without any external component.

They feature extremely low power consumption (200uA) and a wide range of
operating voltages (1.8 - 5.5). They are available in a SOT23 (6) package

The strappable variant allows up to 4 different combinations of operating
configurations to be selected for the single switch input:
- active sense (open/closed, high/low)
- output inversion
- toggle capability (push on, push off)
- output timing
- pulsed operation (pulse per switch actuation)
- pulse width
- input enable/inhibit
- output enable/inhibit
Additionally, the selection of options, pulse width, debounce time can be
field-altered, /in situ/.

[Note: these options are also configurable for the dual device on a per
input/output basis]

The part is multiple sourced to ensure continued future availability.

 

[Three Jeeps]

On Wednesday, August 12, 2020 at 4:56:41 AM UTC-4, Pimpom wrote:

The MAX681* series of switch debouncers look useful but they\'re
quite pricey at >$1 for the single MAX6816 and several $$ for the
octal 6818. Are there any cheaper, easily available alternatives?

For those who are not familiar with the series, these are
externally simple devices that have pins for each mechanical
switch and for the debounced output. Nothing else apart from Vcc
and ground. The 40ms debouncing is done internally without any
external component.

From my 35 yo tips and tricks logbook:
My goto debouncer is the MC14490. needs an external cap - useful when dealing with different ranges of bounce. Don\'t know about price comparison.
For double throw switches, there is always the Set-Reset latch - a quad pack is 14043/14044 - cheap, quite effective.
There is also the R-C approach followed by a hysterisis logic gate to square up the signal and clean up any voltage transients around the upper and lower thresholds. I like using either a 74HC14 or 40106.
If you are running line into a micro, sw debounce is cheap and effective.
Good luck

 

[Sjouke Burry]

On 13.08.20 13:52, Klaus Kragelund wrote:

Why would you need debouncing anyway?

Feeding to a flip flop?

If feeding to a micro, then use sw solution

Anyway, a flip flop is more or less the same price as a cheap micro (western micros can be had for around 10 cent in volume)

CAN SOMEBODY DE-BOUNCE THIS TREAD???

 

[John Larkin]

On Thu, 13 Aug 2020 12:43:51 -0400, Phil Hobbs
<pcdhSpamMeSenseless@electrooptical.net> wrote:

On 2020-08-13 12:06, jlarkin@highlandsniptechnology.com wrote:
On Thu, 13 Aug 2020 11:24:05 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

On 2020-08-13 06:24, Don Y wrote:
On 8/13/2020 2:43 AM, Pimpom wrote:
On 8/13/2020 12:18 PM, Don Y wrote:
On 8/12/2020 11:22 PM, Pimpom wrote:
On 8/12/2020 9:13 PM, jlarkin@highlandsniptechnology.com
wrote:
Why do you need to debounce?

Nothing specific at the moment. I just want to have a stock
standalone solution for future applications.

So, you\'re going to FIX the cost and capabilities of that
\"solution\" without concern for the needs of those \"future
applications\"???

shakes head

I don\'t understand the head shaking. Anyone who designs
circuits regularly will have their favourite solution for a
common task. I\'ve used various debouncing techniques but they
all require passive components, a particular type of switch,
and in some cases another active device to complement the main
active devices.

If there\'s an inexpensive standalone part that needs no
external parts and can replace those techniques in most
applications, what\'s wrong with using that as a stock
solution? A 2N3904/BC547 transistor can be used at the heart of
a wide range of low frequency low power applications. Why hunt
for a suitable type every time?

What do you do when you need a greater Icc? Do you have your
\"favorite FET\"?

I sure do. Besides the late lamented BF862 (of which I have
several reels), there\'s the CPH3910 and my new fave, the Mini
Circuits SAV-551+, which (with a bit of circuit magic) makes the
world\'s finest bootstraps.


SAV-551 is a great switch to ground too. 2 ohms, half a pF. Good for
steering current, discharging ramps, changing gain, or switching
caps in and out. It only needs a half volt or so of gate swing, but
you can drive it from bigger stuff with just a series gate resistor.

Have you tried the new 18-GHz Renesas pHEMT parts as switches? They
wouldn\'t have much chance to oscillate in that service.

I didn\'t know about them. Got a part number that you like?

The data sheets are pretty good, for an RF part. I have supplemental
measurements, like Rds-on.

The e_N*C noise goes to zero at DC faster than the 1/f noise rises, so
even though the pHEMT has a lot more than the JFETs, its lower flatband
noise wins at bandwidths above around 5-10 MHz. Plus it has twice the
transconductance.



\"Diode\"? \"Switch\"?

Do you use them regardless of what the *particular* requirements
and constraints of the design happen to be?

You should be reusing ideas, not implementations. So, for
example, a particular amplifier topology -- not a particular set
of components to implement that particular amplifier instance.

That\'s needlessly dogmatic. Engineering time is very valuable, and
there are lots of noncritical jobs where cut-n-paste makes perfect
sense. For instance, a bunch of my small instruments run off our
standard 24V wall wart and use two buck regulators to make +12-14V
and -16-20V. That\'s generally an LMR23630 running at 2 MHz and an
AOZ1282CI respectively. (The AOZ runs off the 23630\'s output so we
can have more negative headroom.) They\'re reasonably cheap, work
well, and we know how to keep the EMI low enough for use in small
form factor ultrasensitive instruments. So power and EMI are
solved problems for the most part.

Nothing wrong with cutting and pasting chunks of a schematic, but
you\'ve got to be careful about things like mixing 0603s and 0805s
accidentally, or net names, or part library changes. Just check.

My cut-and-paste is done with scissors and rubber cement. The ref
des get redone at BOM compilation time--I number the parts as I enter
them into the BOM, which reduces the number of blunders that have to be
caught later.

We physically resequence the ref desigs in a standard pattern, to make
manufacturing and QC happy, and back-annotate the schematic.

We enter our MAX stock number as a part attribute on each part on the
schematic, so PADS can generate the BOM in our format. I wrote a
program that cross-checks each part type and value against the part in
stock, which catches a lot of errors.

Some of our boards have a thousand parts.

The net names are a genuine issue, true.

When I cut and paste in LT Spice, and things get really crazy, I look
for shorts. It just replicates the net names.

 

[Phil Hobbs]

On 2020-08-13 16:26, John Larkin wrote:

On Thu, 13 Aug 2020 12:43:51 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

On 2020-08-13 12:06, jlarkin@highlandsniptechnology.com wrote:
On Thu, 13 Aug 2020 11:24:05 -0400, Phil Hobbs
pcdhSpamMeSenseless@electrooptical.net> wrote:

<snip>

Engineering time is very valuable, and there are lots of
noncritical jobs where cut-n-paste makes perfect sense. For
instance, a bunch of my small instruments run off our standard
24V wall wart and use two buck regulators to make +12-14V and
-16-20V. That\'s generally an LMR23630 running at 2 MHz and an
AOZ1282CI respectively. (The AOZ runs off the 23630\'s output
so we can have more negative headroom.) They\'re reasonably
cheap, work well, and we know how to keep the EMI low enough
for use in small form factor ultrasensitive instruments. So
power and EMI are solved problems for the most part.

Nothing wrong with cutting and pasting chunks of a schematic,
but you\'ve got to be careful about things like mixing 0603s and
0805s accidentally, or net names, or part library changes. Just
check.

My cut-and-paste is done with scissors and rubber cement. The
ref des get redone at BOM compilation time--I number the parts as I
enter them into the BOM, which reduces the number of blunders that
have to be caught later.

We physically resequence the ref desigs in a standard pattern, to
make manufacturing and QC happy, and back-annotate the schematic.

We enter our MAX stock number as a part attribute on each part on
the schematic, so PADS can generate the BOM in our format. I wrote a
program that cross-checks each part type and value against the part
in stock, which catches a lot of errors.

Simon did one of those for us as well, and our parts DB is in SQLite to
make it easy to back up. I generate a paper schematic and CSV BOM using
our database\'s master list, then somebody lays the board out (Simon for
our internal stuff, vendors for customer stuff).

> Some of our boards have a thousand parts.

We haven\'t gone beyond 250 or so, which is still a lot.

The net names are a genuine issue, true.

When I cut and paste in LT Spice, and things get really crazy, I
look for shorts. It just replicates the net names.

Yup. So does Diptrace.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC / Hobbs ElectroOptics
Optics, Electro-optics, Photonics, Analog Electronics
Briarcliff Manor NY 10510

http://electrooptical.net
http://hobbs-eo.com