Can I do this with a uProcessor?

[Rich Grise]

The human ear routinely picks out such signals with surprizing ease and
regularity. A trained ear can dig a signal out from under as much as 40
dB stronger noise. But in context i do not think that direction that
Rich intended to apply the ratio.

OK, Lucy, lemme 'splain. I was basing my figures on something I heard
about 15 or 20 years ago when I was all into new-age stuff, and yuppies
were spending lots of money on aromatherapy and self-help books and that
sort of thing. One of the products that the new-age yuppies were buying
was "subliminal tapes", where you listen to a tape of maybe the waves at a
seashore, or a forest, or something like that, while, dubbed in with of
the sound of the surf, there's a human voice mixed in, but attenuated by
~20 dB, which supposedly sneaks its way into your subconscious and tells
your mind how to make lots of money or whatever. It's kinda like hypnosis
- there's some guy saying stuff like, "you are successful, you are
beautiful, the Universe loves you..." and crap like that.

Well, I looked into that -20dB "sumliminal" level because my clever idea
was to offer people a service where they could read their own affirmations
onto a tape, and I, in my mad metaphysicist's la-bohr-a-tory, would dub
it onto a tape of the ocean, for a fee.

And I just had an adder where the voice tape was attenuated 20 dB or so.

The thing never got off the ground, and the whole new age thing seems to
have fizzled out, but I _did_ notice that -20dB made it "virtually"
inaudible to me, but that's 20 dB down from the sound of the ocean,
which is basically noise in the first place. ;-)

Thanks,
Rich

 

[Rich Grise]

This is so true. The background bzzzzz that scanning can create is unacceptable.

But, did tempus fugit (the OP) ever say what the actual goal was? If he's
controlling cooking cycles in a microwave oven, a little scanning noise is
irrelevant.

And even if it is audio, an "intelligent design" could also make the
scanning noise irrelevant. John Larkin might have some ideas here - I'd
think that if scanning noise is a problem, it'd be much more expensive of
a problem on a jet engine test stand than on a rock band's mixer board. ;-)

Cheers!
Rich

 

[Rich Grise]

The penny hadn't dropped earlier but I see what you're doing now.

I'm just wondering if it can be upset by a simultaneous depression of 2 or more
keys.

it _can_, but it doesn't have to if you put a little thought into your
firmware. Or maybe use isolation diodes.

Sometime last millennium, I had a 64-or so raw keyboard, with SPST
contacts - actually it was kind of a cool design - it had "quad-furcated"
contacts which naturally wanted to contact each other, but each keytop
had a little plastic baffle that kept them apart - when you pressed the
key, it lowered the baffle and let the contacts contact. I wired it up,
and I'm almost sure I didn't use any diodes at the keyboard itself, but
I do know I didn't have any trouble at all writing in n-key rollover
using a 6502 to do the scanning. I think I only looked at one pair at a
time, and whichever was closed, I queued. Of course, whatever was looking
for the keystrokes had to have enough smarts to use them one at a time.

When I built my "TV Typewriter" knock-off ( ;-) ), I put in typomatic.

Ah, the good old days!

Cheers!
Rich

 

[Eeyore]

But, did tempus fugit (the OP) ever say what the actual goal was?

He's been posting about this project of his for maybe 6 months or so.

I've simply been following it, so I'm more familiar with what he's trying to achieve.
I know therefore that he wants good click, hum and buzz free audio.

Graham

 

[Jasen]

Whether or not tempis fugit "scans" in matrix fashion or just checks state
one by one fast enough to be interactive, say "on the measure", all the
switches must be sampled at least 100 times a second. Thus some "digital
noise" is generated.

that's incorrect, there is no requirement to scan continuously,

there is no requirement to scan at all unless he needs to be able to
distinguish two keys down at the same time.

connect rows and columns to bidirectional IO pins, fit (or enable) pullup
on the rows and columns.

configure rows to out and columns to input make the rows low.
if you see a low column then a key is pressed (else not)

unless a key is pressed do nothing.

if a key is pressed swap it round. (make columns low outputs and rows
inputs) read the rows an see which one is low

however doing that will create a pop when the swap is done (which makes
it unsuitable for tempus fugit)

 

[krw]

The human ear routinely picks out such signals with surprizing ease and
regularity. A trained ear can dig a signal out from under as much as 40 dB
stronger noise. But in context i do not think that direction that Rich
intended to apply the ratio.

I didn't say I agreed with rich. I was simply reading what Rich
wrote, to the dumb donkey, again.

 

[Eeyore]

that's incorrect, there is no requirement to scan continuously,

there is no requirement to scan at all unless he needs to be able to
distinguish two keys down at the same time.

If you were really clever, the 'keyscan' could even be interrupt driven.

No need though.

Graham

 

[Eeyore]

I didn't say I agreed with rich. I was simply reading what Rich
wrote, to the dumb donkey, again.

How often do you talk to donkeys ?

Graham

 

[MooseFET]

that's incorrect, there is no requirement to scan continuously,

there is no requirement to scan at all unless he needs to be able to
distinguish two keys down at the same time.

connect rows and columns to bidirectional IO pins, fit (or enable) pullup
on the rows and columns.

configure rows to out and columns to input make the rows low.
if you see a low column then a key is pressed (else not)

unless a key is pressed do nothing.

if a key is pressed swap it round. (make columns low outputs and rows
inputs) read the rows an see which one is low

however doing that will create a pop when the swap is done (which makes
it unsuitable for tempus fugit)

Said "pop" may be very quiet. The human ear can pick up a tone at a
much lower amplitude than a single click. The speed of the edges can
be quite low so the higher frequencies that zoom right through
capacitances needn't be there.

 

[MooseFET]

If you were really clever, the 'keyscan' could even be interrupt driven.

No need though.

Perhaps there is a reason to do this. It would let you park the micro
in a halt or low power state most of the time. This would lower the
current it takes and avoid the problem of noise from its Icc getting
into the audio.

 

[tempus fugit]

But, did tempus fugit (the OP) ever say what the actual goal was?

He's been posting about this project of his for maybe 6 months or so.

I've simply been following it, so I'm more familiar with what he's trying to achieve.
I know therefore that he wants good click, hum and buzz free audio.

Graham

Thanks again for clarifying for me Graham. In retrospect, i really should've
mentioned that it was an audio application in the original post, but (again
showing my ignorance in any number of areas) I had no idea it would be an
issue.

 

[tempus fugit]

Like fet switches ? It's 'charge injection' usually. Jfets don't exhibit this
effect like CMOS swiches, although there is another way they can click but this
can be reduced by driving them carefully. Do you recall these were the type I
originally recommended ?

I've been thinking about this JFET idea again, as well. Is it possible to
configure JFETs to work like a relay though, or would I use them to switch
the signal to ground (if the latter, I'm back to the 'fair amount of
rewiring' stage...)?
Thanks

 

[MooseFET]

I've been thinking about this JFET idea again, as well. Is it possible to
configure JFETs to work like a relay though, or would I use them to switch
the signal to ground (if the latter, I'm back to the 'fair amount of
rewiring' stage...)?

The gate of some JFETs can be swung to large voltages beyond pinch
off. This means that it is not a problem to ensure that the OFF
device remains off regardless of the swing of the signal. It is
preventing large forward biases on the gate that is a bit trickier.

For the ON switch you want to drive the gate to exactly the same
voltage as the output side of the switch. Sometimes simply having it
connected via a resistor to a buffered version of the output is the
best way to go.

 

[Eeyore]

Perhaps there is a reason to do this. It would let you park the micro
in a halt or low power state most of the time. This would lower the
current it takes and avoid the problem of noise from its Icc getting
into the audio.

Yes, that would work.

Graham

 

[Spehro Pefhany]

If you were really clever, the 'keyscan' could even be interrupt driven.

No need though.

Graham

This is commonly done to wake up a microcontroller that's been put in
sleep mode to save battery power. Many PICs, for example, have an
interrupt on change function that will detect changes on any of a
number of pins. This could give you the pin-saving advantages of
scanning, without scanning continously.



Best regards,
Spehro Pefhany

 

[Eeyore]

I've been thinking about this JFET idea again, as well. Is it possible to
configure JFETs to work like a relay though, or would I use them to switch
the signal to ground (if the latter, I'm back to the 'fair amount of
rewiring' stage...)?

Yes you can use jfets in series with the signal (like a relay as you put it) as
well as to switch to ground so there's no problem there.

Just remind me what voltages you have available.

Graham

 

[Eeyore]

This is commonly done to wake up a microcontroller that's been put in
sleep mode to save battery power. Many PICs, for example, have an
interrupt on change function that will detect changes on any of a
number of pins. This could give you the pin-saving advantages of
scanning, without scanning continously.

Cute. I like that.

Graham

 

[krw]

How often do you talk to donkeys ?

Here, quite often. Always dumb ones.

 

[tempus fugit]

Yes you can use jfets in series with the signal (like a relay as you put it) as
well as to switch to ground so there's no problem there.

Just remind me what voltages you have available.

Graham

ATM I have +5, +9 and +12v available.

How would you wire the JFETs so that they could work as bypasses for the
effects without grounding the entire chain?

Thanks

 

[Michael A. Terrell]

Here, quite often. Always dumb ones.

Of course! That is where the term, "Dumb Ass" comes from. Look in
the dictionary under Dumbass and you'll find Eeyore's picture.

--
Service to my country? Been there, Done that, and I've got my DD214 to
prove it.
Member of DAV #85.

Michael A. Terrell
Central Florida