Jon Slaughter wrote:
"Joerg" wrote in message
...
Jon Slaughter wrote:
"Joerg" wrote in message
. net...
Jon Slaughter wrote:
"Joerg" wrote in message
digy.net...
Jon Slaughter wrote:
Audio mute switch
In the real world this may generate noticeable clicks, mostly coming in
via Cgd and due to the huge input voltage swing. You might want to slow
down the gate drive a bit, for example via an RC lowpass.
Thats not what I asked. I do realize that will happen but I first need
to figure out why the circuit isn't working like I want. Utlimately I'm
going to use an audio analog ic switch that takes care of all those
issues but I first want to understand the basics and I can't seem to
figure out whats wrong.
Well, it was just a comment. There was no question in your post, if there
was then it didn't propagate to my news server. So, what don't you like
about it?
Yes, not that it wasn't a good comment but no one has yet mentioned any
problem with my circuit ;/ This is the real issue I'm after. I want to
know why its not working like it should(or why my reasoning wrong).
In AOE its mentioned as an application for fets in the beginning of the
fet chapter(3 I think). But the issue is that either some audio is passed
when its not suppose to or there is a large amount of signal loss. (about
1/5 of whats input) Maybe this is just the way fets work? Maybe its a bug
in mutlisim? Maybe my circuit is wrong? Hopefully someone will eventually
answer these questions 
A single FET is not going to be ideal. There will always be leakage when
it is turned off, mostly caused by the Cgd capacitance in this case. This
forms an AC-voltage divider with the 1K resistor. It'll be small but you
can run the numbers for each frequency of interest once you know Cgd.
And when it is on there will always be a remaining channel resistance,
commonly called Rdson. FETs aren't ideal switches. If, for example, Rdson
is around 200 ohms then you'll lose 1/5th of the signal.
Well, I'm mainly worried about passing an undistorted signal as possible
with the least amount of attenuation. Unfortunately for my application I'll
be running the signal through many fets(maybe 20+ or even more depending on
how complicated I make it) and so the noise and attenuation might make the
project impractical with fets. (using dips now which I think are ok but I
haven't done any tests)
20? Wow, that would warrant some thinking about the architecture.
Don't count on all audio IC switches to be perfect. I had to re-design
several squelch circuits because of unacceptable popping noise and some
were using "professional" audio chips.
Yeah ;/ I think at this point anything will do. I didn't realize they
were far from ideal. The IC I was looking at only has a -80db rejection
of the signal(if I read the datasheet right). While this is very low it
could cause some problems for my circuit.
Look at a T-switch architecture. But just FYI, layout is not trivial when
you want more than 80dB over the whole audio range. And don't do the whole
T-switch in a single chip because mundane things such as bond wire
impedance can spoil the broth, stuff that doesn't show up at all in Spice
unless you make it part of your model. Also, consider differential signal
paths.
Can you find this architecture in multiple switch IC's? I'm looking for
something thats 6 or larger(the larger the better) as I'll have to use
several.
I am an ultrasound guy when it comes to switches. Crosspoints etc.,
rather expensive or full custom. Take a look around the DG series that
Graham mentioned.
I don't think I can use differential signals as it will make it overly
complicated but I'll try to keep it in mind when working out some of the
details.
What I have now is a bunch of dip switches and I have to manually switch
them into different combinations. My goal is to replace the switches with
something much smaller and can be electrically controlled. Fets seem to
solve the problem on the surface(well, atleast in ic form as doing it
discrete makes it to physically large for my project).
I'm going to go over the datasheet again for the ic I was looking at and
see what I can learn from it.... but I'd still like to know why my
circuit isn't functioning properly. (I did implement another version that
was mentioned for an audio mute but it has even worse issues)
One FET alone might not cut the mustard here ;-)
Well, for the simulation I'm not trying to make it perfect but surely the
issue is pretty significant? I mean, its reducing the signal to 1/5 is
original amplitude... thats pretty significant and would seem to make fets
useless?
No, doesn't make them useless. You just can't operate them in an
environment with impedances as low as 1K if you want less loss. Or use
buffers.
The reason though I'm so concerned with the circuit is because I've seen in
3 places already that say you can use a signal fet as an audio mute or even
volume control. I'm sure that in these circuits there is no attenuation by
anything close to what I'm getting. (I haven't implemented them exactly
though cause I'm sure they would work but wondering why mine is not(Since it
probably means I'm not understanding something)).
I'm going to try a p-jfet though as I think it will make the circuit
easier(no need for -V).
Just watch that reverse junction breakdown. But seriously, I'd scope out
a multiplexer series. DG-series of the CD4051 through 4053, to see if
anything fits the bill here. If your voltage range is lower the latter
also comes in 74HC flavors.
--
Regards, Joerg
http://www.analogconsultants.com