amplify 40kHz audio signal using TL082: first two stages are

[Joerg]

Hello Larry,

I think the cost of those parts will come down as
newer (and faster) fab lines come online and the
ones currently dedicated to higher margin parts
become available for "cheap" stuff.

That's what I always hope for but sometimes it never happens, other
times that price decline is too slow. Take the AD603. It actually seemed
to have gone up a little bit. Well, it doesn't have any competition and
its market is kind of small. So I did a lot of these designs with
transistors and PIN diodes. One upside with that approach is that you
can achieve a dynamic range from here to the Klondike, almost like what
tubes can do.

Designing for volume over years is a special challenge,
one which often precludes using the niftiest new parts.

Yes. Typically 90% or more of the stuff that receives glitz and glamour
in EE Times or EDN is off limits in that design field. Plus nearly
anything that is single source.

I put a discrete amplifier in front of a predecessor
to the AD603. It added cost, but got the noise
down enough that management agreed with its use
in a moderately high volume circuit. Later, I found
out that a system level foulup had discarded about
6 dB of SNR after all my effort to glean a several
dB SNR improvement over what the IC could do.
(I won't go in the bureaucratic snafu behind that!)

Somebody messed up their algorithm? The AD603 is quite remarkable in
noise figure. But I usually have a preamp in front of it with a "hot
rod" RF transistor. Just to squeeze out the last dB.

Regards, Joerg

http://www.analogconsultants.com

 

[Ken Smith]

In article <6wm%d.26263$OU1.18406@newssvr21.news.prodigy.com>,
Joerg <notthisjoergsch@removethispacbell.net> wrote:
[.. snip ..]

It depends on the amount of information you need to process. I was
thinking more along the lines of ultrasound as used in the marine or
medical world. There you often have to log a whole enchilada of echoes
in short sequence. The usual scenario is to use a burst of 2-3 cycles
followed by a long enough listening phase. The receiver bandwidth is
then matched to that burst.

Yes, depending on other factors, you need a wide bandwidth if you need to
resolve multiple targets.

In the marine case, some side scan sonars effectively end up with a narrow
band width. The data from several pulses are averaged together to improve
the SNR.

It also depends on how you define "wide band". The signal is all within
the same decade, in most such systems. There is usually as you said 2-3
cycles at least. In a few systems, the bandwidth really is wide and the
signal can't really be said to have cycles. There is more than one zero
crossing but the time between them increases as you go along.

--
--
kensmith@rahul.net forging knowledge