Cute amplifier - bootstrapped

[[email protected]]

Hmmm... it shouldn't be, unless the generator impedance is high enough tolet
some feedback work, and then the gain goes down. Free lunch and all that.


BCX70K has a min beta of 380. Go for it.

The 2n5089 has min beta of 450 @ 1mA. Lovely part. I like the BCX70K
too.

At ic=60uA, beta = 400, we'll need ib=150nA across [500k bias network
z] = 75mV. Okay, I guess that's not a big deal. Wheeee!

 

[josephkk]

On 15/01/2013 16:50, Jim Thompson wrote:
On Sun, 13 Jan 2013 13:48:15 -0700, Jim Thompson
[snip]

More elaborate...

http://www.analog-innovations.com/SED/MC1552LookAlike.pdf

but works nicely. Been selling for about 50 years now >:-} ...Jim
Thompson


On the face it may be more elaborate, but isn't rail to rail.

[snip]

It's more rail-to-rail than the OP circuit.

Like I say, anyone who thinks the OP circuit is marvy, just simulate
it, then come back and tell me about it.

...Jim Thompson

Ya know what, i may simulate it to see if it is as bad as you say. It
certainly seems to have problems from what i have posted already.

?-)

I've sim'd James' version and my PNP spin. With bigger caps, they make nice
audio-range amps with tons of gain. JT doesn't like it because he doesn't
understand it.

Not realistic. JT understands it better than you ever will/could. Your
narcissism is outshining everything else about you again.

?-)

 

[[email protected]]

No, that's not the purpose. The purpose of the bootstrap is to

increase voltage gain. The input impedance doesn't matter--the input

source is a low-impedance highly capacitive transducer.

Well- you don't use that kind of feedback with that kind of source unless you use a series input resistor for voltage to current conversion.

 

[[email protected]]

I wish you would tell us what this is for, at least in terms of gain, bandwidth, and load.

 

[Jamie]

On Thursday, January 17, 2013 4:32:37 PM UTC-5, [email protected] wrote:


I wish you would tell us what this is for, at least in terms of gain, bandwidth, and load.

What it is for is quite obvious and that is to get a ruckus going, and
by all accounts, it's working !

Good job..

Jamie

 

[[email protected]]

[email protected] wrote:






What it is for is quite obvious and that is to get a ruckus going, and

by all accounts, it's working !



Good job..



Jamie

LOL

 

[[email protected]]

Well- you don't use that kind of feedback with that kind of source unlessyou use a series input resistor for voltage to current conversion.

I really don't understand all the controversy--there ISN'T ANY
FEEDBACK. With a low-Z input the amp is open-loop. The 220k resistor
is for d.c. bias only; any affect on gain is acknowledged, but
incidental and undesirable. You could replace it with

---R---+---R----
|
C
|
===

if that makes you feel better, but that's more parts for little gain.

If you want to kill all the gain we just went to great lengths to
obtain, sure, you can make a standard inverting amplifier + summing
junction amplifier, with gain set by R2/R1.

--R1-+----| ...--+--
| |>. |
| |
'-------R2------'

Again, the purpose was not to present a general-purpose amplifier
design (or any particular amplifier application), just explore the
bootstrap. It's cute.

 

[[email protected]]

The feedback is just to set the DC bias point. It's an open-loop amp at
frequencies of interest, and James wants lots of gain. You and Jim refuseto
accept that intent.

Exactly right. Thanks. (I should've read ahead before my last post.)

 

[[email protected]]

Aha!  We've entered the "Larkin excuse phase" >:-}

He'll never accept that he is wrong.  I await his "analysis", but
it'll never happen... he's not competent to do it.


In...

Message-ID:
<d36b68ba-ddac-4b67-b087-59c013f25...@b11g2000yqh.googlegroups.com>
Subject: Re: Cute amplifier - bootstrapped
From: [email protected]

James admitted that C2 is actually a capacitive transducer and that
his original intent was not to make a general purpose amplifier.

Correct. I said early on the real amplifier for the application would
most likely just use a dual op amp, with two inverting amplifiers.

The interesting aspects of this circuit--and the reason for posting--
were the bootstrap, the easy ability to greatly increase gain, convert
a resistor into an simulated current source, increase output swing,
etc. It's cute.

It has nothing to do with gyrators, feedback, etc. Those all totally
miss the boat.

 

[[email protected]]

No they aren't.


Which is why one generally makes the input impedance large enough to
make source impedance insignificant.

A two-transistor amp has all sorts of compromises, naturally. You
can't get max gain if you use feedback resistors to limit the gain and
present a defined input impedance.

But, the goal wasn't to design a general-purpose hi-z input, wideband
low-distortion amplifier with particular performance. I was just
illustrating the bootstrap. That's it. Two-transistors, lots of
gain.

That has been suggested but since he's never said what the amp is for
you don't really know and that's not the only cap. Unless the amp is
intended for above the audio band operation then the 'maximum gain'
you speak of is compromised because the bootstrap cap is a low value.

The application doesn't matter. I only explained the input impedance
being low as an attempt to get Jim to understand that the "feedback"
resistor is for bias only, not a.c. feedback.

I just said that, although one should define the input impedance so
gain doesn't go flying all over the place depending on what you plug
into it.

The original amp takes the converse route of a high feedback resistor
but that compromises the DC bias feedback.

If you 'really' want maximum gain with undefined source impedances
then AC bypass the DC bias feedback.

Sure, but that takes two extra parts. Stripping them out saves two
parts, with little loss of gain.

I haven't heard him say what it is he 'wants to do'. What I heard him
say is it's a cute bootstrapped amp.

I do note he mentioned 40 kHz, but initially missed the feedback
aspect. The boostrap is clearly what he found 'cute', as did I when
first seen.

If I gave the impression of missing the "feedback" aspect, the fault
is my writing. The feedback path is quite completely obvious, but
relatively inconsequential the test frequency.

You're both so interested in one-upmanship that you never hear what
the other is saying,

I think Fred and Jim are treating this as a circuit application design
problem--values and all--as a specific, optimized, general purpose
amplifier design exercise. It's not--it's a bootstrap topology
showcase.

The bootstrap cap is one way to do that. John's constant-current
collector load also bootstraps the collector resistor impedance, a
different way of skinning the cat.

As far as the original problem which inspired this--which is entirely
irrelevant to the thread, and which was never going to use this ckt
anyhow--all I wanted was a spike out when the transducer got pinged.

 

[[email protected]]

Huh?  Write the equation.


No it's not, there's a low pass feedback to the base.

Shame upon you.

Obviously--that's the d.c. bias loop. It's unimportant at 40KHz.
Ignoring the base impedance, 220k / (1/ 2pi*40KHz*10nF) = 553. That's
enough higher than the amplifier's actual voltage gain as to be
unimportant. If it mattered, use a higher series resistor or the "T"
network below.

For C2 being a transducer, it's fine (when "input" is grounded)... as
a voltage amplifier it's mis-analyzed... and!  I must say, I'm
disappointed in you dagmar... James :-(

I guess being a sycophant pays better than being a good engineer?

You just misunderstood the posting. I'm interested in topologies and
techniques (in this case the bootstrap). John understood.

I think you got off on the wrong track when you thought it was some
sort of gyrator.

 

[[email protected]]

As far as the original problem which inspired this--which is entirely

irrelevant to the thread, and which was never going to use this ckt

anyhow--all I wanted was a spike out when the transducer got pinged.

That sounds more like a comparator app than an amplifier app.

 

[[email protected]]

Just write out the whole loop equation, then tell me what's important
and what's not.  And tell me what feedback is not feedback.

I'm really disappointed in you :-(

You should be ashamed, kissing Larkin's ass like that.

You misunderstood the purpose of the bootstrap cap, and now you're
simply being unpleasant.

The bootstrap cap turns Q1's collector resistor into a virtual current
source. That's neat.

 

[[email protected]]

That sounds more like a comparator app than an amplifier app.

Pretty much, yes, but the signals might've been sub-millivolt, so a
little gain first makes the comparator less critical and switch
faster.

The extra super duper stuff was all just for fun.

The bootstrap efficacy is limited by Q2's sub-unity gain.
Substituting a Sziklai pair greatly increases the bootstrapped virtual
current-source's impedance, at which point yes, you need the t-ee bias
network to take advantage of it.

 

[[email protected]]

That's stupid.

Okay, we're done.

 

[[email protected]]

It now occurs to me that my PNP bootstrap is the same as (Tim's?)
"shunt current source" and it's a follower, too, for free.

Yes, your PNP "current source" is the same as Tim's "shunt current
source." It was featured as design idea in one of the rags some years
back--Electronic Design perhaps?

It's pretty cute too, the principal drawback being that the output
can't swing as low.

 

[[email protected]]

Pretty much, yes, but the signals might've been sub-millivolt, so a

little gain first makes the comparator less critical and switch

faster.

You can make a comparator for submillivolt signal levels but you're going to have to do some serious bandlimiting and low noise design, and you'll also be in the realm of thermal emf, air current, and component placement shielding.

 

[[email protected]]

That's a mixed metaphor but with the circuit in question, in fact, yes
you can and I explained how below.

Per your assumptions below, yes.

And I explained it was the 'boostrap' you found 'cute'. So far I don't
know what you're trying to 'argue' about.

I'm just clarifying where there seems to have been uncertainty.
There's nothing to argue about AFAICT.

Of course it does. If one wants an audio amp it stinks but if one
wants to amplify 40 kHz then not so much.

The application doesn't matter for the intended subject, which was
showing that you can bootstrap a collector load resistor.

Feedback doesn't 'come and go' simply because of your 'intent',
feedback is feedback, and, while you might argue it's 'small', unless
the source impedance is 0 (good luck on that) then there's feedback.

Granted. I mentioned it in passing around post 25 or so, but only in
passing since it simply wasn't germane to the thing being
demonstrated: bootstrapping a collector load. I try to strip
extraneous detail out of descriptions, don't you?

The raw circuit didn't give an 'application' and, without that, the
source impedance is unknown and, so, the amount of negative feedback
is unknown.

Sure, but that takes two extra parts.

Potentially one, but who's counting?

 Stripping them out saves two
parts, with little loss of gain.

You have no idea how 'little [the] loss of gain' is because the
premise states "with undefined source impedances."

I missed that qualifier. Yes, with that assumption, you're correct.
That wasn't my assumption; I assumed a capacitive, low-impedance
source.

What test frequency at what source impedance?

f as spec'd in the drawings and all the results--40KHz. Source
impedance WAG'd @ 10nF, e.s.r. unknown (believed < 100r).

Clearly not 'irrelevant'.


You don't think people ever look at a design to see if maybe they
might want to use it, or parts of it?

YES, the BOOTSTRAP!!!! Increases gain and swing, hence the post. The
rest of it is bog-standard practice. (Apparently the bootstrap is
standard too, I just hadn't seen it.)

 

[[email protected]]

You can make a comparator for submillivolt signal levels but you're goingto have to do some serious bandlimiting and low noise design, and you'll also be in the realm of thermal emf, air current, and component placement shielding.

I've never done that--the feedback from the output switching is such a
nuisance (the dang things want to oscillate), it's easier to
preamplify a little.

I discovered the oscillations in detail in the comparator of my first
homemade dual-slope a/d (and that wasn't even anything hot, just an
LM311). 16 bits in an industry when/where 12 bits reigned. (I
thought it'd be hard, but it wasn't.) Linear Tech had app notes
saying you should include a few mV of hysteresis to avoid that...

Still seems like good practice, when practical.

 

[[email protected]]

That 500 KV output; is that peak or RMS?

Obviously means V/V