Cute amplifier - bootstrapped

Adapted from something I saw on the web...

I had to scratch my head until I understood C1: it bootstraps Q1's
collector load (R2) impedance, turning it effectively into a current
source. Gain is accordingly greatly increased. G = ~220 @ 40KHz.

+3.3V
---
|
.-------+-------.
| |
5K R1 C1 |/ Q2
| 100n .------| 2n5089
+---||-|--. |>.
| | | |
15k R2 | '-----+------>
| | |
C2 +------' R3 820R
10n |/ Q1 |

| |>. | |
| | | R4 470R
| === | |
| | ===
'-----R5------'
220k

R3-R4 set the d.c. output level.

Gain is component-sensitive, but the bootstrap is a cute technique.
That with feedback could stabilize the gain.

 

That's another advantage.

 

You need to turn that around, it is the emitter follower that is being bootstrapped and not the CE. So the CE + CC composite looks like the classic very high gain single pole amplifier inside the feedback loop formed by the resistor shunt feedback divider, everything is simplified. See the textbook write-up here, section 1.17.2:
http://books.google.com/books?id=1-...epage&q=bootstrapped emitter follower&f=false

 

How did that eventually work out?

 

Huh? You bootstrap the follower to increase its input impedance, not lower its output impedance.

 

I remember that trick of yours. Fun.

One of the neatest I've seen was a VLF JFET amplifier. It used
transformers to bootstrap its input capacitance to nada.

James

 

I made a cascode version, but that was getting silly. A dual op-amp
makes more sense at that point with fewer parts.

Talking about silly, a little positive feedback really makes it
sing...

+3.3V
---
|
.-------+-------.
| |
5K R1 C1 |/ Q2
| 22n .------| 2n5089
+---||-|--. |>.
| | | |
15k R2 | '-----+------>
| | |
C2 +------' R3 820R
10n |/ Q1 |

| |>. | |
| | | R4 470R
| | | |
| +---------------'
| | |
| R6 3r3 |
| | |
| === |
| |
'-----R5------'
220k

G= 1050 @ 40KHz.

There actually is some negative feedback via R5--that stabilizes the
gain a little bit.

 

That's plenty cute--a three-transistor inverting amp, with feedback,
then an emitter follower, plus the bias stuff. But is there a
bootstrapped collector load in the signal chain? I didn't see it.

 

Okay, those are fine points, but in this case the designer is trying to eliminate the shunt effect of the emitter follower input bias resistor network, and the bootstrapping succeeds well there.

 

I believe that is how that circuit is put together. Whatever the application was, they knew they needed controlled gain with buffering for the loading.. So someone figured out they can use the super high midband impedance of the bootstrapped CC as a super high gain setting loading for the CE, all of which makes for an easily compensated relatively accurate feedback amplifier.

 

Yep, that's a classic way to raise a.c. input impedance.

ISTM there's really no distinction compared to describe it as
bootstrapping the collector load. The collector load R2, being in
parallel with the Q2 input impedance, bootstrapping R2 raises QA2's
input impedance, that's true. But, the key effect in this circuit is
raising the first stage gain, so it seems most appropriate to look at
it from that perspective.

Either / or.

 

Okay, look at it under a microscope if you prefer it that way.

 

Yep. Jim William's description of the Philbrick varactor amp inspired
R6 above. IRL Jim's throwing a handful of transistors at it makes
more sense for an IC designer--less component-sensitive--but it's
always fun to play.

As mentioned before, I'll likely use a dual op-amp instead, cascaded
inverting stages, a.c.-coupled, d.c. offset. That's one op-amp, two
caps, six resistors--boring, reliable, and repeatable.

 

Q2's input resistance is gain(Q2) x (R3+R4) anyhow, which is on the
order of half a megohm.

So, Q2's un-bootstrapped input impedance doesn't limit Q1's gain, R2
does. Boosting Q2's input impedance per se wouldn't help.

 

I've seen that ages ago, a faded memory, I think likely from that same
G.E. databook. Yours is much linear-er than the circuit I posted.

Hmm. Could even bootstrap the first collector load. Only drawback is
possibly a limited output swing compared to the emitter-follower Q2
version...

I might goof around thinking about that.


James

 

That's not even remotely close. That is the two transistor job with superstablized Q-point self-bias and series feedback.

 

Yep- really- and that circuit has come up as a topic before. You couldn't get to square one figuring out the DC bias then IIRC.

 

You're very welcome.

 

Fred's talking about "another bizarre audio circuit", 3/2/2011

https://groups.google.com/group/sci...read/973bb460733ddab7/67d85dc141f9c4ef?hl=en&



James

 

If I'm remembering right, I first saw that in Microwave Journal, late
'80's. Neat stuff. I never used it, but I found the VLF thing later
(Burhan, Radio Electronics (of all places)) and was reminded of it.