9
\$\begingroup\$

In the OPA454 datasheet I found an interesting circuit idea that I'm not sure what is it usually called (if there is a common name for it). enter image description here

It involves using two opamps to shift the rails of a 3rd one. Note that unlike a bridged [tied load] configuration this circuit requires a doubling of the voltage supply rails. On the other hand, the load is not floating in this circuit, so you can combine ts idea with the bridge to get 4X the output voltage swing (relative to using a single opamp). I'm omitting the 6-opamp bridged picture here as rather obvious; you can find one in the aforementioned datasheet.

My question is just what's a/the common name (if any) for this circuit idea. If I were to coin a name, "dynamic rails", "dynamic operating point" or something like that seems reasonable to me. (But these names don't get any sensible results back via google search.)

EDIT: I also saw something similar, but cheaper, with two BJTs instead of opamps for the "rail shifters" (A1 and A2) in a 1999 EDN article titled "Bootstrapping your op amp yields wide voltage swings" written by [then] AD employees Grayson King and Tim Watkins. Using BJTs would introduce some more non-linearities, no doubt. So maybe "bootsrapping" might be the name for this technique... although Rod Eliott's page discussing the issues with this approach never calls it that, so I'm not convinced "bootstrapping" is the name for it... (EDIT3: Well, this was an incorrect reading of the purpose of that circuit; see comment below the question.)

EDIT2: In another article and in AD app note AN-232 (cited in that article), "supply bootstrapping" or "substrate bootstrapping" refers to something similar (altering the rail voltage via "feedback"), but in these articles it is done for a different purpose: a reduction of the input capacitance non-linearity for opamps with FET input stage... So, I'm guessing "bootstrapping" encompasses the idea I've asked about, but can generally mean the use of this supply-voltage-shifting technique for other purposes.

\$\endgroup\$
  • 1
    \$\begingroup\$ Nice(+1), The link to Rod Eliotts page is not PS boot strapping, but is a power booster... more current. \$\endgroup\$ – George Herold Dec 28 '14 at 15:06
  • \$\begingroup\$ You're right. I didn't look at that schematic carefully enough. There's no signal from the transistors output to the opamp power rails in that circuit. The Zener diodes just feed a (hopefully) constant voltage to the opamp, less than that's used to power the discrete/power stage. \$\endgroup\$ – SX welcomes ageist gossip Dec 28 '14 at 15:32
6
\$\begingroup\$

I've seen "bootstrapping" used in a few more books/articles for this technique:

  • Small Signal Audio Design, 2nd ed., by Douglas Self pp. 136-137; shows both an opamp-based boostrap and a BJT-based one. EDIT to add: As it turns out, there some free excerpts from the first edition of the book published in EE Times; the relevant circuits [aimed at JFET input C/V tweaking] are published in an article called "Op amps in small-signal audio design - Part 2: Distortion in bipolar and JFET input op-amps. Rail bootstrapping to reduce CM distortion".

  • Analog-to-Digital Conversion, 2nd ed., by Marcel J.M. Pelgrom, pp. 210-211 uses "boostrapping" for an NMOS transistor operating above the rails. There are more CMOS books that use this term for the same purpose, e.g. Uyemura's CMOS Logic Circuit Design p. 319.

  • "Advanced techniques tackle advanced op amps' extremely low distortion" by J. Graeme, also reproduced in The EDN Designer's Companion, Here p. 213 the bootstrap is done for measurement purposes.

So I guess that clinches "(rail/[power-]supply) bootstrapping" as a relatively common name for this, even though not every source utters it when discussing such circuits... And the purpose of the rail bootstrap may not always be a voltage swing increase.

If someone finds another (relatively common) term, please contribute another answer.

EDIT: As interesting titbids (found thanks to a rather vague mention on a forum), I found two UK patents on this idea:

  • "Zero input capacitance amplifier" (1988) by Bernard / ABI Systems ; uses the 3-opamp version. Title is self-explanatory.
  • "Interface circuits" (1980) by Mohapatra / Sandman. Rather uninformative title, if you ask me; shows several variants, some with two BJTs and some with opamps all-around. The stated goal is to increase input impedance and CMRR.

Neither of these UK patents uses the term "bootstrap" though... as far as I can tell, anyway; old UK patents are not OCR'd, so search doesn't work in those PDFs.

There's also an output voltage boost circuit based on this idea in Linear appnote 67 in the article titled Extending Op Amp Supplies to Get More Output Voltage by Dale Eagar (starts on p. 58 in this long document). This one uses MOSFETs on the opamp rails. But it never uses the term "bootstrap". So yeah, "bootstrap" is a common enough term for this, but not universally used... The preferred term at Linear for this circuit idea is extended supply mode, which (if you google it) is found exclusively in their documents... so this alternative term appears to have less traction across the industry as whole.

| improve this answer | |
\$\endgroup\$
  • \$\begingroup\$ N.B.: I have marked my own answer as the right one because the system keeps bugging me about picking one and because the other/competing answer is simply a critique of the somewhat established terminology I found. But I'm willing to reconsider if new answers are contributed. \$\endgroup\$ – SX welcomes ageist gossip Jan 4 '15 at 11:42
2
\$\begingroup\$

IMO "bootstrapping" is not an appropriate term for this circuit configuration. "Bootstrapping" means "following"... or to insert a following (opposing) voltage source in series to the input voltage source and a resistor (this old circuit technique is based on one of the Miller theorem applications).

Here (roughly speaking), the op-amp supply voltage varies nonlinearly from 50 V to almost 100 V when the op-amp output voltage changes from 0 V to 100 V (i.e., it is not following). I think the only purpose of this dynamic power supply (50 - 100 V) is to decrease the power dissipated in the op-amp output transistors versus the conventional constant power supply (100 V here).

This idea resembles the variable power supplies with a switchable secondary winding.

| improve this answer | |
\$\endgroup\$
  • 2
    \$\begingroup\$ "Bootstrapping" is unfortunately polysemic, even in just EE circles. In the sense used here it means (quoting from Uyemura): "puling up the value of a physical parameter", which is exactly what happens in the NMOS transistor case. I guess the same term got extended to both positive and negative "pull up" (for opamps), which is obviously a bit upsetting as far as the English language goes, but, hey, it's not me who came up with "bootstrapping" for this. \$\endgroup\$ – SX welcomes ageist gossip Dec 28 '14 at 8:39
  • \$\begingroup\$ The excessive power is still dissipated... but in the other op-amps (the voltage followers). It is a kind of a power (voltage) redistribution between the op-amps. \$\endgroup\$ – Circuit fantasist Dec 28 '14 at 9:08
  • 1
    \$\begingroup\$ The legend say that, several centuries ago, Baron Munchausen applied this technique for the first time... pulling up his shoelaces... and so lifting himself above the ground:) With the same success, he could pull down himself. So, if we present the voltage by a height (potential energy), we can imagine that the positive source "pulls up" the load... and the negative source - "pulls it down". But note that although the second source "pulls" the resistor in the same direction, it does this from the other side of the resistor... which means that it opposes the input source to change the current. \$\endgroup\$ – Circuit fantasist Dec 28 '14 at 12:19

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.