Forums

OpAmp Feedback Resistance vs. Stability

Started by Tim Wescott March 28, 2017
So, about a decade and a half ago, I worked with some very ill-tempered 
rail-rail opamps.  They were among the first-generation rail-rail amps, 
and they had the "nice" feature that in spite of their specified 
behavior, they'd go unstable as a voltage follower if you just connected 
the output to the negative input -- but they'd work fine if you did so 
through a 1k-ohm resistor.

I got into the habit of putting a resistor there in my voltage followers.

I just discovered an error in a circuit of mine that uses TSV912 op-amps, 
connected as voltage followers, with 22K-ohm resistors from output to 
input.  I'm having a very "d'oh" moment, because I'm thinking that if you 
DO want resistors there at all, you really want something in the 200-ohm 
to 1k-ohm range.  Habit, I think, kneecapped me.

So, I'm going to experiment around with the best value for this, but does 
anyone have any suggestions as to how to figure out what to use?  Was 
that old ill-tempered op-amp that I used Back Then a fluke, or is it 
generally advantageous to use a resistor in the feedback path of a 
voltage follower?

Suggestions welcome.

-- 

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

I'm looking for work -- see my website!
Den onsdag den 29. marts 2017 kl. 00.12.54 UTC+2 skrev Tim Wescott:
> So, about a decade and a half ago, I worked with some very ill-tempered > rail-rail opamps. They were among the first-generation rail-rail amps, > and they had the "nice" feature that in spite of their specified > behavior, they'd go unstable as a voltage follower if you just connected > the output to the negative input -- but they'd work fine if you did so > through a 1k-ohm resistor. > > I got into the habit of putting a resistor there in my voltage followers. > > I just discovered an error in a circuit of mine that uses TSV912 op-amps, > connected as voltage followers, with 22K-ohm resistors from output to > input. I'm having a very "d'oh" moment, because I'm thinking that if you > DO want resistors there at all, you really want something in the 200-ohm > to 1k-ohm range. Habit, I think, kneecapped me. > > So, I'm going to experiment around with the best value for this, but does > anyone have any suggestions as to how to figure out what to use? Was > that old ill-tempered op-amp that I used Back Then a fluke, or is it > generally advantageous to use a resistor in the feedback path of a > voltage follower? > > Suggestions welcome.
https://e2e.ti.com/blogs_/b/precisionhub/archive/2014/07/15/resistors-in-the-feedback-of-a-buffer-ask-why
On Tue, 28 Mar 2017 15:17:14 -0700, Lasse Langwadt Christensen wrote:

> Den onsdag den 29. marts 2017 kl. 00.12.54 UTC+2 skrev Tim Wescott: >> So, about a decade and a half ago, I worked with some very ill-tempered >> rail-rail opamps. They were among the first-generation rail-rail amps, >> and they had the "nice" feature that in spite of their specified >> behavior, they'd go unstable as a voltage follower if you just >> connected the output to the negative input -- but they'd work fine if >> you did so through a 1k-ohm resistor. >> >> I got into the habit of putting a resistor there in my voltage >> followers. >> >> I just discovered an error in a circuit of mine that uses TSV912 >> op-amps, >> connected as voltage followers, with 22K-ohm resistors from output to >> input. I'm having a very "d'oh" moment, because I'm thinking that if >> you DO want resistors there at all, you really want something in the >> 200-ohm to 1k-ohm range. Habit, I think, kneecapped me. >> >> So, I'm going to experiment around with the best value for this, but >> does anyone have any suggestions as to how to figure out what to use? >> Was that old ill-tempered op-amp that I used Back Then a fluke, or is >> it generally advantageous to use a resistor in the feedback path of a >> voltage follower? >> >> Suggestions welcome. > > https://e2e.ti.com/blogs_/b/precisionhub/archive/2014/07/15/resistors-
in-the-feedback-of-a-buffer-ask-why OK. That explains why it should be zero (easy enough, there's other zero- ohm resistors on the board). I still wonder why that way-back-when amp needed one. I wish I knew the part number -- all I remember is that it was one of the first wave of rail-rail op-amps, and we were all pretty bitter about it by the time the product shipped. -- Tim Wescott Wescott Design Services http://www.wescottdesign.com I'm looking for work -- see my website!
On 03/28/2017 06:12 PM, Tim Wescott wrote:
> So, about a decade and a half ago, I worked with some very ill-tempered > rail-rail opamps. They were among the first-generation rail-rail amps, > and they had the "nice" feature that in spite of their specified > behavior, they'd go unstable as a voltage follower if you just connected > the output to the negative input -- but they'd work fine if you did so > through a 1k-ohm resistor. > > I got into the habit of putting a resistor there in my voltage followers. > > I just discovered an error in a circuit of mine that uses TSV912 op-amps, > connected as voltage followers, with 22K-ohm resistors from output to > input. I'm having a very "d'oh" moment, because I'm thinking that if you > DO want resistors there at all, you really want something in the 200-ohm > to 1k-ohm range. Habit, I think, kneecapped me. > > So, I'm going to experiment around with the best value for this, but does > anyone have any suggestions as to how to figure out what to use? Was > that old ill-tempered op-amp that I used Back Then a fluke, or is it > generally advantageous to use a resistor in the feedback path of a > voltage follower? > > Suggestions welcome. >
There are lots of good op amps out there. Just find one that doesn't need the resistor. ;) Seriously, I've never ever done that in a follower. (Well, once long ago I was in a jam and used a decompensated op amp as a follower by jacking up the noise gain. But not recently.) Seriously, apart from some weirdness with input protection diodes and fast transients, I can't imagine an amp needing such a resistor. It just adds noise and high frequency peaking. Cheers Phil Hobbs -- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC Optics, Electro-optics, Photonics, Analog Electronics 160 North State Road #203 Briarcliff Manor NY 10510 hobbs at electrooptical dot net http://electrooptical.net
On Tue, 28 Mar 2017 17:12:46 -0500, Tim Wescott
<seemywebsite@myfooter.really> wrote:

>So, about a decade and a half ago, I worked with some very ill-tempered >rail-rail opamps. They were among the first-generation rail-rail amps, >and they had the "nice" feature that in spite of their specified >behavior, they'd go unstable as a voltage follower if you just connected >the output to the negative input -- but they'd work fine if you did so >through a 1k-ohm resistor. > >I got into the habit of putting a resistor there in my voltage followers. > >I just discovered an error in a circuit of mine that uses TSV912 op-amps, >connected as voltage followers, with 22K-ohm resistors from output to >input. I'm having a very "d'oh" moment, because I'm thinking that if you >DO want resistors there at all, you really want something in the 200-ohm >to 1k-ohm range. Habit, I think, kneecapped me. > >So, I'm going to experiment around with the best value for this, but does >anyone have any suggestions as to how to figure out what to use? Was >that old ill-tempered op-amp that I used Back Then a fluke, or is it >generally advantageous to use a resistor in the feedback path of a >voltage follower? > >Suggestions welcome.
I don't do that, unless it's a current-feedback opamp. In theory, balancing the input resistances can cancel Ib-driven errors, but I seldom do that either. OPA197 is a nice gumdrop 32-volt RRIO amp. It clips/recovers cleanly and it's stable with a big cap from output to ground. -- John Larkin Highland Technology, Inc picosecond timing precision measurement jlarkin att highlandtechnology dott com http://www.highlandtechnology.com
On 03/28/2017 06:27 PM, Phil Hobbs wrote:
> On 03/28/2017 06:12 PM, Tim Wescott wrote: >> So, about a decade and a half ago, I worked with some very ill-tempered >> rail-rail opamps. They were among the first-generation rail-rail amps, >> and they had the "nice" feature that in spite of their specified >> behavior, they'd go unstable as a voltage follower if you just connected >> the output to the negative input -- but they'd work fine if you did so >> through a 1k-ohm resistor. >> >> I got into the habit of putting a resistor there in my voltage followers. >> >> I just discovered an error in a circuit of mine that uses TSV912 op-amps, >> connected as voltage followers, with 22K-ohm resistors from output to >> input. I'm having a very "d'oh" moment, because I'm thinking that if you >> DO want resistors there at all, you really want something in the 200-ohm >> to 1k-ohm range. Habit, I think, kneecapped me. >> >> So, I'm going to experiment around with the best value for this, but does >> anyone have any suggestions as to how to figure out what to use? Was >> that old ill-tempered op-amp that I used Back Then a fluke, or is it >> generally advantageous to use a resistor in the feedback path of a >> voltage follower? >> >> Suggestions welcome. >> > > There are lots of good op amps out there. Just find one that doesn't > need the resistor. ;) > > Seriously, I've never ever done that in a follower. (Well, once long > ago I was in a jam and used a decompensated op amp as a follower by > jacking up the noise gain. But not recently.) > > Seriously, apart from some weirdness with input protection diodes and > fast transients, I can't imagine an amp needing such a resistor. It > just adds noise and high frequency peaking. > > Cheers > > Phil Hobbs >
A unity-gain-stable op amp that's laid out correctly but can't remain stable in a basic follower configuration in 2k17 is a garbage-tier amp, throw it in the garbage.
On Tuesday, March 28, 2017 at 6:12:54 PM UTC-4, Tim Wescott wrote:
> So, about a decade and a half ago, I worked with some very ill-tempered > rail-rail opamps. They were among the first-generation rail-rail amps, > and they had the "nice" feature that in spite of their specified > behavior, they'd go unstable as a voltage follower if you just connected > the output to the negative input -- but they'd work fine if you did so > through a 1k-ohm resistor. > > I got into the habit of putting a resistor there in my voltage followers. > > I just discovered an error in a circuit of mine that uses TSV912 op-amps, > connected as voltage followers, with 22K-ohm resistors from output to > input. I'm having a very "d'oh" moment, because I'm thinking that if you > DO want resistors there at all, you really want something in the 200-ohm > to 1k-ohm range. Habit, I think, kneecapped me. > > So, I'm going to experiment around with the best value for this, but does > anyone have any suggestions as to how to figure out what to use? Was > that old ill-tempered op-amp that I used Back Then a fluke, or is it > generally advantageous to use a resistor in the feedback path of a > voltage follower? > > Suggestions welcome. > > -- > > Tim Wescott > Wescott Design Services > http://www.wescottdesign.com > > I'm looking for work -- see my website!
Not rail to rail, but I've put 500 ohms in the OPA227's following figure 37 here, http://www.ti.com/lit/ds/symlink/opa227.pdf as current limiting resistors. George H.
On Tue, 28 Mar 2017 15:17:14 -0700, Lasse Langwadt Christensen wrote:

> Den onsdag den 29. marts 2017 kl. 00.12.54 UTC+2 skrev Tim Wescott: >> So, about a decade and a half ago, I worked with some very ill-tempered >> rail-rail opamps. They were among the first-generation rail-rail amps, >> and they had the "nice" feature that in spite of their specified >> behavior, they'd go unstable as a voltage follower if you just >> connected the output to the negative input -- but they'd work fine if >> you did so through a 1k-ohm resistor. >> >> I got into the habit of putting a resistor there in my voltage >> followers. >> >> I just discovered an error in a circuit of mine that uses TSV912 >> op-amps, >> connected as voltage followers, with 22K-ohm resistors from output to >> input. I'm having a very "d'oh" moment, because I'm thinking that if >> you DO want resistors there at all, you really want something in the >> 200-ohm to 1k-ohm range. Habit, I think, kneecapped me. >> >> So, I'm going to experiment around with the best value for this, but >> does anyone have any suggestions as to how to figure out what to use? >> Was that old ill-tempered op-amp that I used Back Then a fluke, or is >> it generally advantageous to use a resistor in the feedback path of a >> voltage follower? >> >> Suggestions welcome. > > https://e2e.ti.com/blogs_/b/precisionhub/archive/2014/07/15/resistors-
in-the-feedback-of-a-buffer-ask-why Thanks. Zeroing out the resistance solves the problem, although it does leave me with "d'oh" stamped on my forehead. Unfortunately, it doesn't solve the problem that my customer is fighting in Baltimore, at the same time that it _does_ solve the symptoms that I was seeing in my test setup here on the west coast. We've been plagued by this all along, alas. -- Tim Wescott Control systems, embedded software and circuit design I'm looking for work! See my website if you're interested http://www.wescottdesign.com
Tim - the early opamps (lm707??) required "compensation" so that the loop 
gain was less than unity at the frequency where the phase shift was 180 
degrees. The first opamp that completely compensated itself, ie could be 
used at unity gain with no weird behavior, was, I think, the single 
version of the lm747. 
   Those versions that required compensation had special pins for
capacitors. But the circuitry between the output and the negative input 
could also be used for compensation, that is to reduce the loop gain with a 
low frequency rc filter. That is probably what you were doing with the 
rail to rail opamp. A 1K resister and 20 pf creates a 3 db point around 8 
mhz or so.

Hul

Tim Wescott <tim@seemywebsite.com> wrote:
> On Tue, 28 Mar 2017 15:17:14 -0700, Lasse Langwadt Christensen wrote:
> > Den onsdag den 29. marts 2017 kl. 00.12.54 UTC+2 skrev Tim Wescott: > >> So, about a decade and a half ago, I worked with some very ill-tempered > >> rail-rail opamps. They were among the first-generation rail-rail amps, > >> and they had the "nice" feature that in spite of their specified > >> behavior, they'd go unstable as a voltage follower if you just > >> connected the output to the negative input -- but they'd work fine if > >> you did so through a 1k-ohm resistor. > >> > >> I got into the habit of putting a resistor there in my voltage > >> followers. > >> > >> I just discovered an error in a circuit of mine that uses TSV912 > >> op-amps, > >> connected as voltage followers, with 22K-ohm resistors from output to > >> input. I'm having a very "d'oh" moment, because I'm thinking that if > >> you DO want resistors there at all, you really want something in the > >> 200-ohm to 1k-ohm range. Habit, I think, kneecapped me. > >> > >> So, I'm going to experiment around with the best value for this, but > >> does anyone have any suggestions as to how to figure out what to use? > >> Was that old ill-tempered op-amp that I used Back Then a fluke, or is > >> it generally advantageous to use a resistor in the feedback path of a > >> voltage follower? > >> > >> Suggestions welcome. > > > > https://e2e.ti.com/blogs_/b/precisionhub/archive/2014/07/15/resistors- > in-the-feedback-of-a-buffer-ask-why
> Thanks. Zeroing out the resistance solves the problem, although it does > leave me with "d'oh" stamped on my forehead.
> Unfortunately, it doesn't solve the problem that my customer is fighting > in Baltimore, at the same time that it _does_ solve the symptoms that I > was seeing in my test setup here on the west coast. We've been plagued > by this all along, alas.
> -- > Tim Wescott > Control systems, embedded software and circuit design > I'm looking for work! See my website if you're interested > http://www.wescottdesign.com
On Wed, 29 Mar 2017 22:54:13 +0000 (UTC), Hul Tytus <ht@panix.com>
wrote:

>Tim - the early opamps (lm707??) required "compensation" so that the loop >gain was less than unity at the frequency where the phase shift was 180 >degrees. The first opamp that completely compensated itself, ie could be >used at unity gain with no weird behavior, was, I think, the single >version of the lm747. > Those versions that required compensation had special pins for >capacitors. But the circuitry between the output and the negative input >could also be used for compensation, that is to reduce the loop gain with a >low frequency rc filter. That is probably what you were doing with the >rail to rail opamp. A 1K resister and 20 pf creates a 3 db point around 8 >mhz or so. > >Hul >
[snip] My first OpAmp, MC1530/31, early '60's... <http://www.analog-innovations.com/SED/MC1530-DataSheet.pdf> was like that, external compensation, sliding-class-A output stage, still being sold by Lansdale (licensed from Motorola), 53-years later ;-) <http://www.analog-innovations.com/SED/MC1530-TeachingExercise.pdf> The military love it... absolutely stable, high-slew-rate for the era, indefinite-time short-circuit proof. ...Jim Thompson -- | James E.Thompson | mens | | Analog Innovations | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | STV, Queen Creek, AZ 85142 Skype: skypeanalog | | | Voice:(480)460-2350 Fax: Available upon request | Brass Rat | | E-mail Icon at http://www.analog-innovations.com | 1962 | Thinking outside the box... producing elegant solutions.