Author Topic: Unity gain amplifier stability  (Read 6151 times)

0 Members and 1 Guest are viewing this topic.

Offline ricko_ukTopic starter

  • Super Contributor
  • ***
  • Posts: 1158
  • Country: gb
Unity gain amplifier stability
« on: February 10, 2020, 09:27:56 pm »
Hi,
how can I:
1) make sure an op-amp is unity gain stable (i.e. to use just as buffer) at a certain frequency and certain amplitude if it does not say "unity gain stable" in the datasheet?
2) how do I determine the minimum gain at a certain frequency and certain amplitude for it to be stable? What infos on the datasheet do I need to check?

Thank you :)
« Last Edit: February 10, 2020, 09:34:51 pm by ricko_uk »
 

Offline moffy

  • Super Contributor
  • ***
  • Posts: 2217
  • Country: au
Re: Unity gain amplifier stability
« Reply #1 on: February 11, 2020, 07:55:54 am »
If you get 180 degree phase shift (open loop) at a gain greater than 1 (open loop) then it is not unity gain stable. The open loop transfer curve showing open loop gain and phase shift vs frequency will show you this.
 

Offline Prehistoricman

  • Regular Contributor
  • *
  • Posts: 216
  • Country: gb
Re: Unity gain amplifier stability
« Reply #2 on: February 11, 2020, 09:38:31 am »
I was always taught in uni that the unity gain operational amplifier configuration is the most unstable (or difficult to stabilise).

1) make sure an op-amp is unity gain stable (i.e. to use just as buffer) at a certain frequency and certain amplitude if it does not say "unity gain stable" in the datasheet?
You may need to add a compensation capacitor (if the part supports that) or a resistor in the feedback path. In fact, a resistor in the feedback path is a good design, and you can see the reasons for that here (look at all the answers, they have good stuff):
https://electronics.stackexchange.com/questions/56727/what-is-the-purpose-of-a-resistor-in-the-feedback-path-of-a-unity-gain-buffer

2) how do I determine the minimum gain at a certain frequency and certain amplitude for it to be stable? What infos on the datasheet do I need to check?
Datasheet won't tell you a lot because the opamp performance depends greatly on the external parts. Simulation is probably the best option.

Offline exe

  • Supporter
  • ****
  • Posts: 2647
  • Country: nl
  • self-educated hobbyist
Re: Unity gain amplifier stability
« Reply #3 on: February 11, 2020, 11:16:59 am »
If someone has a good textbook on how to design compensation networks -- please share. I went through gazillions of manuals, but still don't quite understand the topic. What often missing is a "logical" explanation why it is designed that way. Like, few textbooks explain why openloop gain should have a rolloff of -20db/decade (I figured out myself what this means). A lot of educational material starts with math with explaining much: "just put some zeros with your compensation network and you are good to go".  May be I'm searching wrong things, should I start with a generic textbook on feedback systems?
 
The following users thanked this post: ricko_uk

Offline Wimberleytech

  • Super Contributor
  • ***
  • Posts: 1134
  • Country: us
Re: Unity gain amplifier stability
« Reply #4 on: February 11, 2020, 11:31:17 am »
What opamp are you considering?
« Last Edit: February 11, 2020, 11:41:02 am by Wimberleytech »
 
The following users thanked this post: ricko_uk

Offline exe

  • Supporter
  • ****
  • Posts: 2647
  • Country: nl
  • self-educated hobbyist
Re: Unity gain amplifier stability
« Reply #5 on: February 11, 2020, 11:43:30 am »
if it does not say "unity gain stable" in the datasheet?

Pick a different opamp :)
 
The following users thanked this post: thm_w, Wimberleytech, ricko_uk

Offline Wimberleytech

  • Super Contributor
  • ***
  • Posts: 1134
  • Country: us
Re: Unity gain amplifier stability
« Reply #6 on: February 11, 2020, 11:59:20 am »
If someone has a good textbook on how to design compensation networks -- please share. I went through gazillions of manuals, but still don't quite understand the topic. What often missing is a "logical" explanation why it is designed that way. Like, few textbooks explain why openloop gain should have a rolloff of -20db/decade (I figured out myself what this means). A lot of educational material starts with math with explaining much: "just put some zeros with your compensation network and you are good to go".  May be I'm searching wrong things, should I start with a generic textbook on feedback systems?

I have a great set of lecture notes from, uh, 40+ years ago.  I will scan them and post later.
In the mean time...
You cannot avoid math, poles, and zeroes and such.
At the most fundamental level, for an amplifier to be stable, you cannot feed back a signal that is in phase with the input signal because it will be regenerative.  Once you say the word "phase" you are thrust into the world of poles and zeroes and the associated math.

When designing opamps the game is to push all "phase shifting" nodes to higher frequencies (higher than your desired unity-gain BW).  This involves moving zeroes near non-dominant poles and making sure that the attendant pole is yet farther out in frequency.  Or just brute forcing and pushing nasty poles to higher frequencies by lowering resistance at critical nodes (which means increasing gm with more current).

---Art of Electronics Ch 4 has a pretty good treatment of compensation--
« Last Edit: February 11, 2020, 12:24:31 pm by Wimberleytech »
 
The following users thanked this post: exe, ricko_uk, MathWizard

Offline EEEnthusiast

  • Frequent Contributor
  • **
  • Posts: 381
  • Country: in
  • RF boards, Precision Analog, Carpentry
    • https://www.zscircuits.in/
Re: Unity gain amplifier stability
« Reply #7 on: February 11, 2020, 12:22:55 pm »
A unity gain amp is the most susceptible to oscillations as the entire output is fed back to the input. If the feedback path is just a short, then its frequency response is flat. All it takes is the forward gain to drop to unity with 180deg phase for the OPAMP to become unstable. If it is not specified, then use a RC network in the feedback path. You can play with the poles to make it stable. At low frequencies, the gain would still be unity.
Better to get the model and simulate for a phase margin of at least 45deg.
Making products for IOT
https://www.zscircuits.in/
 

Offline exe

  • Supporter
  • ****
  • Posts: 2647
  • Country: nl
  • self-educated hobbyist
Re: Unity gain amplifier stability
« Reply #8 on: February 11, 2020, 12:33:55 pm »
---Art of Electronics Ch 4 has a pretty good treatment of compensation--

Thank you very much for your suggestions, it's time to revise it. I hope I accumulated a "critical mass" of knowledge to get through it this time. Question: do I need to know math to understand the topic, or it's only needed to calculate component values?
 
The following users thanked this post: ricko_uk

Offline Wimberleytech

  • Super Contributor
  • ***
  • Posts: 1134
  • Country: us
Re: Unity gain amplifier stability
« Reply #9 on: February 11, 2020, 12:44:06 pm »
---Art of Electronics Ch 4 has a pretty good treatment of compensation--

Thank you very much for your suggestions, it's time to revise it. I hope I accumulated a "critical mass" of knowledge to get through it this time. Question: do I need to know math to understand the topic, or it's only needed to calculate component values?

Art of Electronics explains it with a minimal amount of math. 
 
The following users thanked this post: ricko_uk

Offline magic

  • Super Contributor
  • ***
  • Posts: 7453
  • Country: pl
Re: Unity gain amplifier stability
« Reply #10 on: February 11, 2020, 01:31:13 pm »
Like, few textbooks explain why openloop gain should have a rolloff of -20db/decade (I figured out myself what this means).
I'm not familiar with the details but rolloff steepness appears to be tied with phase.
There are opamps where it's faster than 20dB/decade at certain frequencies, see NE5534 or LT1128. Phase margin suffers.

if it does not say "unity gain stable" in the datasheet?

Pick a different opamp :)
These days almost everything is unity gain stable. The datasheet may not say it explicitly.
« Last Edit: February 11, 2020, 01:36:30 pm by magic »
 
The following users thanked this post: ricko_uk

Offline Wimberleytech

  • Super Contributor
  • ***
  • Posts: 1134
  • Country: us
Re: Unity gain amplifier stability
« Reply #11 on: February 11, 2020, 01:38:22 pm »
 
The following users thanked this post: HackedFridgeMagnet, AlfBaz, exe, SilverSolder, RoGeorge, intabits, ricko_uk

Offline Wimberleytech

  • Super Contributor
  • ***
  • Posts: 1134
  • Country: us
Re: Unity gain amplifier stability
« Reply #12 on: February 11, 2020, 01:43:26 pm »

Quote
I'm not familiar with the details but rolloff steepness appears to be tied with phase.
There are opamps where it's faster than 20dB/decade at certain frequencies, see NE5534 or LT1128. Phase margin suffers.


For a second order system, if Aß crosses 0dB at 20dB/decade then you will have a minimum of 45° of phase margin.  No such amplifier exists, but that is where the rolloff rule of thumb comes from.
 
The following users thanked this post: ricko_uk

Offline 741

  • Frequent Contributor
  • **
  • Posts: 406
  • Country: gb
    • Circuit & PCB Design (small PCB quantities OK)
Re: Unity gain amplifier stability
« Reply #13 on: February 11, 2020, 05:55:12 pm »
A very simple approach that can work "resonably" is simply to "pot down" (scale down) the inverting input, then allow for this by introducing a gain of the same. When computing the feedback resistance, note the input impedance will be equivalent to the 2 input resistors in parallel.

The above works by simply reducing the "closed-loop gain", and it compromises op-amp circuit performance accordingly.

Offline rstofer

  • Super Contributor
  • ***
  • Posts: 9964
  • Country: us
Re: Unity gain amplifier stability
« Reply #14 on: February 11, 2020, 10:37:09 pm »

I have a great set of lecture notes from, uh, 40+ years ago.  I will scan them and post later.

Here they are: http://www.wimberleytech.com/techblog/wp-content/uploads/2020/02/Second-order-systems-and-compensation_opt.pdf

Thanks for the notes.  They clearly point out why I walked away from analog and focused on digital in college.
 
The following users thanked this post: ricko_uk

Offline Wimberleytech

  • Super Contributor
  • ***
  • Posts: 1134
  • Country: us
Re: Unity gain amplifier stability
« Reply #15 on: February 12, 2020, 01:00:49 am »

I have a great set of lecture notes from, uh, 40+ years ago.  I will scan them and post later.

Here they are: http://www.wimberleytech.com/techblog/wp-content/uploads/2020/02/Second-order-systems-and-compensation_opt.pdf

Thanks for the notes.  They clearly point out why I walked away from analog and focused on digital in college.

LOL...and I have always been amazed by the digital guys...its a dance: analog--digital "mixed signal!"
 
The following users thanked this post: ricko_uk

Offline exe

  • Supporter
  • ****
  • Posts: 2647
  • Country: nl
  • self-educated hobbyist
Re: Unity gain amplifier stability
« Reply #16 on: February 12, 2020, 08:31:49 am »
A few mandatory pics.

PS I'm a "digital" arduino guy because I'm a software engineer at my real job. Want to learn some analog, but feedback loops drive me crazy.
 
The following users thanked this post: ricko_uk

Offline ricko_ukTopic starter

  • Super Contributor
  • ***
  • Posts: 1158
  • Country: gb
Re: Unity gain amplifier stability
« Reply #17 on: February 12, 2020, 12:39:12 pm »
Thank you all for the infos! :)
 

Offline Vovk_Z

  • Super Contributor
  • ***
  • Posts: 1478
  • Country: ua
Re: Unity gain amplifier stability
« Reply #18 on: February 12, 2020, 02:35:25 pm »
1) make sure an op-amp is unity gain stable (i.e. to use just as buffer) at a certain frequency and certain amplitude if it does not say "unity gain stable" in the datasheet?
Usually if it isn't unity gain stable - it will be in the datasheet. So if there isn't such info - that it have to mean that it is "usual" = "unity gain stable" opamp.
But situation changes if there is something else in opamp feedback loop - any device (even long wires) change (worsen) it's stability.
« Last Edit: February 12, 2020, 03:40:16 pm by Vovk_Z »
 
The following users thanked this post: ricko_uk

Online SiliconWizard

  • Super Contributor
  • ***
  • Posts: 15799
  • Country: fr
Re: Unity gain amplifier stability
« Reply #19 on: February 12, 2020, 02:41:53 pm »
(...)
But situation changes if there is something else in opamp feedback loop - any device (even long wires) change (worsen) it's stability.

Of course. Then again, if there's anything in the feedback loop,  we may argue that you won't get unity gain, strictly speaking.

 
The following users thanked this post: ricko_uk

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 17428
  • Country: us
  • DavidH
Re: Unity gain amplifier stability
« Reply #20 on: February 13, 2020, 01:28:06 am »
You have to raise the noise gain.  An easy way to do this is to place a resistor between the inverting and non-inverting inputs.  If you consider the amplifier as non-inverting with the minimum stable gain, then connect the feedback divider to the non-inverting input instead of ground.  Usually this is done through a coupling capacitor to remove errors at DC and low frequencies.
 
The following users thanked this post: ricko_uk

Offline sam[PS]

  • Contributor
  • Posts: 46
  • Country: fr
Re: Unity gain amplifier stability
« Reply #21 on: February 28, 2020, 04:14:49 am »
Well lot of very in depth answer there, probably too much for an arduino guy. By the way congratulation for daring to venture in analog territory, it's like coming from city to jungle it can be terrifying but so much great things to see here, so much unique species.

I would emphasis on "if the datasheet tells you it's not, use another OPAMP"

But moreover, as a simple basic rule avoid driving a capacitive load with an opamp. And the bad news here is your arduino/mcu input port are basically capacitive load, but it should be low enough capacity in most case.
99% of the time you will be using almost the same circuit arrangement so find an opamp that works for those and just stick to it. The famous maxim "if ain't broken ain't fix it" does apply in analog too. Just as an exemple, when mixing opamp based analog and MCU i've been using LM324 countless time with no stability issu so far. Don't forget good PCB layout practice : short trace on critical signal.

Of course there is the 1% remaining cases and i'm sure some will be prompt to try to prove what i just say is totally wrong based on what might happens in those particular case. Let's put it this way, this 1% is what separate a great EE engineer (aka analog guru) from an average one and it takes more than a textbook to get there. But you can make a decent career just applying simple recipes and just accept that once in a while it will not work.

If you want some kind of textbook on the matter, most of my understanding of opamp stability comes from an excellent white paper on the subject by TI (or it might be AD, just check). The TI "analog engineer cookbook" and/or the older "handbook of operational amplifier" are must read, there is so much in those including practical pcb advice that i never found in any other opamp textbook. In general both TI and AD application engineers are true analog gurus and i strongly encourage anyone who want to get deeper into the subject to check for their literature.

Just my 2cts...
 
The following users thanked this post: EEEnthusiast, ricko_uk


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf