#172
Вставка
- Опубліковано 5 вер 2024
- Op amp gain-BW product and slew rate limiting are defined, discussed and demonstrated on the bench. This discussion applies to the majority of general purpose op amps on the market - as most op amps are internally compensated with a single dominant pole. High speed op amps, unconditionally stable op amps, non-unity gain stable op amps, high power opamps, etc. may not follow these characteristics because they are often compensated differently in their design. An LM358N is used for the example circuit. Other popular op amps like the LM741, etc. will behave in a similar way. Sometimes the slew rate limit of a device will be the dominant factor in determining the bandwidth, and other times the gain-bandwidth product will determine the resulting frequency response. The video demonstrates why this happens. Notes from the video are here:
www.qsl.net/w/w...
Really great explanation! I hadn't looked at slew-rate limited signals on the scope like that -- very good way of showing the problem.
Applied Science Thanks Ben - much appreciated!
I'm finding your channel the best of all of the EE blogs. Dave and signal path are ok but the waffle on to much about smeg, you are straight to the point in an entertaining way
Good way in skewing the problem 👍 nice video and good information!
The way you explained slew rate was the best I've ever heard or read anywhere. Great job!
the demonstration of the slew rate limiting the frequency response of the opamp was so sweet!!!! I have not done the connection between slew rate and freq response.
That all makes a lot of sense. You can tell a great teacher, it seems obvious as they explain things.
Very good demonstration. I have read slew rate in datasheet often but understood it really with this demonstration and explanation. I like the attenuation box you use.
Excellent! You laid out in 15 minuets what it took for me; three months of college study to understand back in 1986, with some help from Walter Jung, of course. Alan, you did an outstanding job! Thanks for your hard work.You make it look so easy however, I'm well aware that it takes a lot of hard work, not only for the concept and math, circuit design and build but the mere production of these videos. Thank you so very much for your hard work to make electronics so easy for saps like me...!
A very, very well done, easy to understand explanation. Link to video instantly forwarded to my hobby network.
Please keep doing. You have found the optimal way to merge theory and practice. I have studied these "boring " things over 20 years ago. Now they are exiting thanks your videos.
Brilliant explanation as usual. Thank you.
A tear fell when you pulled out the datasheet "book". Excellent.
Even though you teach "negative feedback", I am sure no one can ever give you a "negative feedback". Thanks a lot sir. Clear. Compact. Beautiful.
Thank you. I do get negative feedback sometimes. There always seem to be a few thumbs down on my videos...
w2aew they hate you cause they ain't you 😂😂 We love you sir 😊
Excellent!!! It is interesting how the slew rate of the op amp is not affected by the frequency of the signal but basically by the amplitude of the signal.
Excellent descriptions as always. Thanks for taking the time to put this together.
That is my dream oscilloscope. I regret not buying it when i bought an expensive one. I have the tds2002c which i use more haha.
That scope is how old now and still holds its value and has a floppy drive. Tektronix are the best. Im restoring one for a friend. Hes had it in storage for a few decades so im not even sure what the model is. If i knew i could get that scope id sell almost all my gear. I tend to donate my gear when i get new gear. The tek diff probes are so expensive. Id trade my rigol mso5354 and every other piece of gear for that scope and a nice selection of probes. There's a reason you work for them but havent replaced the 3104.
Great video, thanks. Ive found that i had to buy caps that are above my budget to get most opamp circuits to look as perfect as yours.
Sorry for rambling i have severe ptsd. I shall get my ham license in the new year.
Wow that is such a great class! Thank you for sharing such knowledge!
Brilliant video. Suddenly I understand why my home brew function generator acts wired at higher frequencies.
Thanks a lot.
Best explanation of slew rate and GBP
Really great explaination and to the point ,you are inspiration to many
Fantastic teaching.
Great description of GBP. I am jealous of that old databook as well it looks like a great book. Thanks for the vid
I used to have an entire library of the old databooks and applications books from the major semiconductor manufacturers. They were (and still are) a tremendous resource of information. Over the years, my collection is whittled down to just a dozen or two...
@@w2aewyou lent them to other people and never saw them back? 😅
@@keylanoslokj1806 Many of them stayed at the companies that I worked for.
Awesome tutorial. Thank you!
Thank you so much for working on publishing this video. You have marvelously explained the difference between operational amplifiers gain BW limit and slew rate. Hats off.
Awesome practical demo! I really appreciate the balance between theory and real-life measurements.
An Exceptional Narration!
It seems to me, that in most applications we wouldn't want to be slew rate limited. That is, we would want to hit the 3dB corner (f_3dB), or the full power BW, before we hit the slew rate frequency (f_SR). We can check if this is the case from the op-amp's data sheet, which gives us its GBP (gain-bw-product) and its SR (slew-rate).
We have: GBP = (A_CL)(f_3dB) or f_3dB = GBP/A_CL. We also have SR = 2(pi)(Vout_pk)(f_SR) , or f_SR = SR/((2pi)(Vo_pk)). Using Vo_pk = (Vin_pk)(A_CL) this last equation can be written as f_SR = SR/((2pi)(Vin_pk)(A_CL)). So we need f_SR > f_3dB, or SR/((2pi)(Vin_pk)(A_CL)) > GBP/A_CL, or SR/((2pi)(Vin_pk)) > GBP.
If this isn't true then our application needs an op-amp with a larger SR and or a smaller GBP (or we need a redesign where we start with a smaller input signal Vin_pk). Am I thinking about this correctly? Thanks!! And beautiful job on teaching about closed loop f_3dB and SR !!.
awesome job explaining. Super helpful, thanks
Just watched like 10 of your videos, great stuff!
Fantastic video and all in 15:30 minuts !
Exceptional explanation. Precise and compact, thank you very much!!
Thanks! Studying op-amps right now for class and this was confusing me. You made it very clear!
Cool! Be sure to tell your fellow students about my channel, and even your professors! Where are you attending school?
This is absolutely brilliant. Very well explained. Now I can sleep on class and learn at home!
Excellent information and demonstration thank you! I've been having refresh/learn things about op-amps lately for our senior project and this video helped very much!
Many many thanks to you, I'm designing a 2-switch forward converter, and need to do with op-amp feedback compensation. This video make me recall things on my electronics class. :) :)
отличное видео. Спасибо. Как мне не хватало таких видео лет 30 назад :)
That's easy for you to say.
Thanks Alan. It's always a pleasure learning from your videos.
Thank you,I never understood slew rate until now.
Super clear . Thank you for doing this video. Liked and subscribed
your videos are helpful as they are more practical and hands on..thanks for it
Really helpful explanation. Can you make a video on opamp stability and capacitive loading ? It really helps the way you explain it with actual demo. thanks again.
This is great stuff, I just came across phase reversal on a LF353 working as a voltage follower when reaching Gnd. Luckily I also bought TLC272 that did that job perfectly single supply. I know it was a rookie mistake but what better way to learn.
Yes, some op amps have that unexpected phase reversal when you violate the input voltage range. Very nasty, and lock up a system. And, often not specified in a data sheet!
Very well composed video. You are the best in explaining the basic things.
Very helpful video, everything is described in a very calm and demystifying manner.
Btw: whats the purpose of the 2.2k resistor at the Output?
John Matrix The output stage of the LM358 uses a push-pull output stage that has some crossover distortion. Using a 2.2k resistor to ground keeps the stage in the "push" mode (sourcing current) so that I don't incur the crossover issue. Also, the spec sheet for the part specifies the output characteristics with a 2k load, so I figured I'd come close to matching that.
***** thanks for the quick response, now i get it. since i´m a electronics newbie i`m not yet into checking datasheets before asking questions :-)
Interesting info. Will be fun to repeat this experiment
Thank you so much for the explanation of the slew rate, helped me a lot!
Nice explanation about slew rate!!! Very helpful!
Stellar video. Thank you so much for the great lesson & demo.
Very good explanation....Learned the concept practically..just amazing
Your explanations are the best!
A fascinating demonstration.
Regarding the phase delay introduced when the slew rate limit is reached: is this a constant? You pointed out that the falling edge has a slightly different slope than the leading edge and this was clearly seen. Two thoughts on this. (1) is this predictable enough that it has utility in a design? and (2) as a phase delay appears to be present (although you didn't show the waveform just out of the signal generator), is this useful in certain applications - and perhaps undesirable in others?
k1mgy I didn't show the phase delay, but there certainly is a delay encountered as you near the GBP limit, and a further delay as you hit the slew rate limit. It would be unwise to use this in a design because the performance, particularly slew rate, may not be well controlled between devices, and will vary depending on operating conditions. Not shown was that the asymmetry of the rising and falling slew rate limit is also dependent on the input common mode voltage.
Dear w2aew,
I really am a big fan of your videos, this time however I was surprised to notice you made a common mistake at around 2:23. You are saying that for a single pole response the product of the DC-gain and the bandwidth is constant, however this is _not_ true for the inverting amplifier you are demonstrating: it should be: (1+R2/R1)*BW=unity gain frequency. As long as R2/R1 is much larger (10 times or more) than one, the difference in bandwidth will not be that large, but for small values of R2/R1the difference between your prediction and the correct formula can be a factor 2! If the unity gain frequency of your op amp is 1MHz you should expect for a single pole response a bandwidth of 500kHz and not 1MHz. This also means that the -20dB/dec slopes of the closed loop configurations will not touch the -20dB/dec slope of the open loop system, but will be shifted to the left of it. Unfortunately you did not measure the small signal bandwidth for R2/R1=1, you should apply an input voltage whose amplitude is small enough not to have slew rate distortion. I wrote an article concerning this matter "A closer look at the slew rate criterion" in 2004. Maybe you could check it out and remeasure the bandwidth for R2/R1=1. In reality you probably will find a somewhat higher bandwidth due to the fact that a real op amp always is a higher order system, which especially can be seen in the case where R2/R1=1, sometimes noticeable as peaking in the amplitude characteristic, the asymptotic slope of the amplifier being steeper than -20dB/dec and the phase characteristic shifting over more than 90° relative to the phase at low frequencies.
kind regards,
Hugo
Thanks Hugo. I chose not to get into this level of detail in the "Basics" type of video, but I greatly appreciate your reply. Other viewers will benefit from reading your reply here, as well as your article:
lirias.kuleuven.be/bitstream/123456789/240890/1/artikelextended.pdf
Hello w2aew, may I encourage you to add a follow-up video to re-measure the unity-gain BW at a sufficiently low amplitude to stay under the slew-rate limit. It would be very beneficial to test the theory as expressed by Hugo Coolens regarding deviation from constant gain-BW product for closed-loop gains that are under 10. Cheers.
Great as always! Thank you!
This was extremely concise and well paced and informative. Thanks so much for your videos!
I had a few questions about the caps in your test circuit. I'm assuming the 0.1uF cap going from your divider reference to ground is a bypass cap to limit noise from the op amp, right?
As for the 220uF coupling cap, was there any reasoning in picking that value or is it what you had on hand? Additionally, would this cap and the input impedance of the op amp form a high-pass filter? It seems like the 3db point would be nearly 0?
I'm still learning the fundamentals and trying to piece together all these EE ideas in an intuitive way. Your videos have helped immensely in that quest.
Thanks again
Yes, cap values aren't critical. The 0.1uF is to keep the noise down at the non-inverting input. The 220uF is what I had laying on the bench, and was large enough to be "invisible" at the lowest frequency I intended to test with (~1kHz).
Thanks for such quality tutorial!
this a classroom should be thank you sir !
As usual, so clear to make me understand at first time! Are IC switching power regulators in your "to do" list?
Max Petrus That's another that's been on my list for a long time. Just haven't had the hours it takes to put something together on that yet.
I wish i could vote it up a few more times so that Mr.Alan Wolke does it sooner :).
This is briliant explanation. Thank you.
As usual, another excellent and lucid video.
I was wondering why you used attenuator and didn't just change the singal amplitude from the FG.
Thanx for the great video.
I wanted an amplitude (under the high gain case) that was below the minimum output level of the generator.
Thank you Alan. Superb....
Fantastic as always, thanks.
thanks a lot from the depth of my heart
Holy shit mate, i just noticed the scope you have, That's a serious instrument, around $40,000 if i'm not mistaken.......VERY COOL 1GHz on the front end
It helps when you work for the company...
@@w2aew
LOL....... Indeed
hey... Still Seriously fucking cool
see... if, when you leave the company it can "somehow" get misplaced
LOL
Me.. i can afford that, thats' not the problem
JUSTIFYING THE COST is the problem
and when i look at my MSO5000 Rigol
Honestly, when i look at Bandwidth vs sampling rate if all 4 channels are enabled
i can't really see how i justify the additional $40,000 with merely the 1GHz front end
i do lots of scope work, but i can't justify $40,000 if it's just extra bandwidth,
but yeah, still cool
Excellent explanation!
Really as always. Great teacher
Thanks very much.
Thank you for the nice video. Great stuff
Thanks again, Alan. Great vid.
Platinum!!
thank you very much
Thanks for the heads up in the other comment section. I have watched this video a few times before. My op amp tester is based on a similar type of circuit with a dual polarity supply instead of the bias.
I guess my challenge is to learn how to replicate what your doing with the function generator. ..Without a function generator...yet 🤔 ...but with predictable, repeatable results.. 🙄
My snipe hunting list for eBay is getting ridiculous... I'm starting to think hotrodding cars was the cheaper hobby.
Thanks again,
-Jake
You don't really need the complicated waveform when doing op amp slew rate testing. A simple square wave will do.
Thank you.
Thank you for your efforts. This is an excellent video.
As always a great video! :)
Good video!
Why did you stop increasing the frequency when the voltage dropped to 700mV?
Because that is the -3dB point, which is where the bandwidth is traditionally defined.
Very informative, thanks!
thumbs up before I watched it - for the topic
I'm confused on what is the difference in the op amp and Logic TTL IC chips slew rate specs and rise time specs because at a certain frequency the output pin will be a triangle shape. What I'm confused about is Logic IC chips have rise time specs and Op amps have Slew rate specs but they mean the same thing because they are referring to how fast the output signal can change with time. The rise time specs in the datasheets is using +5vdc logic pulse signal to measure the rise time specs and the Slew rate specs in the datasheets is using an analog sinewaveform to measure the slew rate specs. The rite time specs and slew rate specs are mostly in the nanoseconds nS but how do you find the frequency at which the slew rate specs and rise time specs are a problem? To find the slew rate spec/rise time spec is you use a function generator and change frequency until the op amp or IC logic TTL chips output signal is a triangle and that is the "Problem frequency" which is converted into the slew rate spec/rise time spec on the data sheet?
A logic circuit is designed to switch between two levels, so it makes sense to talk about rise and fall time. An op amp circuit is NOT typically used in an application where it is switching between two values quickly like a logic circuit. It is often used as an amplifier, active filter, signal conditioner, etc. and the bandwidth of the circuit will vary with the application. The bandwidth will determine the effective rise/fall time. There is a limit to how fast it can swing, due mainly to internal compensation, and this is slew rate. Since slew rate is the "slope" of the signal, the rise time will be a function of how far it is swinging - since there is no standard output swing level with an op amp, there isn't a fixed rise/fall time spec.
@@w2aew What I'm saying is what formula do you use to compute to get the Rise time Spec and Slew Rate Spec on the datasheets for op amps and TTL logic chips? I think they use a function generator and increase the frequency UNTIL the output signal is a triangle waveform then that is the "bad frequency" which they covert using a formula for the Rise time spec and slew rate spec? The Slew Rate formula is 2 X pi X Frequency X Vp , But when using a function generator and increasing the input pin of an Op amp or TTL chip UNTIL then output pin signal is a triangle waveform that is when you got the "bad frequency" which you convert into slew rate/rise time spec that is on the datasheets of the Op amp or TTL chip?
@@w2aew When a TTL Chip datasheets says the rise time specs it means how FAST the output signal can change states, so the rise time specs is the same thing as slew rate specs for op amps. If you set up a 2 channel Oscope that ch#1 is input pin of the TTL chip or Op amp and Ch#2 is the output pin of the TTL chip or Op amp to measure the Rise time/slew rate it should be in the nS nanoseconds but what is confusing is the datasheets rise time/slew rate specs don't tell you at what " frequency" is this spec at? example if the rise time spec is 5nS but at what frequency? is the datasheets says the op amp is 10uS/1mS at what frequency? Can you make a video lesson using a 2 channel Oscope with various TTL chips and Op amps and measure the rise time specs and Slew Rate specs to MATCH it to the datasheets with the Oscope and how to compute using formulas to calculate the rise time/slew rate specs on the datasheets?
@@waynegram8907 There is no formula for calculating the rise/fall time of TTL logic chips - it simply must be measured. It does not matter what frequency is used, the output rise/fall time will be the same regardless. For op amps, you can estimate the rise time from the formula 0.35/BW, where BW is the 3dB bandwidth of the circuit. This formula assumes that the op amp is not slew rate limited. The slew rate limit is based on the internal design of the compensation network - it can't be computed from any spec on the datasheet.
@@waynegram8907 An op amp output rise time can often be much slower than the slew rate - it all depends on the bandwidth of the op amp circuit.
Thanks so much for your videos
Hey Alan,
Thanks for uploading another great video!
I hate to be picky, but I noticed something in your diagram of slew-rate-limiting at 10:20.
You drew the slew-rate limited waveform being in-phase with the the input signal (or a non-slew-rate-limited output signal).
However, shouldn't the slew-rate-limited signal lag the input signal by some phase?
The drawing wasn't intended to show input vs output, but rather just comparing normal vs. slew rate limited output waveform shapes. If it was showing input vs. output, then yes there would be a phase shift.
At minute 3:38 the formula for BW is very similar to the formula for calculating the scope BW based on the signal rise time. The diff is in the 0.35 constant. Are they really related?
Yes - it's the same single-pole rolloff relationship.
Thank you for the video and clear explanation i have learned soemthing new today.
Great work man,keep up
Thanks for this awesome video
I just spend 5 minutes checking my mail, only to realize the mail notifications are in the video :)
Oops, sorry about that !
Is there any difference in gain bandwidth product between inverting amp and non-inverting amp while calculating V out ?
I mean I watched a video regarding GBP and teller said
Fu = Acl x Fcl in Non-invering amp
Fu= ( |Acl |+ 1 ) x Fcl in inverting opamp
Does it make sense ?
Thanks a lot, exelent video, keep up the good work :-)
This was really helpful
I was thinking, if you put two common emmiter transistor amps on the output of the opamp and take feedback from the collector resistor of the second transistor back to the opamp feedback resistor network, would that somehow enhance slew rate or/and the gBW product?
Nice refreshing content!
Great Video. I did not know that 2 different aspects effect the BW of the opamp. I used to believe that slew rate was the only cause of the band width limitation.Thanks to you video the concept is very clear.
Any reason for using the Attenuator. Can the experiment be done by varying amplitude in the signal generator controls ?
Mainly because my signal generator output control didn't have the range I wanted - plus the attenuator gives the advantage of maintaining the generator's SNR for small signals (since the attenuator reduces the signal and the noise equally).
Oh ok. Thank you :)
Very nice! thanks :)
Excellent video. Few questions though. How did you come up with all the 220uF (reverse polarity here I think), 470 and 0.1 uF ? Aren't the 220 and 470 too big ? Would some bulk decoupling like 10uF instead of 470uF work ? Thanks.
Nothing critical. 470uF for good bulk supply filtering, value not critical. 220uF for input coupling, just a large value laying on the bench that would provide a low impedance a the low-frequency end of the testing, and 0.1uF for decent medium/high frequency decoupling of noise at the non-inverting input. In all cases, values could vary quite a bit with no change. Just handy values found on the mess of a bench...
Excellent.
Awesome! thank you
Thank for share, very usefull for me, for make good preamp
Regards,
can reasons why the slew rate and the frequency response gets worse when increasing the gain on an op amp? what is causing inside the op amp that would make the high frequency and slew rate get worse from increasing the gain higher on the op amp? This is called gain bandwidth product but not sure what is causing it to happen
Mainly due to the limited amount of current available to charge/discharge the internal compensation capacitor.
So even if the power supply can supply double the current to the op amp its still will be due to the limited amount of current available to the charge/discharge the internal compensation capacitor? you can ADD an "external compensation" capacitor to the op amp to fix the gain bandwidth product?
Thank You!
Slew rates are awesome.
Great video! One question: Is your attenuator placed between your function generator and your op amp input? So the poor mans solution is to simply keep reducing our function generator amplitude as we increase the opamp gain?
Yes. My generator couldn't go low enough. If yours can't, a simple voltage divider will do fine.
I see. Thanks for the reply!
How didi u select the value of resistors and capacitors
Nothing special. I wanted two equal values to setup a 50:50 voltage divider for the non-inverting input. I have a big pile of 3.9k resistors, so I used two of them. For the gain setting resistors, I chose values that are simple "in your head" calculations.
Brilliant!!