@@jeremycyclist6269 definitely didn’t understand that in the beginning. I remember my first “Non Filter Camel” My first opamp circuit made me cough just the same. 😂
those are for a load if needed, also a capacitor if needed. many datasheet tests need these locations. of course you don't need to load any or all of the components on this board, it is for testing
Awesome content! I love your op amp theory videos. I have a few questions on a project I am working on, as a fellow ham, would you mind if I picked your brain over a few analog design concepts?
OK, interesting stuff but why is this the case? What is going on internally in the amp that causes these results? In the 10X10 configuration is the output the same at 1kHz as the 100X configuration? You have left me with more questions than answers, but that is not all bad. Makes me have to think about it some.
The bandwidth of an OP amp is specified as GBWP (Gain Bandwidth Product) and it is the product of unity Bandwidth/ Gain. So if an OP Amp has a GBW of 1MHz and you set the feedback gain as 10, then the total GBWP will be 100KHz. It has to do with the feedback loop and the internals. Also, gain of 100 equals to gain of 10x10.
There is also deliberate feedback capacitance built in to roll off the high frequency response so the op amp will be stable when configured for a voltage gain of one (unity gain stable). Some op amps do not have this feature, so they can be used at higher gains with greater bandwidth, but these will be unstable when configured for low gains.
The gain-bandwidth tradeoff is easier to understand when you kind of invert what you're looking at. Don't consider an output signal that can grow infinitely large as the gain approaches infinity, rather the opposite, consider an input signal that can be made infinitesimally small as the gain approaches infinity. As the signal level approaches the noise floor, the wider the bandwidth, the higher the potential noise amplitudes can be in the time domain if all the vector elements of the noise frequency components align just right. This can be enough to interfere with the discrimination of signals of similar amplitude. So we must decrease the bandwidth of the amplifier to chop off some of those noise frequency components, to limit potential time-domain amplitudes of those unwanted components. And whether its analog or digital, all signal detection fundamentally comes down to instantaneous amplitudes.
A good demonstration of GBWP there. I wonder - you made a 100x gain out of two 10x gain stages, but would it behave differently if you did unequal stages - perhaps 20x 5x or somesuch? GBWP is one thing but so is slew rate - maybe it helps to amplify the smaller signal a bit more, knowing that the larger one doesn't have to cope with quite as much slew rate change at the end? Could be a fun follow-up to this one.
I think cascaded the network is ultimately bandlimited by the larger gain (and thus smaller BW), so to maximize your available BW you'd want them to be equal.
Nice! Bandwidth! Laaaaaarrrge! Unless CW, or SSB. Or make an audio chain that is limited to human hearing abilities, then make it an universal standard, so no generally available device could work ultrasound.
i guess its the -3db point or the cut-off frequency at which u half the power of the signal by ur filter, in this case the opamp. if u apply it 2 times in serial its seems to be a good approximation that u get half the cutoff frequency. 300/2 khz
If you had 25 boards running at a gain of 2 x 2, one might get a bandwidth of 500kHz - 1 MHz... lol. I haven't got 25 of those boards so not going to try that concept..
Nice little boards. Great for learning and testing ideas. Thank you.
Never knew that. I’m going to going out your boards that I purchased a while ago and try a few of my op amps. Thanks very much.
thank you!
"It's a thing", I didn't know, now I do. Thanks!
The more you stage, the better they act.
Don’t forget that each additional opamp will add more noise.
@@jeremycyclist6269 definitely didn’t understand that in the beginning. I remember my first “Non Filter Camel”
My first opamp circuit made me cough just the same. 😂
Like my first 'Woodbine'@@jj74qformerlyjailbreak3
no opamp no gain
No woman no cry.
The top left and right op amp configured wrong input resistor in series should be to gound
The bottom left and right op amp configured wrong ouput resistor should be in series not to ground because op amp has low impedance at the output.
those are for a load if needed, also a capacitor if needed. many datasheet tests need these locations. of course you don't need to load any or all of the components on this board, it is for testing
Awesome content! I love your op amp theory videos. I have a few questions on a project I am working on, as a fellow ham, would you mind if I picked your brain over a few analog design concepts?
I try to answer as many questions as I can.
OK, interesting stuff but why is this the case? What is going on internally in the amp that causes these results?
In the 10X10 configuration is the output the same at 1kHz as the 100X configuration?
You have left me with more questions than answers, but that is not all bad. Makes me have to think about it some.
The bandwidth of an OP amp is specified as GBWP (Gain Bandwidth Product) and it is the product of unity Bandwidth/ Gain. So if an OP Amp has a GBW of 1MHz and you set the feedback gain as 10, then the total GBWP will be 100KHz. It has to do with the feedback loop and the internals. Also, gain of 100 equals to gain of 10x10.
www.ti.com/lit/sboa356
I think because there are parasitic capacitances on input/outputs of the amplifier/transistors/whatever that begin to short at certain frequencies.
There is also deliberate feedback capacitance built in to roll off the high frequency response so the op amp will be stable when configured for a voltage gain of one (unity gain stable). Some op amps do not have this feature, so they can be used at higher gains with greater bandwidth, but these will be unstable when configured for low gains.
The gain-bandwidth tradeoff is easier to understand when you kind of invert what you're looking at. Don't consider an output signal that can grow infinitely large as the gain approaches infinity, rather the opposite, consider an input signal that can be made infinitesimally small as the gain approaches infinity. As the signal level approaches the noise floor, the wider the bandwidth, the higher the potential noise amplitudes can be in the time domain if all the vector elements of the noise frequency components align just right. This can be enough to interfere with the discrimination of signals of similar amplitude. So we must decrease the bandwidth of the amplifier to chop off some of those noise frequency components, to limit potential time-domain amplitudes of those unwanted components. And whether its analog or digital, all signal detection fundamentally comes down to instantaneous amplitudes.
A good demonstration of GBWP there.
I wonder - you made a 100x gain out of two 10x gain stages, but would it behave differently if you did unequal stages - perhaps 20x 5x or somesuch? GBWP is one thing but so is slew rate - maybe it helps to amplify the smaller signal a bit more, knowing that the larger one doesn't have to cope with quite as much slew rate change at the end? Could be a fun follow-up to this one.
I think cascaded the network is ultimately bandlimited by the larger gain (and thus smaller BW), so to maximize your available BW you'd want them to be equal.
Nice! Bandwidth! Laaaaaarrrge! Unless CW, or SSB. Or make an audio chain that is limited to human hearing abilities, then make it an universal standard, so no generally available device could work ultrasound.
I'm glad you did this video. Is the gain vs. bandwidth linear? Or does its curve varies? From IC to IC? Each specimen?
I suggest you look at an opamp datasheet. there is a plot of frequency and gain. in general it is linear
Thank You a lot. My love for electronics is renewed.
do you have a video of the top of the board being put together and could i use this with my rtl-sdr ?
I don't think it would work beyond a few MHz.
a few 100s of kHz at best
😊
What bandwidth should have 10x10 version? 3MHz/10, so 300kHz?
www.ti.com/lit/sboa356
@@IMSAIGuy So it should be 300kHz? You measured 150kHz. What's the cuplrit?
read the paper I sent. if you cascade a 300khz and 300khz. you end up with less than 300
i guess its the -3db point or the cut-off frequency at which u half the power of the signal by ur filter, in this case the opamp. if u apply it 2 times in serial its seems to be a good approximation that u get half the cutoff frequency. 300/2 khz
the formula is 1/f is sqrt(1/f1^2 + 1/f2^2) where f1 and f2 are bandwidths of the two amps and f is the resulting bandwidth
If you had 25 boards running at a gain of 2 x 2, one might get a bandwidth of 500kHz - 1 MHz... lol. I haven't got 25 of those boards so not going to try that concept..
3:04 4kHz !