I've been quietly lurking and watching FesZ's videos for some time now, and I have to say this guy is my absolute favorite on this subject. Thorough, showing the listener each step and how to evaluate, model, and build really quality stuff. This guy's a gem, I wish he was the guy teaching me when I learned EE back in the day.
this design saved my butt. I am currently working on a loadcell signal amplifier that is meant to sense rapid changes in force. I am 17 right now so I don't have a lot of knowledge. this is pure gold
I once had to design a bio instrumentation amplifier to pickup eye movement signals, which unconveniently lied in the 0hz -15 hz range, so basically in the flicker noise of almost if not all opamps. I got around it by modulating my signal to 10khz since at that frequency only thermal noise exists, filtering and then demodulating the resulting signal after all the processing. Of course from a simplicity point of view, it isn't so viable since you have to have a dedicated oscillator, two modulators, multiple filtering stages, peak detectors, etc. But it was interesting nevertheless. Keep up the videos, i really enjoy your content!
Sounds like a really interesting design! I guess that is one of the best ways to build the circuit in case you wish to make a really low frequency amplifier.
Bloody hell! I finally know and understand how to to interpret datasheet graph! Only for the noise part, that is. Spike is called 'flicker noise' and we do not want that of we are working at that particular frequency, in this case is 10 Hz. The steady line is called 'thermal noise' and is all right, we can use it.
Although I am struggling a bit with my rusty electronics knowledge and have some trouble completely understanding schematic and basic principle how filtering and amplification is achieved with your op amps circuits. But I have to write that you have excellent channel and I am looking forward for exploring it more. Subscribed... Good luck.
Yes, LM358's suck, but they are very affordable. It's always good to have some of them around. I use them a lot to test if prototypes are wired correctly: No smoke -> plug in the actual op amp.
Nice video again. In theory you can actually make a fourth order filter with just one opamp (the math is a bit of a thing), so also an second order bandpass (for high and low). Depending on your needs it can be beneficial to use a multiple feedback filter instead of a sallenkey.
That will also work of course. The beauty of electronics is that there is no single correct solution. Any problem can have multiple ways of solving each leading to mostly the same result. For me it was this method, but any other method is just as true. I plan on finishing this project as is, but in the future I may revisit the idea and try to improve on it, maybe try out this sort of filter.
I know price was a concern but did you look at any instrumentation amplifiers? We use them alot in biomedical applications where high gain and low noise is a must
Hello Thaej! I guess the main reason why I didn't go for an instrumentation amplifier was because I didn't want to have a differential input. Maybe at some point I will make a follow up to this video with that sort of implementation though. What instrumentation amplifier IC's would you recommend?
Thank you so much. my challenge is how to amplify a thermopile signal (10uv -1mv) using only LM358 that is available in my county nowadays because of the covid-19 pandemic! is there any way to compensate for the amplifier noise?
Hello! Well the thermocouple (thermopile) is not a particularly fast sensor so you can manage the noise by severely reducing bandwidth - you can add a capacitor over the feedback resistor. Another thing to keep in mind is the input offset of the part - the LM358 has a maximum of 3mV. If the offset doesn't vary the circuit will work as long as you compensate this
Hello Dimuthu ! Well there are formulas for this sort of filter - for example here at the end of the page www.electronics-tutorials.ws/filter/filter_6.html (and chapter 5 is about the low pass version) But to be honest, you only have certain resistor values available (depending on tolerance and your personal stock) so I just tried out a few combinations until I got the desired effect. You can do this a bit faster by using .step commands
12:02 - Not multiplied by 8 but by 6. This is due to the theory and your measurements that you present in Part 3 (czcams.com/video/2mGKvGYwWrk/video.html). This gives a 33 percent difference. :-) Good job.
This is quite a relative subject. I've found different values for the relationship between noise and RMS voltage. On page 4 of ( training.ti.com/system/files/docs/1312%20-%20Noise%202%20-%20slides.pdf ) its x6; on page 6 of ( www.renesas.com/eu/en/doc/misc/noise-in-the-signal-chain-webinar.pdf ) and page 3 of ( www.analog.com/media/en/training-seminars/tutorials/MT-048.pdf ) its x6.6 but in the end its a probability topic. The 6x factor comes from 6sigma=99.7% probability that Vpp will cover the 6xVrms. By going up to a factor of 8 you ensure that the value of Vpp will cover your measurement in 99.99% of cases.
@@FesZElectronics Yes. It is indeed so. This factor (6 or 8) depends on the confidence interval, but in practice 6.6 is most often used. You are right and I have some of it. In your measurement from 3 parts (czcams.com/video/2mGKvGYwWrk/video.html) Vrms = 1.44mV and Vpp = 8.72mV. The ratio is therefore 6.06. BTW, why the measurement is cyclic RMS? What your oscilloscope really measures. Where's period in the noise?
To be honest I wasn't paying attention to the type of RMS, I don't think there is much difference between the 2 when measuring noise since there is no cycle to stick to. Anyway I don't really trust the automated measurements on the Hantek, they seem to be off from time to time. Also for a proper noise measurement, that oscilloscope has quite few data points (mine can do a maximum of 16k I think; In theory it has a memory upgrade up to 1M but it just can't process it very well... various measurements like the FFt will limit the memory to 4k even with the memory upgrade)
@@FesZElectronics This is not the point. The measurement of Cyclic RMS is based on the fact that the oscilloscope finds the period of the waveform and integrates it only in this range. This avoids the error of too narrow or too wide integration range, which would cause errors in determining Vrms. And there is no period in the noise, so how does the oscilloscope count crms? Probably takes the first case that would resemble a period - e.g. the time interval between the intersection of the rising edge of the signal by a certain voltage level. In oscilloscopes that not has crms measure, the rms voltage is calculated from the integral of everything that is in the acquisition record. But we're leaving the topic.
@Jarodoli I looked up the manual for this oscilloscope ( www.circuitspecialists.com/content/183992/dso5102p_manual.pdf ) in chapter 5.5.2 on page 33 at the bottom they explain CyclicRMS and PeriodRMS. Apparently the CRMS is just the regular RMS over the entire range and PRMS is what we know as CRMS. Guess its a translation issue, but I tried out looking at some random noise, and the PRMS is much smaller than the CRMS so the description is correct.
Well since you want 1MHz of bandwidth and 20dB of gain the op-amp should have at least 10MHz of bandwidth, the more the merrier. I guess the hardest part will be to select a low noise circuit and re-adapting the filters. But this should be easy to test in the simulator.
Unfortunately the manufacturer does not provide a model on their website, or at least I could not find it; since this is an op-amp, you could look for a similar or equivalent component from a different manufacturer that does have a model and just use that... LTspice has a ton of libraries built in for LT and AnalogDevices opamps, there must be something similar
Why do you say that, in the AC simulations its visible - example czcams.com/video/Car8hk_ubJs/video.html ; the noise simulation, the one that I do most often in the video, does not contain phase information though
I've been quietly lurking and watching FesZ's videos for some time now, and I have to say this guy is my absolute favorite on this subject. Thorough, showing the listener each step and how to evaluate, model, and build really quality stuff. This guy's a gem, I wish he was the guy teaching me when I learned EE back in the day.
this design saved my butt. I am currently working on a loadcell signal amplifier that is meant to sense rapid changes in force. I am 17 right now so I don't have a lot of knowledge. this is pure gold
I once had to design a bio instrumentation amplifier to pickup eye movement signals, which unconveniently lied in the 0hz -15 hz range, so basically in the flicker noise of almost if not all opamps. I got around it by modulating my signal to 10khz since at that frequency only thermal noise exists, filtering and then demodulating the resulting signal after all the processing. Of course from a simplicity point of view, it isn't so viable since you have to have a dedicated oscillator, two modulators, multiple filtering stages, peak detectors, etc. But it was interesting nevertheless. Keep up the videos, i really enjoy your content!
Sounds like a really interesting design! I guess that is one of the best ways to build the circuit in case you wish to make a really low frequency amplifier.
Wow. That's a neat solution! Never would have thought of that!
Genius idea!
@@oojikarasuma1it's a good idea, but a very well-known technique called Chopping
nice content
Pure gold ☺️
Bloody hell! I finally know and understand how to to interpret datasheet graph! Only for the noise part, that is. Spike is called 'flicker noise' and we do not want that of we are working at that particular frequency, in this case is 10 Hz. The steady line is called 'thermal noise' and is all right, we can use it.
Thanks so much!
Although I am struggling a bit with my rusty electronics knowledge and have some trouble completely understanding schematic and basic principle how filtering and amplification is achieved with your op amps circuits. But I have to write that you have excellent channel and I am looking forward for exploring it more. Subscribed... Good luck.
Very nice explanation..!!
Yes, LM358's suck, but they are very affordable. It's always good to have some of them around. I use them a lot to test if prototypes are wired correctly: No smoke -> plug in the actual op amp.
Nice video again. In theory you can actually make a fourth order filter with just one opamp (the math is a bit of a thing), so also an second order bandpass (for high and low). Depending on your needs it can be beneficial to use a multiple feedback filter instead of a sallenkey.
That will also work of course. The beauty of electronics is that there is no single correct solution. Any problem can have multiple ways of solving each leading to mostly the same result. For me it was this method, but any other method is just as true.
I plan on finishing this project as is, but in the future I may revisit the idea and try to improve on it, maybe try out this sort of filter.
I know price was a concern but did you look at any instrumentation amplifiers? We use them alot in biomedical applications where high gain and low noise is a must
Hello Thaej! I guess the main reason why I didn't go for an instrumentation amplifier was because I didn't want to have a differential input. Maybe at some point I will make a follow up to this video with that sort of implementation though. What instrumentation amplifier IC's would you recommend?
Thank you so much. my challenge is how to amplify a thermopile signal (10uv -1mv) using only LM358 that is available in my county nowadays because of the covid-19 pandemic! is there any way to compensate for the amplifier noise?
Hello! Well the thermocouple (thermopile) is not a particularly fast sensor so you can manage the noise by severely reducing bandwidth - you can add a capacitor over the feedback resistor. Another thing to keep in mind is the input offset of the part - the LM358 has a maximum of 3mV. If the offset doesn't vary the circuit will work as long as you compensate this
@@FesZElectronics Thanks so much for your help and fast reply.
Hi .is it pussible to amplify a fast edge squre wave (rise time=30ns) with 10 Vp-p to 100 Vp-p and with rise time being less than 60ns?
can you explain how you calculate R and C values of the Low pass and High pass filter
Hello Dimuthu ! Well there are formulas for this sort of filter - for example here at the end of the page www.electronics-tutorials.ws/filter/filter_6.html (and chapter 5 is about the low pass version)
But to be honest, you only have certain resistor values available (depending on tolerance and your personal stock) so I just tried out a few combinations until I got the desired effect. You can do this a bit faster by using .step commands
12:02 - Not multiplied by 8 but by 6. This is due to the theory and your measurements that you present in Part 3 (czcams.com/video/2mGKvGYwWrk/video.html). This gives a 33 percent difference.
:-)
Good job.
This is quite a relative subject. I've found different values for the relationship between noise and RMS voltage. On page 4 of ( training.ti.com/system/files/docs/1312%20-%20Noise%202%20-%20slides.pdf ) its x6; on page 6 of ( www.renesas.com/eu/en/doc/misc/noise-in-the-signal-chain-webinar.pdf ) and page 3 of ( www.analog.com/media/en/training-seminars/tutorials/MT-048.pdf ) its x6.6 but in the end its a probability topic. The 6x factor comes from 6sigma=99.7% probability that Vpp will cover the 6xVrms. By going up to a factor of 8 you ensure that the value of Vpp will cover your measurement in 99.99% of cases.
@@FesZElectronics Yes. It is indeed so. This factor (6 or 8) depends on the confidence interval, but in practice 6.6 is most often used. You are right and I have some of it. In your measurement from 3 parts (czcams.com/video/2mGKvGYwWrk/video.html) Vrms = 1.44mV and Vpp = 8.72mV. The ratio is therefore 6.06.
BTW, why the measurement is cyclic RMS? What your oscilloscope really measures. Where's period in the noise?
To be honest I wasn't paying attention to the type of RMS, I don't think there is much difference between the 2 when measuring noise since there is no cycle to stick to. Anyway I don't really trust the automated measurements on the Hantek, they seem to be off from time to time. Also for a proper noise measurement, that oscilloscope has quite few data points (mine can do a maximum of 16k I think; In theory it has a memory upgrade up to 1M but it just can't process it very well... various measurements like the FFt will limit the memory to 4k even with the memory upgrade)
@@FesZElectronics This is not the point. The measurement of Cyclic RMS is based on the fact that the oscilloscope finds the period of the waveform and integrates it only in this range. This avoids the error of too narrow or too wide integration range, which would cause errors in determining Vrms. And there is no period in the noise, so how does the oscilloscope count crms? Probably takes the first case that would resemble a period - e.g. the time interval between the intersection of the rising edge of the signal by a certain voltage level.
In oscilloscopes that not has crms measure, the rms voltage is calculated from the integral of everything that is in the acquisition record.
But we're leaving the topic.
@Jarodoli I looked up the manual for this oscilloscope ( www.circuitspecialists.com/content/183992/dso5102p_manual.pdf ) in chapter 5.5.2 on page 33 at the bottom they explain CyclicRMS and PeriodRMS. Apparently the CRMS is just the regular RMS over the entire range and PRMS is what we know as CRMS. Guess its a translation issue, but I tried out looking at some random noise, and the PRMS is much smaller than the CRMS so the description is correct.
I want to amplify a 1 mhz signal from a transducer in two stages of 20db. Which ic can I use?
Well since you want 1MHz of bandwidth and 20dB of gain the op-amp should have at least 10MHz of bandwidth, the more the merrier. I guess the hardest part will be to select a low noise circuit and re-adapting the filters. But this should be easy to test in the simulator.
Hi, great content. I'd like to simulate the NCV33272ADR2G but I cannot find a .model or .subckt. Do you know where to find it or how to build it?
Unfortunately the manufacturer does not provide a model on their website, or at least I could not find it; since this is an op-amp, you could look for a similar or equivalent component from a different manufacturer that does have a model and just use that... LTspice has a ton of libraries built in for LT and AnalogDevices opamps, there must be something similar
@@FesZElectronics TI suggest TLE2142 but they don't look so similar... and it exhibits a lot of noise on low frequencies too (LTSpice).
why there is no phase shift in output as you are using RC filter....
Why do you say that, in the AC simulations its visible - example czcams.com/video/Car8hk_ubJs/video.html ; the noise simulation, the one that I do most often in the video, does not contain phase information though