ECE4450 L4.1: Voltage Controlled Amplifiers: Operational Transconductance Amps (ACMS)
HTML-код
- Опубликовано: 2 авг 2024
- Support this channel via a special purpose donation to the Georgia Tech Foundation (GTF210000920), earmarked for my work: • Support This Channel v...
[Whoops: The title slide should say "Voltage-to-Current," not "Current-to-Voltage"]
I prepared this slides deck for a lecture in the Spring 2020 offering of my Georgia Tech class, ECE4450: Analog Circuits for Music Synthesis. Enough people on youtube liked my Moog Ladder lecture that I thought I'd start doing more of these, where I try lecturing into a microphone instead of to a live class. OTAs will appear in future posted lectures on voltage controlled oscillators and filters.
Near the end, I think I accidentally say "SSM2161" when I meant to say "SSM2164."
As far as I know, the circuit at 6:10 has the advantage of keeping the OTAs output pin exactly at GND potential, which makes the OTA output much more linear whereas the simple resistor at the output makes the OTA output pin move up and down with the signal and creates additional distortion.
Good point! (I think there's similar reasoning behind holding the output of the SSM2164/SSI2164 at virtual ground)
Holy! What did just happen?! I just understood wave shapeing in 30 minutes without any questions left, thank u so much for this great work! I love ur lectures! Even as a non native English speaking person and i would say i'm also not the best at this, i could easy follow ur video!
I found an amusing sentence on the wikipedia page for the Wilson Current mirror - the 3 BJT mirror used in the TCA. 'Wilson devised this configuration in 1967 when he and Barrie Gilbert challenged each other to find an improved current mirror overnight that
would use only three transistors. Wilson won the challenge'. wow that is serious cleverness (or accident?!)
From South Korea-Thank you very much Aaron
I wish I had also lecturers at college that explained stuff with analogies to fictional characters from retro sci-fi movies.
Your online lectures are a treasure, even for experienced analog designers. Thank you!
One way to think about the Moog Taurus using the hyperbolic tangent function of the OTA is that it's distorting, but very gently so it's acting kind of like compressor and ends up increasing the RMS value of the waveform, especially with respect to the fundamental (presuming your cutoff frequency is higher than that). People always complain about how adding resonance to the Moog VCF causes bass loss, which is true in the sense of relative level if you don't compress/clip the resonant peak of the filter and instead compensate it by reducing overall level (which is how the resonance knob works). This is why people often would feed the headphone output back into the external input to drive the filter and clip the resonant peak so that the fundamental was relatively louder at the cost of distorting /squashing the resonant peak.
So, I think, since this is specifically a bass instrument, Moog was probably emulating that trick using the OTA hyperbolic tangent curve to act as kind of a resonance limiter/"bass preserver." Just a guess :)
Thank you for lectures! Respect from Wrocław PL
INformative and understandable for a former electronics technician now expoloring modular synthesizers.
I'm so glad you're uploading again! So jealous of everyone that gets to be in your class.
I'm going to keep putting these up as I have time. :)
I absolutely love your Tina Turner and Master Blaster comparison- you have awesome teaching skills.
Thanks! That's a highly underrated movie. :)
I wish I would have had this course in my EE undergrad. Synthesizers are the entire reason I studied EE to begin with. Absolutely fantastic course!
Thank you for your kind words!
Excellent lecture. Thank you
Your comment about "Warmth" and nonlinearities in DSP at 9:40 made me cheer out loud :D Thank you so much for uploading your lectures!
I am going through all the modules and today it was VCA time. A recap on the theory, recreating all the formulas myself and then creating my own schematics is much more satisfying than just asking google for some schematics to copy.
After calculating my component values, simulation and a quick test on a breadboard I changed the design a bit.
There was a DC offset on the output signal and that DC offset increases when increasing Iabc, most likely due to a small offset in the opamp's output. Fixed that by: less attenuation on the audio input opamp, a larger attenuation on the resistor network on the OTA input and reducing those resistors to 10k/680 ohm. This resulted in a (I think too large) linearity error which I then removed by adding a resistor on the diode bias input. This also removed the DC offset (or reduced it far enough). I noticed that the datasheet places a potentiometer between the + and - inputs with the wiper to ground. As far as I know that is done to trim the DC offset.
Yup that's a DC offset trim.
thanks so much for putting in the effort of these videos - I graduated EE 6 years ago uninspired and went into a boring software job, but now I am re-learning all the analog fundamentals that made me want to do the degree to start and this is providing the inspiration!
Thank you for your kind words!
Thank you for making all these !!
These videos are fantastic! Thanks for sharing.
Great video, thank you!
It's not obvious to the newbie that Iabc is referenced to the V- input instead of ground. That was the first lesson. The next lesson was that the input has to be divided down so far that an enormous makeup gain is necessary. This makes the design very sensitive to small distortions and CV feed through. I guess then the temperature dependency needs to be considered. Perhaps in audio signals this all works out, but for CV VCAs, it isn't easy. I was unable to make this device work adequately. Because I failed, I suspect that I didn't fully come to understand it. Instead of digging further, I just moved to the SSM2164 or (eventually) to a THAT300 based log/antilog VCA. Great lecture and thanks for mentioning what appears to be a useful book.
After watching a few of your videos I did something that I haven't done in like five years. I searched for a youtube downloader! :D I'm guessing you are not doing this for the money but if you put up a patreon or something I'll be the first one to sign up!
Whoa. Was not expecting a Mad Max metaphor in this.
Tina Turner is just so awesome in that movie. I love her last words to Max at the end. She's an antagonist in that she and Max are at odds, but she's just doing what she thinks is best for her people.
Oh my gosh I am so excited for the exponential converter lecture, this is exactly the info I need right now. Looking at Musical Apps of MPU's for the meantime, thanks for the recommendation.
Alas the expo converter is a topic I don't already have slides made up for, but I will hopefully get around to it at some point. Next up I have some filter lectures ready to record.
Math math math math math. Thumbs up if you are surrounded by HP48s. I will accept newer TI's (I suppose). I have 8 or 10 HP48s. Each one has different equations programmed into it. Some do electronics. Some do acoustics. A couple of them do long range rifle ballistics, and a couple do land surveying, and fluid dynamics. My dad was a surveyor and a hydrologist. I still have RCA CA3094 OTAs kicking about. Some sort of a envelope detector/filter controlled phase shifter effect? Prophet 5s!! Love it. The Prophet 10 was just two Prophet 5s built atop one another. Yes, one hand for each keyboard.
What are the benefits of the presented linear V-to-I converter over a buffering opamp and a resistor between Vcin and the OTA's I abc input? Thanks for the video!
The main thing is that the voltage at the input of the OTA's Iabc will slide around a bit -- it will be "around" a "diode drop" or two (depending on the OTA -- you can look at the rough internal schematic given on most data sheets), but that "diode drop" of 0.65V or 0.7V is an approximation. So you can roughly rely on it, but not precisely rely on it. Using the op amp lets you generate the current by putting a voltage through a resistor to a fixed voltage reference.
Amazing lecture! One question. Is it possible to use this (V - I) linear converter with single voltage supply (0/+9v) ?
I'd need to think about how to modify it to work with a unipolar supply; the raw circuit requires the op amp to have the ability to spit out negative voltages. I'm sure it's possible to come up with something... you might need a 2nd op amp.
Hi, thanks for these lectures. You say at 2148 that for a current sink you could re-arrange the circuit to use an NPN transistor instead of the PNP shown. I'm not good enough at electronics yet to do this myself. Do you know of any schematics showing the VtoI converter as a current sink using an NPN transistor?
Thanks again.
You can just flip the order of all the current arrows (including obviously the one inside the transistor) if you first *invert* the voltage going in. You could put a 2nd op amp in front to do that. There's probably more elegant approaches but that's the first thing I thought of.
Isn't scaling down the signal in front of the OTA worsening the signal-to-noise ratio of the circuit? If so, is it significant? Could we avoid it somehow without introducing distortion? Use a different OTA maybe?
It totally is making the SNR -- it's why you generally won't find OTAs used in studio compressors; the only one I'm aware of is earlier versions of the BSS DPR-402. You will see it used in guitar stomp boxes. The 3280 is better than the 3080/17000 but not lots better. For higher quality, one could go with a more modern "VCA" chip like the SSI2164. In general, though, I think the weird grunginess of the OTA is considered part of the sound of some of those classic filters.
At 22:00 - what do we need R3 for? The whole calculation would work without it, or am I missing something?
EXCELLENT question!!! It's there to help with practical issues on picking component values. Without it, R2 drops out, and you have to pick R1 to get all the current you want, but this often R1 value winds up being very low, so you wind up with your circuit having a very low input impedance than might load down the circuit feeding it. In some applications could go ahead and leave out R2 and R3 and pick an R1 that's low if you have an op amp voltage buffer feeding R1 -- but then the op amp feeding it has to supply all of the current. By using R2 and R3, the transistor is doing the "heavy lifting" of moving the current around.