The resonance stems from the bootstrapping effect, and will be present with all active devices. It is going to be stronger with devices having a gain closer to unity, but it will always be present.
The topology is a variety of active filter, and this behaviour is inherent to it. It can be attenuated by various means, like a proper choice of time constants or dumbing down the bootstrapping effect with a series resistor with the bootstrapping cap.
That said, the resonance looks very spectacular and harmful on the impedance, but it is in fact mostly benign for the rest, as this sim shows:
There, there is no hint of resonance, and in fact, having a higher impedance at some frequencies could even be beneficial. That said, one has to be careful since such resonances can sometimes turn a circuit into an oscillator
The topology is a variety of active filter, and this behaviour is inherent to it. It can be attenuated by various means, like a proper choice of time constants or dumbing down the bootstrapping effect with a series resistor with the bootstrapping cap.
That said, the resonance looks very spectacular and harmful on the impedance, but it is in fact mostly benign for the rest, as this sim shows:
There, there is no hint of resonance, and in fact, having a higher impedance at some frequencies could even be beneficial. That said, one has to be careful since such resonances can sometimes turn a circuit into an oscillator
here... maybe it is my biased view from the daily work. But, if I see a Lorentzian shape and phase changing 180°, then in my interpretation will sound like: ok, "something there" is acting as true resonance (LCR for example), with impedance before f0 being capacitive and after f0 inductive, or viceversa (similar cases purely in mathematical sense). It would be very useful to know if that is indeed the case in your opinion.
Yes, high risks of oscillations, depending on the physical interpretation of the resonance and less on which parameter it acts upon. Whether or not this is the case at the end of the day, it is designer's job and a complex circuit story to solve.
Well, for my level it is good to know when they exist by nature on a given circuit. Other minds have squeezed their heads around, maybe a life time career, to figure out and explain in simple educational terms how this effect comes about. I am not ready for this level, on the bootstraping//active filter effect. That's why I am simply happy to at least know these resonances do exist intrinsically, especially when the external circuit does not explicitly looking like a LCR (where is the stored energy?).
In a sense, the resonance can explain a lot of the intrinsic noise topic.
LV, I am totally humbled to experience this high level of teaching&expertise well explained. I feel the same about Marcel's posts. I'llstick around to learn more, if you can put up with my beginner questions.
P.S: what could be the resonance about for PSRR case? I mean, resonance for Z_in is still rather kind of intuitive and ubiquitous ... sort of.
Yes, high risks of oscillations, depending on the physical interpretation of the resonance and less on which parameter it acts upon. Whether or not this is the case at the end of the day, it is designer's job and a complex circuit story to solve.
Well, for my level it is good to know when they exist by nature on a given circuit. Other minds have squeezed their heads around, maybe a life time career, to figure out and explain in simple educational terms how this effect comes about. I am not ready for this level, on the bootstraping//active filter effect. That's why I am simply happy to at least know these resonances do exist intrinsically, especially when the external circuit does not explicitly looking like a LCR (where is the stored energy?).
In a sense, the resonance can explain a lot of the intrinsic noise topic.
LV, I am totally humbled to experience this high level of teaching&expertise well explained. I feel the same about Marcel's posts. I'llstick around to learn more, if you can put up with my beginner questions.
P.S: what could be the resonance about for PSRR case? I mean, resonance for Z_in is still rather kind of intuitive and ubiquitous ... sort of.
in what sense?
I guess it depends what are the parameters which are impacted.
For example, if H2 is proportional to Z_in, then probably yes it is a useful tool to be able to control the level, the frequency and the bandwidth over which this H2 increase is made.
Please detail or send me yearch keywords hints where I can find relevant literature links.
beginner here : I vaguely remember that I have seen just the other evening a crazy stuff in my search results, an OpAmp active filter which simulated an inductor using only capacitors 
Maybe it is the inverse function of C, obtained by current driven stages circuits. I do not know, just fooling around.
Maybe it is the inverse function of C, obtained by current driven stages circuits. I do not know, just fooling around.
Educated guesswork: I think it is no resonance, but rather a cancellation between two paths with approximately opposite phase (complex zero pair rather than a complex pole pair).
I remember that I have seen just the other evening an OpAmp active filter which simulated an inductor using a capacitor
When you make a bootstrap circuit with an ideal voltage follower, resistors with values R1 and R2 and a bootstrapping capacitor with capacitance Cb, you emulate an inductance L = R1 R2 Cb with a series resistance R1 + R2.
Apparently innocent.
But, if the remarkable similarity with a complex resonance is not followed up by Z_in being zero or infinite, then probably it is indeed not a resonance.
As usual, Marcel answered most of the questions.
Regarding the PSRR, here is the plot again:
About the (beneficial) effect of the increase in impedance at one or some frequencies, I'd say at first that you cannot have too much of a good thing -except that you can-. In some instances, phaseshifts or negative impedance due to additional parasitic effects can cause instabilities.
Regarding the PSRR, here is the plot again:
About the (beneficial) effect of the increase in impedance at one or some frequencies, I'd say at first that you cannot have too much of a good thing -except that you can-. In some instances, phaseshifts or negative impedance due to additional parasitic effects can cause instabilities.
Me, being in my hobby audio world a fanatic of zero-phase circuits, I challenge to know more details and insights how resonances can be beneficial in an amplifier
As a result, when you connect a coupling capacitor to such a circuit without any extra damping measures, there is always an impedance minimum R1 + R2 occuring somewhere, normally at some subsonic frequency.
You can get rid of that by adding a resistor in parallel with the bootstrapping network (or in series with the bootstrapping capacitor) and using an input coupling capacitor with much smaller capacitance than the bootstrapping capacitor. The disadvantage is a smaller input impedance at higher frequencies, but it can still be fairly high.
I do not know if this is 101 or high level electronics, but definitely this looks to me as high level. Needs time to digest it to understand its meaning.
with "physics DLL" it outputs a contradiction: smaller input capacitance should give a bigger (relative) input impedance at any frequency.
Maybe it is the late hour and the remains of a difficult day for me. I'll look again on it tomorrow.
looks like I need an "electronics DLL" to understand this.
with "physics DLL" it outputs a contradiction: smaller input capacitance should give a bigger (relative) input impedance at any frequency.
Maybe it is the late hour and the remains of a difficult day for me. I'll look again on it tomorrow.
You can: just scale all the caps accordingly:
Of course, it will also exacerbate the pseudo-resonance effects.
That said, for a supply having the maximum rejection on 100Hz (or 120Hz) is advantageous: there are very few half-wave supplies around
That's right, my PS has almost completely removed the 100Hz (so I don;t need actually any additional concern on PSRR for this frequency)
The remains of 50Hz fundamental + odd harmonics is killing though.
Thanks for values!
PS, you said splitting R1 and shunting from middle to Ground with a cap will help the PSRR. So, 2x 2.2K with shunt of 220uF would be fine?
The remains of 50Hz fundamental + odd harmonics is killing though.
Thanks for values!
PS, you said splitting R1 and shunting from middle to Ground with a cap will help the PSRR. So, 2x 2.2K with shunt of 220uF would be fine?
Yes, it will.
In my example, I used more extreme values, but that's probably overkill:
https://www.diyaudio.com/community/threads/one-bjt-line-amp-buffer.409364/post-7613009
Another option, for a very high PSRR is to use a two-pole filter. Overkill too, IMHO.
At some point, the noise caused by the regulator will drown the hum
In my example, I used more extreme values, but that's probably overkill:
https://www.diyaudio.com/community/threads/one-bjt-line-amp-buffer.409364/post-7613009
Another option, for a very high PSRR is to use a two-pole filter. Overkill too, IMHO.
At some point, the noise caused by the regulator will drown the hum
I have tested the bootstrapped version, and it behaves exactly like in the sim: the input impedance is raised to several megohms, with a huge increase at ~1kHz (probably not 300 meg, but very high).
The circuit remains well-behaved in all conditions, including open-input, and the frequency response and distortion remain unaffected.
The circuit remains well-behaved in all conditions, including open-input, and the frequency response and distortion remain unaffected.
I started to clear the PCB and count the nodes, number of holes per node, etc.
Maybe I get a chance to put as much features as possible.
But, an input node is giving me troubles. From the same one I need to start two circuits, one for LineOut and one for VU-meters.
Problem is, that VU-amp leaks back. It is a CE topology, inspired/copied from A77. There, its position right in the heart of the RecPreamp+EQ (feeding signals directly in the feedback of preamp) seems that it did not pose any real worries to the designers, or any concerns to performance. However, in my tests I see huge leaking back, making the node 'dirty' (full of harmonics). I am puzzled.
I wonder, how is proper to measure the isolation in Spice simulations?
Maybe I get a chance to put as much features as possible.
But, an input node is giving me troubles. From the same one I need to start two circuits, one for LineOut and one for VU-meters.
Problem is, that VU-amp leaks back. It is a CE topology, inspired/copied from A77. There, its position right in the heart of the RecPreamp+EQ (feeding signals directly in the feedback of preamp) seems that it did not pose any real worries to the designers, or any concerns to performance. However, in my tests I see huge leaking back, making the node 'dirty' (full of harmonics). I am puzzled.
I wonder, how is proper to measure the isolation in Spice simulations?
Last edited: