bootstrapped opamp buffer

[Gus]

I circuit I have not seen in a guitar/bass effect

Lower value input resistors that are bootstrapped to a higher value input resistance.  Note how C2 value changes things.  The opamp in the screenshot is in the stock download of LT spice.

[John Lyons]

Interesting. Is the input cap really 470pf? 

[PRR]

> Is the input cap really 470pf?

I can't read it.

[arawn]

.00047uf yep thats 470 pf 

[Gus]

Link to sim
http://www.aronnelson.com/gallery/main.php?g2_view=core.DownloadItem&g2_itemId=46187&g2_serialNumber=1

Yes 470pf.  Note the use of lower input resistor values and the high input resistance.  This might help with noise.  I should build this with something like a 741 as a test.  Need to think about stability with different opamps.  Use the opamp in the sim because it was in the stock LT spice down load.  

9VDC with 10 ohm output resistance

[earthtonesaudio]

Might want to check the effect of varying R6.  Taking the signal for the bootstrap cap from the junction of R5/C5 will make it more sensitive to output loading than if (for example) the cap came directly from the output of the op-amp, or through a separate resistor.

[Gus]

The sim gets ugly if you take the feedback from the output of the opamp before the resistor.  Yes you need to be careful with the loading but IMO it is less ugly this way.  R6 below 100K loading you lose some lows above 100K to 1meg the peak moves in frequency with the same cap value and the bottom peaks a little more.

Not a circuit for everyone but if you can put it in a sim and adjust it for you needs it might be something interesting.  Sensitive to values used for resistors and caps.

This might be useful as a high input resistance buffer with some bottom peaking at a set frequency that goes into a known load it can be adjusted for you get a EQ(bottom bump) and high resistance in one stage.

EDIT earthtoneaudio good idea with the separate resistor (R12)  Sim is less sensitive to loading with R12

[R.G.]

A similar and very interesting problem is to raise the input impedance of an *inverting* opamp input.

Doing the simple thing and just raising the input resistor leads to substantial noise increases and also makes the feedback resistors become very large indeed. For instance, a 1M input resistor on a gain-of-ten inverting opamp stage would require a 10M feedback resistor, and even if you can find one of those with a decent tolerance, PCB surface contamination and stray capacitances play havoc with what the resistor does in the circuit.

There is not a huge amount that can be done about the input resistors simply, but one can lower the values of the feedback resistor by going to a T attenuator for the feedback network to lower the resistive impedances, making strays much less important.

The issues of bootstrapping a noninverting opamp stage and also the inverting stage are covered in Walt Jung's "Audio IC Op-Amp Applications".

[Gus]

R.G. I have the Jung book "Audio IC Op-Amp Applications".  When I simmed a single supply circuit with the values in the book I got a big about 10 hertz peak.

[CynicalMan]

Try using ".ac oct 50 5 20k" for clearer peaks.

[Gus]

CynicalMan  thanks, 50 does help with the low end

Anyone going to build it?   If you do I would build the circuit with R12.

[PRR]

It is two 90 deg phase-shifts, a 180 deg inversion at some *infinitely* low frequency. It won't outright oscillate, but it can peak-up to nearly-infinity bump. (and your C1 is the start of a third phase-shift and potential oscillation.)

Offset C2 and C3 (as you did) a LOT to stagger the shifts. However the available LF gain of chips means the offset has to be insane.

You need R12 to limit the amount of bump.

Why bootstrap? Even a '741 can be run with huge bias resistors. The input bias current in shirtsleeve is <0.1uA typ 0.5uA max. Take the 0.5uA max and feed through a 2Meg resistor: 1V DC error. Which means your "4.5V" could be 3.5V or 5.5V. (In fact for '741 it will be down, 3.5V; and at 9V supply it won't have 0.5uA bias but something less.) This is still well between 0V and 9V rails. Signal will pass.

Yes, 2Meg is not as high as what you got (5Meg?) but "very large" compared to pickup and cable impedances.

[Gus]

PRR posted "Why bootstrap?"  good question.

One, something different
Two, I found by accident that I like what the NPN boost ver2 sounds like at min gain as well as other gains it has a peak in the lower frequencies at min gain so a one stage EQ buffer circuit might be fun to try
Three, by bootstrapping lower values for R1 and R2 like 100K.  I was thinking noise might be lowered.  I would use a JFET opamp if I built it

The last sim is from the Jung book "Audio IC Op-Amp Applications" using the resistor and cap values in the book 470Ks .0022uf and .47uf green trace converted to single supply(R3, R4 and C1).  R.G. posted about "Audio IC Op-Amp Applications" so I showed a circuit close to the one in the book with the 10Hertz peak

For the first sim red trace is a 2.2uf looks to be the flat buffer cap value.  The high end peak is from the guitar

[merlinb]

PRR posted "Why bootstrap?"  good question.

One, something different
Two, I found by accident that I like what the NPN boost ver2 sounds like at min gain as well as other gains it has a peak in the lower frequencies at min gain so a one stage EQ buffer circuit might be fun to try
Three, by bootstrapping lower values for R1 and R2 like 100K.  I was thinking noise might be lowered.  I would use a JFET opamp if I built it

I used this sort of bootstrapping in early versions of the Glass Blower, but I later removed it because there were situations in which it would motorboat. Decreasing shunt resistors like R1/R2 does not significantly lower noise (in any case, bootstrapping is positive feedback, so the noise just gets fed-back and reinforced anyway).

[johngreene]

I used this sort of bootstrapping in early versions of the Glass Blower, but I later removed it because there were situations in which it would motorboat. Decreasing shunt resistors like R1/R2 does not significantly lower noise (in any case, bootstrapping is positive feedback, so the noise just gets fed-back and reinforced anyway).

If it was motorboating then it must have had gain. Gus's circuit does not have gain. I don't see how noise is getting fed back? What I see is the output of the opamp being used to provide a reference for the input that follows it at a reduced level, therefore reducing the amount of current needed from the source, i.e. raising the input impedance. By reducing the amount of current flowing through the input resistors, the amount of noise generated by them is reduced. Now if the opamp was configured for gain, I can see noise being added by the amplification. But not in this case. If anything it is a noise reduction (noise from current through resistors anyway).

[merlinb]

I don't see how noise is getting fed back?

The opamp generates noise, and there is some noise from the source too. The opamp can't tell the difference between signal and noise, so it is fed back via the bootstrap capacitor and re-amplified just as much as the signal. Intuitively this is what you shoudl expect, since negative feedback tends to reduce noise, positive feedback increases it.

By reducing the amount of current flowing through the input resistors, the amount of noise generated by them is reduced.

No DC current flows in the input shunt resistor even with the standard configuration, so that's not an issue. Also, the excess noise of a carbon film is about 0.1uV/V, so even if you somehow had the full 9V across one, that's only 0.9uV, which is less than the Johnson noise of a 3k resistor.

[johngreene]

I don't see how noise is getting fed back?

The opamp generates noise, and there is some noise from the source too. The opamp can't tell the difference between signal and noise, so it is fed back via the bootstrap capacitor and re-amplified just as much as the signal. Intuitively this is what you shoudl expect, since negative feedback tends to reduce noise, positive feedback increases it.

By reducing the amount of current flowing through the input resistors, the amount of noise generated by them is reduced.

No DC current flows in the input shunt resistor even with the standard configuration, so that's not an issue. Also, the excess noise of a carbon film is about 0.1uV/V, so even if you somehow had the full 9V across one, that's only 0.9uV, which is less than the Johnson noise of a 3k resistor.

But the opamp output is fed back to its inverting input. So, by your description, with unity gain all the noise generated by the opamp should be cancelled out. I don't think the noise being generated is of any importance. Bootstrapping is used on microphone preamps all the time. If noise was an issue, I don't think that would be happening.

Anyway, I think Gus's point was to show how the frequency response can be shaped via bootstrapping, not about noise reduction. I think its great when people 'experiment' with techniques for the sake of experimenting. Many times this is how great effects are 'discovered'.

[noisette]

Yes, old, I know, but interesting.

Actually the feedback is into NON-inverting.