bootstrapped grid resistance?

Reading this http://sound.westhost.com/dwopa.htm#highz

I thought of this:


would this do the same thing? I mean would it cause the upper grid resistance to appear higher than it actually is, because of the small voltage across it?
Vin - [Vin * 0.9] or whatever the gain is at the cathode....? Certainly it wouldn't work as well as an opamp with scads of open loop gain but what the heck.

Just thinking out loud here. Can't make any claims as to the quality though :grin:

Yes, that works. The linked circuit, as well as this one, actually makes the input impedance look like an inductor with a series resistor. When the gain is really close to unity like the opamp version, the inductance is equal to the product of the two resistors and the capacitor, and the series R is the sum of the two Rs. This assumes that the opamp input impedance is very high relative to those resistances.

> Vin - [Vin * 0.9] or whatever the gain is at the cathode....?

For simple happy cathode bias, gain at cathode is closer to 0.5 than to 0.9.

Oversimplified proof: 12AX7, book-conditions 100V -1Vg 0.5mA, Gm=1,250uMho. Bias resistor is 1V/0.5mA= 2K. Cathode impedance is 1/Gm or 800 ohms. Gain at cathode is 2K/(800+2K)= 2000/2800= 0.7. In real life, plate resistor divided by Mu adds to 1/Gm, say 100K/100= another 1K seen looking up into cathode, now more like 2K/(1800+2K)= 0.52. The 'AX7 is one of the best, 12AU7 at high current gives 0.64.

So you can maybe triple your "input impedance". But you can thoughtlessly use 1Meg. You can usually go higher. Ampeg liked 2M7 or 3M3 grid resistors. No danger of runaway because inputs tend to have large plate resistors and "can't" go over-current.

If that cathode is really the NFB injection point around a high-gain amplifier, then you may really have midband gain input-cathode near 0.9. But also a lot more phase shifts in the system.

However in Audio, the real input resistance is the capacitance. Miller on a 12AX7 gives 100pFd input C, and that's like 80K at the top of the audio band. i.e. that "1 Meg" is true only to 1,600Hz. Jack the resistance to 3Meg, and you still have the 100pFd, 80K at 20KHz, impedance falling from 533Hz.

As a general rule in audio: if you think you need an input impedance over 100K, you are probably wrong.

If input Z is your over-riding goal, do a cathode follower with grid suspended from two 22Meg resistors B+ to ground. The low-frew impedance is near 11Meg. As CF, the tube's capacitance becomes just a few pFd. You will have to go nutz with low-C layout to avoid a big increase due to wiring parasitic C. And what about your cable?

If your transducer is capacitive, then it may be valid to have a high impedance at low F and a low impedance at high F. But as bcarso says, bootstrapping with a capacitor gives an inductive input. Drive it from a capacitive source, and you may (or may not) wind up with a big resonance at a troublesome frequency.

For DC-mostly measurements, standard VTVMs used 10Meg grid paths. Because of the nature of the beast, grid leakage offset was always "zero": the user had a knob for that and a big needle-scale to read it on. The better breed had twin-triode and balanced grids for first-order correction. You still get 300mV offset for tube-swap, 150mV cold/hot, 20mV day-day always-hot. But these drifts are sub-audio and can be ignored.

Bootstrapped tubes are fairly rare. Even in condenser mikes and crystal phono inputs, where it might seem useful. Transistors did it more, but that fell out of fashion when per-each price fell to a dollar and another stage allowed ample impedance leverage at low cost.

G9 input stage is bootstrapped (R8) for very-high input impedance..

http://www.gyraf.dk/gy_pd/g9/g9_sch.gif

Jakob E.

[quote author="PRR"]
However in Audio, the real input resistance is the capacitance. Miller on a 12AX7 gives 100pFd input C, and that's like 80K at the top of the audio band. i.e. that "1 Meg" is true only to 1,600Hz. Jack the resistance to 3Meg, and you still have the 100pFd, 80K at 20KHz, impedance falling from 533Hz.
[/quote]

12AX7 is a twin triode, so a single toob may be used as a cascode.

The problem solved. :wink:

Thanks all. My current project is actually a guitar amp, but I was just generally curious if I was thinking about this bootstrapping thing right. For sure the transducer is reactive, it's a guitar pickup. An old gibson mini-humbucker, "Johnny Smith". With a volume pot. Understanding what this signal really looks like, in terms of Z vs. frequency has been a kind of stumbling block.

> it's a guitar pickup. ... With a volume pot.

5K+5H, loaded in ~~200pFd and 250K(?) pot, the pot wiper loaded with ~~300pFd of guitar cord, gridleak, grid capacitance.

I did the curves long ago, and if the guitar pot is 250K and anywhere in the middle, it's basically 60K resistive and the cord-capacitance does not influence the pickup winding resonance. So any amp-jack over 270K and under 150pFd is all the same. If you work the pot full-up, you might like 750K-1Meg and <150pFd.

Without the pot, there is a sub-octave resonance peak in the 3KHz-6KHz range, sensitive to cable length. The amp input resistance affects the height of the peak, but high resistance gives a very narrow peak, like 3 notes (actually harmonics).

A very low resistance will suck-off highs due to coil inductance. There is ample empirical evidence that 130K is "high enough" for most guitar tone. Look at the Fender 2-jack input affair which became common when humbuckers appeared and were overloading inputs. The lo-gain jack gives two 68K in series to ground with grid tapped at the halfway point.

Keep your resistance well above 100K, it's all the same. For non-active guitars, capacitance is all about the guitar cord. Don't wet-dream "high impedance" and then use a 30-foot cord. Even a 3-foot cord is 90pFd, and fairly awkward unless you sit with your amp at your right hip.

Jack's site has an overview.

Douglas Brown's paper at U Urbana has carefully measured values. Note however that cable capacitance is not included (yet generally essential); real stage use does not yield 9KHz and 850K.

And you do NOT want 9KHz response. The overtones of a string (or pipe) are not on the harmonic series. You "need" a low-pass to knock-down inharmonicity. The speaker does some, but high partials in the amplifier will intermodulate into the passband, with really unharmonic result. The pickup top-resonance is as important as the diameter of a flute or clarinet, or the thickness of a reed, factors which blow-musicans wrestle with.

More at Daten & Resonanzfrequenzen von Pickups, also BuildYourGuitar.com

------------
> a single toob may be used as a cascode. The problem solved.

Analysis suggests that input C of the tube is not limiting. However guitar input stages tend to run near overload even with 250V supplies; a cascode suggests ~~500V supply between stacked triodes and increased gain. Also I am believing that the curvature of 12AX7 is part of "the sound"; a cascode has a different curvature. And in a real-world, every dollar and ounce in a guitar-amp hurts your income and your shoulder. Solving a non-problem is not the best use of a spare section.

[quote author="PRR"]
------------
> a single toob may be used as a cascode. The problem solved.

Analysis suggests that input C of the tube is not limiting. However guitar input stages tend to run near overload even with 250V supplies; a cascode suggests ~~500V supply between stacked triodes and increased gain. Also I am believing that the curvature of 12AX7 is part of "the sound"; a cascode has a different curvature. And in a real-world, every dollar and ounce in a guitar-amp hurts your income and your shoulder. Solving a non-problem is not the best use of a spare section.[/quote]

Sure. Also, I personally don't like the sound of cascodes, but "If Zulu customer wants beads, ..." :green:

[quote author="RogerFoote"][quote author="Wavebourn"]"If Zulu customer wants beads, ..." :green:[/quote]

Hmmm, never heard that one before...[/quote]

Do you mean bootstrapped input, or cascode?

Thanks for the links PRR. Maybe I won't worry about the input z. But....

I'm not trying to clone a Fender, I'm more going for that "acoustic amp" type thing. Alot of these have tweeters and claim full range response. Active tone controls are pretty standard. Many have class-d power amps. AER is a company making good ones. I'm aiming somewhere in between; I'm not using a piezo, and I don't think I need baxandall tone circuits. But neither do I need distortion, master volume control, the intentionally saggy power supply, etc. I'm not particularly married to the idea of having a tube preamp actually. I'm one of those philistines who doesn't care much for the all-tube amps. I hear distortion, exaggerated boing-y dynamics, plus I don't want to carry it. For a preamp it might be nice, as long as it's in a fairly linear region.

Maybe you can see why I ask questions here instead of at guitar forums (fora?); my wants don't align with 99% of other guitarists. I hope no one feels I'm hijacking here, I know this is supposed to be for studio pro audio stuff.

My instrument is basically an acoustic fitted with a pickup. So, if there's some inharmonic effect that will come through an amp, but that I can't hear when it's not plugged in (20 hz - 20 khz, or with my ears probly 30-13k), I definitely don't want that. One simple experiment I have done is to record my guitar direct, just to see what adjustments I would have to make to get an acceptable tone, starting from a supposed flat response. Loaded by 102 Kohms at the "universal input" of my ADC, it's actually pretty close; a few dB mid cut here, a little hi shelf there. Maybe there's some other reason this wouldn't translate well to an actual speaker. But based on this experiment I was imagining a fairly flexible mid cut (freq, depth, 2 position q switch?), a simple hi shelf cut, and dispensing with the bass control entirely. It'll just do one thing, with slight adjustments for different rooms. And have a really really cool power indicator - a need shared by all guitarists!

Hi Jeffrey;
I made a thingy for acoustic guitars with pickups recently:

http://www.groupdiy.com/index.php?topic=26133&highlight=vysotsky

Thanks Wavebourn, interesting. I'm still going through PRR's links too. I slow.

Nice crest. Is that a hedgehog?

> that "acoustic amp" type thing. Alot of these have tweeters

OK, you want all the string-ding and squeak that a pure-acoustic guitarist uses for accent where an electric gitarist would just ease into overdrive. Valid.

> I'm not using a piezo, ... basically an acoustic fitted with a pickup.

What KIND of pickup?

The Rickenbacker/Gibson/Fender many-turn wound pickup will not catch nor deliver very high frequency; to get much past 4KHz you need to decouple the cable capacitance from the winding (active buffer in/at guitar). Even then the self-resonance is ~~9KHz. The impedance may be ~~800K, a 1Meg or 2Meg input is plenty high. With a few feet of cable, 102K is probably "high enough".

Piezos don't care about cable capacitance (it reduces overall level, not treble extent). Their high mechanical impedance means they must couple through bridge or other woodwork, which may shave the top octave. (OTOH, their high mechanical impedance and the lack of anything "solid" in an acoustic instrument means bass response is often saggy...)

What KIND of pickup?

Click to expand...

Right, it's a Johnny Smith - the form factor is particular to archtops, where it mounts by a slim bracket to the end of the neck/fingerboard, so as to avoid routing openings in a very nice piece of carved spruce. Otherwise I believe it's similar to other humbucking pickups. Wound to about 7.2K. I don't know about the inductance, but I think I know someone who does.

you want all the string-ding and squeak

Click to expand...

Well let's say, I want some of it. I'm sure I'm pretty limited without actually using a piezo or a microphone. I'm not against the use of a buffer but it makes things rather more complex. I think I'm just going to start building things and see what happens. I'm ordering a Hypex class-d amp and making a box.

> the form factor is particular to archtops

OK, but the guts is a plain old telephone coil or Majestic speaker-coil adapted to a 6/12-pole situation. Both you and a site I hunted-for say it is many-K impedance (probably DC resistance), well within the general ballpark. There seem to be good and mediocre models, but I bet electrically you treat them all the same.

For extended top end, keep your cable short. Rock whangers can stand 30-foot cords and 3KHz cut-off, because they can add artificial zing in the amp overload. 10 feet is long for full-range well-fingered guitar and this class of pickups.

> Well let's say, I want some of it.

I trust you to use taste and discretion. The general aim on the pickup-cable-amp interface would be to get as much as reasonably possible (compromised with cable length and buffer-battery on the guitar), then shape it both with finger technique and amp/speaker tone.

Ordinary tubes and FETs can easily stand 1Meg-2Meg grid resistors, and this is well above any impedance the pickup/cable combination will reach. Your main variable is cable length.

Well, to my surprise Kent Armstrong didn't have much to say about guitar pickups. He rewound mine and is fairly well known for producing great pickups, but I gather he's not a theorist, he just knows how to do it, by doing it many thousands of times. He is the son of Dan, of lucite guitar fame. A nice friendly guy too.

I had forgotten what a hassle that Miller capacitance is. Need more practice. The falling impedance of the input combined with rising impedance of the pickup doesn't seem conducive to a flat response. The capacitance of the cable and the Miller are effectively parallel to each other?