JFET pre question

In regards to the Tape-Op Hamptone JFET module:

An ideal JFET has infinite input impedance, right? What is the point of hanging a 10M resistor on the gate to ground? Why would you want to lower the input impedance? Thanks,

You need that resistor there to get rid of the leakage currents, and to allow the JFET's gate to bias at some voltage around 0 volts.

> Why would you want to lower the input impedance?

In most real-world audio, 10 Megs is perfectly infinite.

A yard (or meter) of cable has about 100pFd of capacitance. The reactance of 100pFd at 20KHz is around 100K ohms. So any source designed to drive even a short cable with full treble has to be designed for a source impedance lower than 100K, or at least much-less than 10Meg.

The main place we need higher impedance: a condenser microphone capsule acts like a 30pFd source. To get response down to 20Hz, we have to keep the load above 250Megs or so. We also have to keep the load to less than a few pFd, so it has to be very close to the capsule (typically inside the body). For complex reasons, tube condenser mikes use 100Meg-1,000Meg (1Gig), small-gate FETs can use higher values of 1Gig and up.

But the Hamptone is clearly intended for use from Line Outs, guitars, transformers, and pots, all much-much less than 10Meg sources.

And as Dale says: you need to set the DC bias on the Gate some how. FET leakage is low but not zero; anyway it is likely to leak and float to halfway between Drain and Source, not to a DC ground lower than the Source.

David;

Think about it this way the 10Meg pulles the gate down to
zero volts i.e. ground.
The source lead with resistor to ground develops
a few volts of DC drop across the resistor.
So the gate is more negative than the source and the
FET biases correctly. Remember the solid state device
has three legs and you bias each leg relative to the other.

Hope that helps.

Thanks guys, you've explained it brilliantly. I was guessing it was probably analogous to the grid leak resistor on a triode. Cheers!

> I was guessing it was probably analogous to the grid leak resistor on a triode.

Exactly the same. But there are so many children of the 1980s around, vacuum-tube analogies no longer work as well as they used to.

And like a grid-leak, as long as it is "much bigger" than source impedance, it has little effect on the circuit.

Different, because the leakage effects in most FETs are far smaller than most hollow-state devices. Typical small tubes should not use grid resistors higher than 1 or 2 Megs unless the plate has a large resistance: if leakage in a high-Meg resistor makes the grid positive, makes the tube hot, increases leakage, it can run-away and melt the tube. FETs have low leakage so can use a bigger resistor.

sometimes words like biasing can be intimidating and confusing. I'll put it in plainspeak.. really all you are doing is using resistors to pull-up or pull-down voltages on FETs and BJTs. putting a pulldown on the gate of the FET is pulling it's potential to ground. this is useful on an N-channel FET, both JFET and MOSFET topologies. for the Nchannel the gate is OFF when at the same potential as the source, turning ON at a specific potential above the source. N channels are usually used "lowside" which means that they usually switch the ground side of a device. this is why the Gate is usually pulled down on a N-ch FET, to keep the gate OFF. The gate is capacitive and can develop and hold a charge unless this is done.

One more note on this.

I agree with the others about why that resistor is there. And I agree with PRR that 10meg is essentially infinite impedance for the application.

The disadvantage to a big R to ground at the input is increased noise. I am not familiar with that preamp, but 10meg sounds really high. 1/10th the value would significantly lower the thermal noise contributed by this resistor. And it would still be high enough to not worry about loading effects on almost all sources including passive guitar pickups.

> The disadvantage to a big R to ground at the input is increased noise. I am not familiar with that preamp, but 10meg sounds really high. 1/10th the value would significantly lower the thermal noise contributed by this resistor.

The thermal noise of that resistor is shunted by the source impedance. Yes, open-input it is about 44 microvolts across the audio band. But connect it to a 10K source (or the center position of a 40K pot with low-Z at top and ground at bottom) and the 10Meg resistor's noise is 1,000 times lower (10M/10K). A 1Meg resistor has around 15uV audio noise, which is lower, but is only shunted 100 times by the 10K source, so it actually is noisier than the 10Meg resistor. (In fact noise is now dominated by the 10K source, about 1.4uV across the audio band.) Knowing that audio source impedances are rarely over 10K (except inside compact high-gain amplifiers), we see that external inputs or pot-sourced amps rarely need grid/gate resistors much over 100K-1Meg, but a higher resistor actually means (immeasurably) lower total noise.

Yes, guitar-pickups, phono-pickups, and hi-Z mike transformers are about the highest impedance sources we usually meet. But even phono-pickups, designed for just 4 feet of cable-loading, peak at around 20K-100K impedance at the top of the band (hence the standard 47K load). Guitar pickups are larger, typically drive more cable, and don't really want 15KHz bandwidth: 1,000pFd at 6KHz is 30K impedance; at 800Hz it might be 240K. Guitar pickups may tend to resonate somewhere in there, and we may not want to damp too much, but the many-K copper resistance also damps the coil resonance and I doubt impedance ever gets much over 100K, hence the popular 470K input resistor.

I have ignored current noise. (You did say thermal noise.) In a typical thermionic device, noise
current is low but not negligible in resistors higher than a Meg or so. Again this is reduced by practical source impedances so it hardly matters. (Anyway typical tubes can become thermally unstable with big grid resistors combined with small plate resistance and high supply voltage.) Most FETs have much lower current noise and it can usually be ignored. The one big audio exception is the condenser microphone capsule: a 30pFd source that is too weak to be shunted. We have to do things in condenser mikes that we would not do in any other audio stage.