Effective OP-AMP Input Impedance, JFET vs. BiPolar

[matt239]

I'm curious about the effective input impedance of various op-amps, in various circuits.

Consider, as an example a TL072 vs. an LM833, -the LM833 has better specs overall. -The selling point for the TL072 (in a guitar stompbox..) is the bi-FET input's ultra-high input impedance,
-but, configured as a non-inverting amplifier, won't the LM833 have very high input impedance also?

For an ideal non-inverting op-amp the input impedance is infinite, so for a real LM833 it should still be at least a couple MegaOhms yes?

Is an LM833 a good choice for a guitar level input?
- such as in a buffer-less screamer type circuit?
Or, will it require a buffer before?

[phatt]

Hi Matt,

A TS9 has a single BJT buffer and the input imp of that one transistor buffer is likely to be pathetic when compared to the opamp input Z.
So why bother with the stupid BJT buffer?

Well that reduces the Z to the opamp input. (notice the 10k hanging on the input of the opamp?)
Do the test yourself?
Drop the buffer,,, hang a 1meg R off the opamp input now insert a 1 meg pot around the opamp (Gain). Now turn up the gain and listen to the **NOISE**
Now go back to the pathetic little BJT buffer circuit and you will find there is far less Noise in your circuit.

Took me quite a long time to get my head around designing out noise issues as any one can build massive gain from a single opamp but more often than not the noise is horrendous when full up.

The quality of the opamp pales when compared to good circuit design, even a well designed 741 chip can be stunning when you understand how it works.

Very high speced chips may well be a problem for distortion units and basic guitar circuits.
Phil.

[mictester]

Hi Matt,

A TS9 has a single BJT buffer and the input imp of that one transistor buffer is likely to be pathetic when compared to the opamp input Z.
So why bother with the stupid BJT buffer?

Well that reduces the Z to the opamp input. (notice the 10k hanging on the input of the opamp?)
Do the test yourself?
Drop the buffer,,, hang a 1meg R off the opamp input now insert a 1 meg pot around the opamp (Gain). Now turn up the gain and listen to the **NOISE**
Now go back to the pathetic little BJT buffer circuit and you will find there is far less Noise in your circuit.

Took me quite a long time to get my head around designing out noise issues as any one can build massive gain from a single opamp but more often than not the noise is horrendous when full up.

The quality of the opamp pales when compared to good circuit design, even a well designed 741 chip can be stunning when you understand how it works.

Very high speced chips may well be a problem for distortion units and basic guitar circuits.
Phil.

Spot on, Phil.

There are a couple of other issues, too. The transistor buffer does a good job of separating the op-amp (or the CMOS IC you're pushing into service as a "tube-sound" op-amp) from the outside world. That isolation can be the difference between reliability and a bad reputation for spontaneous failures!

[matt239]

Hmm.. Thanks guys.
I understand what you're saying, but not completely.. (I figured there must be a reason for the buffer, so that's part of what I was trying to understand..)

Why is the noise different? what causes that?
What about op-amps as buffers? does that work as well as a transistor? (obviously is a more expensive part, but a little simpler, & sometimes you have 1/2 a dual left-over..)
Would there be some reason to choose a TL072? Or is the main point to use a buffer with either?

Perhaps the best case would be a JFET buffer??

- I'm not too obsessed about op-amp specs, or hung up on the "mojo" of different op-amps,
I take your point that overall circuit design is probably more important.

My thinking was mostly that the LM833 has low noise, & a reasonable price, so using it, especially early in the circuit, might help in designing a low noise device??

I'm interested in any methods you can suggest for reducing all types of noise.
Even though we're designing pretty simple, not-too-critical devices, why tolerate excess noise?

P.S. @ Mic; What failure modes does the buffer protect the I.C. from? - Will the transistor then fail instead??

[DrNomis]

As posted in a similar forum thread, typically, the input impedance of an Op-Amp IC is about 10 Megaohms, well it's high enough to be regarded as a "Virtual-Earth", most Op-Amps these days have very low input current specs which can be as low as 1nA (one nano-amp) in the case of Op-Amps with JFET inputs, because of the very high input impedances some Op-Amps can be sensitive to ESD......

[mictester]

P.S. @ Mic; What failure modes does the buffer protect the I.C. from? - Will the transistor then fail instead??

The inputs to most op-amps are very sensitive (particularly the CMOS or FET input types), and spikes (from static build-up, for example) can destroy the input of these devices. Bipolar transistors are much more tolerant, and will provide good protection. They can also be used to provide impedance conversion, too.

[matt239]

Will discrete FETs be sensitive to ESD, once in circuit?

How to use BJT for impedance conversion?

I'm thinking to keep input impedance @ at least 1M.
-but also low noise..

The first stage can have a low gain; maybe 5. (Yes, the same very-low-gain "tube sound" O.D. I'm always talking about..)
I'm planning on not allowing the Op-amps to clip. -- Diode "ladders," series resistance, germanium diodes.
Cool thread here: viewtopic.php?f=28&t=12212

[phatt]

Will discrete FETs be sensitive to ESD, once in circuit?

I'll let better qualified Folk Answer that one Calling mictester

How to use BJT for impedance conversion?

TS9 frontend

I'm thinking to keep input impedance @ at least 1M.
-but also low noise..

TS9 again or use an opamp buffer front end if that 1 meg thing bothers you.

with multi stage Higain rigs the 1 meg input thing is just going to add to noise problems.
If using an opamp buffer stage then Even 220k input R works fine and keeps things under control.

The 1meg thing is just a left over from old Valve teck.
Bare in mind they used the bare minimum Valve count and the input stages where always high to get a decent signal swing but with most gear today being multi stage setups,, be they SS or bottles you are bound to end up with noise issues if you just keep hanging 1meg resistors of everything thinking you are adding some mojo.
sorry but that's just a fantasy trip for the ignorant masses

I was researching the Carvin Legacy Valve Amp a while back and noticed the input grid leak was down to a more SANE 220k on V1 and the PI input was also reduced to 100k rather than the ignorant 1meg.
With so much gain you have no choice but to reduce sensitivity somewhere otherwise the noise is horrendous.

The first stage can have a low gain; maybe 5. (Yes, the same very-low-gain "tube sound" O.D. I'm always talking about..)
I'm planning on not allowing the Op-amps to clip. -- Diode "ladders," series resistance, germanium diodes.
Cool thread here: viewtopic.php?f=28&t=12212

Just stick a buffer in front of that first circuit posted by *billonious* and reduce R5 down to 10k,, then you may have to increase P1 up ,,say 100k should deliver.
For me I've moved on from those all in one pedal tricks. Some can sound ok but they tend to lack in the sustain department.
Hope it helps,, Phil.

[DrNomis]

Will discrete FETs be sensitive to ESD, once in circuit?

How to use BJT for impedance conversion?

I'm thinking to keep input impedance @ at least 1M.
-but also low noise..

The first stage can have a low gain; maybe 5. (Yes, the same very-low-gain "tube sound" O.D. I'm always talking about..)
I'm planning on not allowing the Op-amps to clip. -- Diode "ladders," series resistance, germanium diodes.
Cool thread here: viewtopic.php?f=28&t=12212

Once a discrete Fet is soldered into circuit, the device will be relatively safe from ESD damage because static charges will not be able to build up to dangerous voltages because they will have an electrical path to circuit ground, via the high-value pull-down resistor from the Fet's Gate terminal to circuit ground.....

[juanro]

Some reading about noise:

http://focus.ti.com.cn/cn/lit/ml/sloa082/sloa082.pdf

Juanro

[matt239]

1M input Z does still seem to be a popular design goal for guitar inputs..
The idea is to avoid loss of treble in combination with cable capacitance. You would only do this @ the input, not in subsequent stages..
I'm not married to >1M input Z, but it does seem important not to let it get too low.. Seems like it should be >500k ??
Am I missing something?

JFET buffer is starting to sound like a good idea..

(If ESD is not a problem for the FET transistor once it's in circuit, because of path to ground, does this not apply also to op-amps??)

[mictester]

1M input Z does still seem to be a popular design goal for guitar inputs..
The idea is to avoid loss of treble in combination with cable capacitance. You would only do this @ the input, not in subsequent stages..
I'm not married to >1M input Z, but it does seem important not to let it get too low.. Seems like it should be >500k ??
Am I missing something?

JFET buffer is starting to sound like a good idea..

The losses inherent with high impedance pickups, feeding passive volume and tone controls, in most guitars can be mitigated slightly by having high input impedance to your effects or amplifiers . However, this really isn't a great solution. Active pre-amplification inside the guitar giving a balanced, low impedance output is the way to go! Low impedance will drive cables with little loss, and a balanced audio feed will prevent the pick up of hum and other noise on the cable.

(If ESD is not a problem for the FET transistor once it's in circuit, because of path to ground, does this not apply also to op-amps??)

To a large extent, yes. However, both FETs and op-amps are susceptible to input spikes. Bipolar transistors are much more robust in this respect. This is why the Tubescreamer separates the FET switch from the input socket by a transistor buffer stage.

[merlinb]

you are bound to end up with noise issues if you just keep hanging 1meg resistors of everything thinking you are adding some mojo.

Not true. Basic noise theory tells us that shunt resistors need to be as *large* as possible for maximum signal-noise ratio (and series resistors should be as small as possible). High *node* impedances are susceptible to hum and other capacitive interference, but in any normal setup the source resistance will shunt the 1Meg resistance, thereby reducing the node impedance.

A 220k input impedance will cause noticeable damping of pickup resonance and treble, especially with humbuckers. 400k is pretty much the bare minimum you want to aim for.

[DrNomis]

There's also something called the Maximum Power Transfer Theorem, basically, it means that if the output impedance of a signal source and the input impedance of an input are the same, you will get maximum power transfer between them, if the impedances don't match, the power transfer will be less than maximum, this happens because the input and output impedances form a divider......


Here's a link to an article on Wikipedia about the Maximum Power Transfer Theorem, and goes into it in a bit more detail:


http://en.wikipedia.org/wiki/Maximum_po ... er_theorem

[juanro]

I think maximum power transfer is relevant when speaking of, right, power transfer, as in the case of the output stage of a tube amp. But in the input stage just voltage is of importance, and noise, and resonance or highs damping. If one could use active guitars with balanced output it would be ideal; but for generic or standar stuff, one should aim for the higher input impedance possible without becoming too noisy (unless low input impedance is part of the effect itself as in a fuzz face)

Regards
Juanro

[matt239]

Hmm. This is getting good..

Can we reach a group consensus?

1: Do we need a buffer? (Or can we rely on high op-amp input impedance, & trust the op-amp won't be damaged in-circuit??)
2: A FET buffer or BJT?
3: Is 1M input impedance appropriate? ( ... /necessary/overkill ??)

Thanks for all the great replies!

[phatt]

Hmm. This is getting good..

Can we reach a group consensus?

1: Do we need a buffer? (Or can we rely on high op-amp input impedance, & trust the op-amp won't be damaged in-circuit??)
2: A FET buffer or BJT?
3: Is 1M input impedance appropriate? ( ... /necessary/overkill ??)

Thanks for all the great replies!

I doubt you will ever reach **group consensus**

Here's what my circuits sound like,,,, your choice?
The buffer input is only 390k and the OD when engaged is only 68k,,,, go figure???/
I can post the circuits if you wish/
Cheers,,, Phil.

[phatt]

Hi Matt,,, sorry forgot to mention.
Re damage to TL072/LM883.
I've built uncountable circuits of all sorts ,,pulled opamps while voltage present,, you name it I've probably done it.

These are very robust devices and unless you are *really cleverly stupid* you will have a hard time blowing them up when used as intended.
It's just not an issue. Go for it and have some fun and learn along the way
Phil.

[DrNomis]

Hmm. This is getting good..

Can we reach a group consensus?

1: Do we need a buffer? (Or can we rely on high op-amp input impedance, & trust the op-amp won't be damaged in-circuit??)
2: A FET buffer or BJT?
3: Is 1M input impedance appropriate? ( ... /necessary/overkill ??)

Thanks for all the great replies!

Well, I reckon you probably will need a buffer of some sort to provide a high enough input impedance for a guitar, the output of a guitar is usually at a high impedance (for passive pickups), a buffer will make the effect sound brighter, personally I think it's good design practice to design for a high input impedance, and a low output impedance anyway....

I think you could make the buffer using a JFet or an OpAmp, it depends on your design budget, either will work equally well...

And I think any impedance from 1M to say 4M7 should work fine.....

[merlinb]

1: Do we need a buffer? (Or can we rely on high op-amp input impedance, & trust the op-amp won't be damaged in-circuit??)

Opamp buffers are perfectly reliable. A small series input resistor (100R to 1k) is good practice.

2: A FET buffer or BJT?

A single BJT emitter follower cannot easily attain 1Meg input impedance in a pedal circuit. You really need a FET or opamp to do that. Discrete buffers usually have less than impressive headroom though, unless you designed them carefully (which requires more parts), in which case you might as well go with the opamp anyway. The BJT buffer in the tubescreamer barely manages 1Vp-p headroom, which is kinda lame.

3: Is 1M input impedance appropriate?

1Meg is ideal (nice round number that pickups were originally designed to work with), but any value from about 500k upwards will do.