Effective OP-AMP Input Impedance, JFET vs. BiPolar

merlinb · Post by **merlinb** » 09 Sep 2011, 15:00

kleuck wrote: Try a LP Junior in a 5 meters cable straight in a simple vintage amp, and switch to the same cable, a Boss pedal or a buffer, and another 5 meters cable, if you get the same sound, ie the same presence peak, you're lucky, i never did.

I suspect that might have more to do with the crummy BJT input buffer on those pedals, which have an input resistance in the 300-400k range, which becomes noticeable in bypass mode. Also, the input of some amps are not favorably designed towards being driven by pedals- I'm thinking mainly of modern high-gain amps here.

kleuck · Post by **kleuck** » 09 Sep 2011, 15:09

Right, but my own tests has been made with a good homemade buffer (fet based, with the choice between 1M and 2 M input imp) and two amps that are vintage-inspired, and modded by myself, not the point but the two have now a regular 1M input impedance, though one is full tube (and therefore the input is a tube) when the second one has an AOP in the input (stock a TL072, now an OPA2134, but probably a OP275 when i really did compare) and they are both simple, not 12 triodes in the preamp or complicated voicing nets.
One is an EL84 PP where the other is a class A SE (6V6 at the time, 6V6, 6L6, 6P3S, 6P1P or EL84 now...thanks a lot to the Great Merlin

)

merlinb · Post by **merlinb** » 09 Sep 2011, 16:06

kleuck wrote:Right, but my own tests has been made with a good homemade buffer (fet based, with the choice between 1M and 2 M input imp)

When you say 'good', what kind of buffer circuit do you mean? The bootstrapped kind?

kleuck · Post by **kleuck** » 09 Sep 2011, 16:26

Buffer with high voltage capabilities (about 4 volts on source), tried a BF245C at first (not so great noise performance, too much idle current)

then this one :

Not sure if it's what you call "bootstrapped"

earthtonesaudio · Post by **earthtonesaudio** » 09 Sep 2011, 19:22

I'd certainly call that bootstrapped (the hand-drawn schematic).

kleuck · Post by **kleuck** » 09 Sep 2011, 20:09

Sorry, i'm french and can't always find the best way to express my exact ideas in English, nor do i always perfectly understand.
To (try to) say more precisely what i mean when i say that output buffers are not a good thing most of the time, i do not want to say that they kill the sound in the same way as a crappy or too long cable, taming the highs and lowering the presence raise of the PU, but they sound to me most of the time (not only with my little buffers) like there's no more peak, or a way too high one (frequency, say 7Khz or more)
It's pretty obvious with my guitars as i use 1M pots, mini-humbucker and P90, there’s a huge presence peak, and you can hear the peak moving with different cables or length.
Same with my buffer -actually it was not 1M and 2M, but 1M and 3M as the schematic reminded me (did not draw the exact thing)
I like to use 50 to 100K pots on the contrary, cause i think it's a good range to roughly emulate the PU/cable combination.

matt239 · Post by **matt239** » 11 Sep 2011, 01:54

phatt wrote:Matt Your the one asking ,,, Not me.

OK; Once again I was trying to be funny, - joking when I asked: "Have I got you all talked into this topology" - Making fun of myself a little for having switched from asking a question, to supporting/defending a certain approach. Hence the:

Indeed, based on all the responses to my question, & on my continuing reading, I have convinced myself, (& maybe some of you??) that this is A valid, solid, technically correct approach.
I'm not at all claiming it to be the only valid approach.
I planned to build a circuit this way, & I wanted to see if you guys thought there was some big reason NOT to build it this way.
I hasten to add it's not "MY" method. A similar approach has been used by many of our D.I.Y. community's "Respected Elder Statesmen"

Remember, the value of the pull-down resistor does not = the total load the guitar P/U sees. A 1M resistor doesn't give us an input Z of 1M when connected to a typical electric guitar.
And, as I have shown; a 100k pull-down resistor, paralleled with the controls on a Strat, actually presents a load of 55k to the Pickup, creating a significant low pass filter effect.

It strikes me that some of you who prefer the very low pull-down resistors are actually using this as a special effect, - a totally valid approach if you find that filtering musically useful/pleasing.

I also find that treble roll-off musically useful/pleasing much of the time, but I want the OPTION to have the twang and jangle also..
With my setup (Strat w/bridge humbucker) I can get an almost Tele-like brightness/twang, or I can roll back the guitar VOL control to "7," crank up the amp volume a little and, VOILA! : instant AC/DC!

matt239 · Post by **matt239** » 21 Sep 2011, 02:55

I have convinced myself,

mostly...

I'm still not married to this, if there's something I'm missing.
Just, nobody has explained why not, in a way that makes sense to me, & I feel like I answered some of the concerns.

Originally I was trying to decide between an LM833, & a TL072, & choose a pull-down resistor, etc.

I know we used to choose JFet input op-amps for higher source impedances, & they would give lower noise performance in that specific case than bi-polars with lower noise specs, but lower input resistance, but I'm under the impression some newer bi-polar input op-amps will perform better..
- See R.G.'s posts here: https://www.diystompboxes.com/smfforum/ ... c=77130.20

I think an LM833 will work better than a TL072 for my guitar input, & I think a 500k to 1M pull-down resistor will give an acceptable compromise between noise performance, & frequency response with a guitar (pickup loading..)

I still don't know how to determine the input resistance of the LM833 from the data sheet..

merlinb · Post by **merlinb** » 21 Sep 2011, 11:45

matt239 wrote: I think an LM833 will work better than a TL072 for my guitar input, & I think a 500k to 1M pull-down resistor will give an acceptable compromise between noise performance, & frequency response with a guitar (pickup loading..)

I still don't know how to determine the input resistance of the LM833 from the data sheet..

The TL07x has an equivalent input noise voltage 18nV at 1kHz. It's current noise is practically nil at 0.01pA at 1kHz. That means you could use a source impedance of over 800k before current noise added even 1dB to the total noise. In other words, you can ignore current noise for FET opamps.

The LM833 is the same as the NE5532. Its voltage noise is 4.5nV at 1kHz, and its current noise is 0.7pA at 1kHz.

The question is, how big does the source impedance have to be before the LM833 gives more noise than the TL (18nV)?

Well, 18 - 4.5 = 13.5nV. This is the extra noise we are looking for when the LM833 noise current flows in the source impedance. Form Ohm's law: 13.5nV/0.7pA = 19.3k ohms.

So to a rough approximation, if the source impedance is less than 20k the LM833 is quieter. Above 20k the TL is quieter.

So what is the source impedance of a guitar pickup? If the pickup has a self inductance of 3 henrys then that's a reactance of 19k straight at away (at 1kHz), and most signal coils are exactly around the 3H mark! Humbuckers even higher. Added to that is the DC resistance (a few k ohms), so we are tipping into the TL07x region. But if the guitar vol pot is turned down a bit then this will add a LOT to the source resistance, and we are well into TL07x territory.

At lower frequencies the pickup inductance has an even higher reactance, which is a bad thing because opamp noise current also increase at low frequencies! This clinches the deal- the TL07x is going to be quieter 90% of the time.

Input resistance of the NE5532 is quoted at 300k typical, 30k minimum, and the LM833 is practically identical to the NE.

By the way, the TL07x is quoted as having an equivalent input noise voltage of 4uV across the audio band (a discrete FET or triode would be closer to 1uV). Anyway, that's the same average noise as a 100k resistor (although resistor noise is white so it would probably sound a bit worse). You should keep any series input resistor to less than about 20k, otherwise it will add an extra 1dB or more to total noise (the fact that this is the same as the 20k used earlier is a coincidence).

matt239 · Post by **matt239** » 21 Sep 2011, 23:54

AH HA!
That gets down to it! Most helpful. Thanks!

I knew not to just look at the voltage noise spec, so I used an online calculator to compare them, but still got LM833 as quieter, but I knew people must be still choosing TL072 for a reason; now I understand what else needed to go into the calculation.

So it seems I should use TL07x..

Question 1: Are we any less likely to destroy the LM833 than TL, from ESD etc? (heard some differing opinions on that sort of thing..)

Question 2: Assuming we need another I.C., would an LM833 be a better choice to follow after our TLxx input stage than an additional TL ??

juanro · Post by **juanro** » 22 Sep 2011, 00:46

Hi matt,

Regarding question 1 I have this bookmarked since long ago.

Hope it helps, cheers.

Juanro

kleuck · Post by **kleuck** » 22 Sep 2011, 02:08

matt239 wrote: Question 2: Assuming we need another I.C., would an LM833 be a better choice to follow after our TLxx input stage than an additional TL ??

If input, bias, and feedback resistors are kept of low values, yes.

merlinb · Post by **merlinb** » 22 Sep 2011, 09:07

matt239 wrote: Question 1: Are we any less likely to destroy the LM833 than TL, from ESD etc? (heard some differing opinions on that sort of thing..)

ESD is a non issue. Most of the world's audio gear uses opamp inputs, and I've never even heard of a damaged one.

Question 2: Assuming we need another I.C., would an LM833 be a better choice to follow after our TLxx input stage than an additional TL ??

What Kleuck said.

gus · Post by **gus** » 22 Sep 2011, 10:35

Some threads One circuit in this thread looks like one I posted the Schoeps like JFET buffer.
https://www.diystompboxes.com/smfforum/ ... ic=92274.0

https://www.diystompboxes.com/smfforum/ ... ic=92591.0
https://www.diystompboxes.com/smfforum/ ... ic=93793.0
https://www.diystompboxes.com/smfforum/ ... ic=93712.0
https://www.diystompboxes.com/smfforum/ ... ic=93699.0

matt239 · Post by **matt239** » 25 Nov 2011, 04:05

OK so how does this look?

phatt · Post by **phatt** » 25 Nov 2011, 08:31

matt239 wrote:OK so how does this look?

Input Section.JPG

Your wasting your time and effort on silly stuff. If this was the input for a scope or a circuit that makes the space program work then yeah sure cover every base ,, but for gadz sakes man,,, a million of opamps function just fine * even when badly designed*.
I guess It matters little what I say but if you wish to chase rainbows then fine by me as it's your time wasted not mine.

In my View and 30 plus Years experience,,,,Anyone who uses 1 or 2 meg pots on there strat does not fully grasp what is happening.
Let me know when you find the design flaws in the universe

Phil.

matt239 · Post by **matt239** » 25 Nov 2011, 17:52

a circuit that makes the space program work

No, it ain't for the space shuttle..

I hope I'm not wasting my time.. sure I could just build something and it might "work" but I may as well try to understand what I'm doing..
So partly this is just an exercise. - I don't know if you know just how much of a noob I really am.

Before I started this thread, I couldn't design an op-amp guitar input stage from scratch at all..

I didn't get why the TL072 would be quieter than LM833.. (the part I was really missing was how to calculate a value for the source impedance, which made all the difference.)
(THANKS merlinb!)
It may not be a critical difference, but since both parts are cheap & widely available, might as well choose the one best suited to my purpose.
- Once I have a method for doing this that satisfies me, I can use it over and over as the input for various circuits.

Noise annoys me. Some familiar circuits have terrible noise, so I wouldn't want to just copy those..
I don't believe in MOJO, (no carbon comps or tropical fish will appear in my pedal builds..) & I don't care about anything that I, or most people with normal hearing can't actually HEAR..
So.. I wouldn't pay twice as much for a component that had lower noise that was measurable, but not audible, or spend weeks agonizing over it..

The reason I spent so much time on this wasn't so much perfectionism, as it was cluelessness.. I don't have the knowledge & experience yet to always know what matters, & what doesn't..

I appreciate everyone's help & patience.

I tried to simulate the circuit I posted in TINA, but I'm not getting results that make sense. Either I've done something wrong with the circuit, or I just don't get how to use the software.. so I thought I'd ask you guys to have a look.

phatt · Post by **phatt** » 26 Nov 2011, 14:42

matt239 wrote:OK so how does this look?

Input Section.JPG

Matt,
Point taken,,, and thanks for explaining your position.

Sorry if come across a bit rash

But I'm trying to save you wasting time on pointless stuff.
I hate to see people spending x years working on mindless maths when all you may need is a few pointers.
Leave all the terms like slew rate, .000001 % distortion silly stuff for HiFi nuts. LOL

The mistake on your schematic has the pos input DC coupled to both Common and bias/Vr.
C1 need to be *After R3*.
The output needs decoupling as well,, depends a bit on what comes after.
(Even with mistakes Sims can still give results which may, to the novice look like it works)
Phil.

matt239 · Post by **matt239** » 27 Nov 2011, 03:07

Cool. Thanks.
I changed the schem, but it still doesn't work in the sim.
It shows it not having any voltage gain.
I think it should have about x6 gain..

What am I missing?

PlutoniumBob · Post by **PlutoniumBob** » 27 Nov 2011, 04:42

Connect R2 to Vr instead of GND (or use a DC blocking cap).