In the pursuit of accuracy, I must admit to a mistake in calculations. A 4KΩ pickup has 8 nV√Hz noise. Added to the TL072's 18nV√Hz gives 19.7nV√Hz, not the 26 I quoted earlier. The OPA's 4.5nV√HZ will give a total of 9.2nV√Hz, not 12.5. Thanks to Gnobuddy for reminding me that noises don't add linearly; it's the square root of the sum of the squares. (Didn't the Strawman say something like that to the Wizard?). So it's 6.6db difference.
Head of Quilter Labs, a manufacturer of guitar amps. Have you not heard of them?
Actually, it wasn't in any advertisements; it was a response on a guitar forum to a complaint from a guitarist about the excessive hiss from their previous amps; he responded that they had lowered the noise in their current production by switching to OPA1654 opamps from the TL074 ones used previously.
So there are different ways of judging the effective noise of an opamp used as a guitar preamp. But IT'S STILL QUIETER with an OPA1652/54 rather than a TL072.
That's an unrealistic experiment---we've already established that the source resistance of a guitar pickup is ~ 4KΩ, with the addition of PRR's 3 KHz peak which would raise it a little more, but certainly NOT 23 KΩ.
Yes, it does. And I still say it's a relatively inexpensive way to improve a circuit.
Perhaps it is more $$$ in Canada. I found a PC sheet of adaptors plus pins for about $1 apiece. The OPA 1652 at $1.97 makes it ~ $3. At any rate, $7 or $3 for an audible (to me) improvement is CHEAP!!!
Last edited:
You've moved the goal-posts a long, long, LONG, way in order to come to this very unrealistic conclusion.
We started out with a guitar with an 8k DCR humbucker and a 500k volume pot set to half-resistance or 85% volume; that has an output impedance of 127k, which I rounded down to 100k for convenience.
There are Fender guitars with 250k volume pots and some 4k DCR, so I'll happily accept half that - 63.5k - as a realistic output impedance for some guitars. Round that to 50k for convenience, it makes no practical difference.
Anything below 50k is increasingly contrived. And now somehow we've magically arrived at the magic-unicorn-pixie-dust-guitar with an output impedance of 4k, which is more than thirty-two times less than an honest 127k, and about sixteen times less than an honest 63.5k. This is complete fantasy, not reality.
I don't know a single guitarist who only plays 4k pickups with the guitar volume permanently set on maximum. Perhaps that's how a beginner with her first Squier Bullet plays during her first week; but if she has any musical talent, things won't stay that way for long.
But as long as we're in fantasy-land, why stop at 4k? Let's wire sixteen of those 4k guitar pickups in parallel. Now we have only 250 ohms source impedance.
And while we're at it, let's immerse the 16-pup guitar in liquid nitrogen at 77 Kelvin, or about (-200) degrees C. The DC resistance of copper drops to about 13.5% of its room temperature value at this temperature, so our 250-ohm guitar is now down to 34 ohms source impedance.
See, I told you so! An OPA1652 isn't quiet enough for an electric guitar! A 16-pickup Fender guitar with all 16 pups wired in parallel and immersed in liquid nitrogen, that is!
Of course not. It's actually 34 ohms!
-Gnobuddy
I should perhaps mention that I often play the guitar with vol pot at max. And the resulting impedance is far above 4k. And the same is the equivalent noise resistance as explained some postings above. The signal as well as the underlying noise are max when vol pot is in the max position. At least this applies for the wiring of my strat. Maybe you have this crude two PU signal mixing wiring scheme where the wiper of the pot shorts the pickup. Anyway - your calculations based on dc-resistance are true for dc only and irrelevant here.
Seems resistor current noise has not been taken into consideration, or it was implied somehow. The values must be infinitesimally small, but they are still there. Moreover, current noise predominates exponentially approaching 0 Hz.
Also, the .047u capacitor does not itself contribute to noise, but it has an affect on the resistances in the circuit.
Given these facts, perhaps we should be looking at resistor and capacitor types.
Incidentally, I have the single vol pot setup. At the output jack, the range is 5R at zero, and 17k with the treble-only pickup switched on. The box if extremely quiet at zero, noise increases from there to about the 85% knob range as stated above, where it produces the most hiss. There is some noise at 10, which would be the pickups, plus any end resistance, also inconsequential.
So the pot is noisy. I just ordered the Bourns 500k Pro Audio pot. We'll see what happens.
Also, the .047u capacitor does not itself contribute to noise, but it has an affect on the resistances in the circuit.
Given these facts, perhaps we should be looking at resistor and capacitor types.
Incidentally, I have the single vol pot setup. At the output jack, the range is 5R at zero, and 17k with the treble-only pickup switched on. The box if extremely quiet at zero, noise increases from there to about the 85% knob range as stated above, where it produces the most hiss. There is some noise at 10, which would be the pickups, plus any end resistance, also inconsequential.
So the pot is noisy. I just ordered the Bourns 500k Pro Audio pot. We'll see what happens.
Last edited:
Not really - at all settings except max volume, the pot swamps out much of the reactive component of the impedance.
And at max volume there is additional noise around the pickup resonant peak - but that peak impedance is in the neighborbood of 100k, which is, practically speaking, the same impedance as the 500k volume pot at half-resistance.
IIRC, PRR's more sophisticated simulation didn't show an enormous difference in noise voltage vs the simple resistive calculation I did. PRR used 4k where I had used 20k (at Dotneck's suggestion), which did make some difference - roughly one-quarter the resistance, so we expect the thermal noise to roughly halve.
Even before that, in post #114, I showed the thermal noise from the guitar at both 10k source impedance (raw humbucker DCR) and 100k source impedance (500k vol pot at roughly half-resistance.) The conclusion was already clear in that post - to get a 1 dB noise figure (NF) at 10k source impedance, you need an op-amp with 6 nV/root Hz noise; the TL072 has 18 nV/root Hz instead, and NF is around 6 dB.
But turn the volume knob down a little, and the guitar source impedance leaps to 100k; the TL072 now falls on the 1 dB NF line, which means there is essentially no room for audible improvement.
Of possible interest: Joe Bonamassa showing why using the controls on your e-guitar is a good idea: YouTube
-Gnobuddy
I'm not sure I understand "resistor current noise". If you mean the noise current from the op-amp, I've neglected it because, with either TL072 or OPA1652, it has no effect until source impedance becomes far, far higher than you get from a guitar.
It's an easy calculation, and I'll do it later in this post.
This is getting into the "irresistible force meets immovable object", and "How many angels can dance upon the head of a pin?" territory.
Philosophers and mathematicians worry about infinitesimally small things. Philosophers go nowhere because of the nature of the thing itself, so philosophy has stayed in the same ridiculous place for a few thousand years now. Mathematicians sometimes make headway, as with calculus, the mathematics of infinitesmal things, which has turned out to be incredibly useful, because it also describes things that change with time, and that includes almost everything interesting on the planet.
But for engineers and scientists, a thing that has infinitesimal effect is irrelevant. BUT, it's very important to make sure that the effect is in fact infinitesimal, before you put it out of your mind and forget about it.
If I understood you correctly, you're worrying about the effect of op-amp input noise current, and what it does to the total noise voltage.
A preamble: we know multiple noise sources add as the square root of the sum of their squares. That being the case, if you have two noise sources, you can easily verify that if the second one is, say, less than half as strong as the first, its effect becomes negligible:
1) Consider a noise source with a strength of 1 unit, and a second one with a strength of 0.5 unit.
2) Square and add, you get 1.25.
3) Take the square root, you get 1.118.
4) Calculate the increase in noise (in decibels) going from 1.0 to 1.118. This equals 20*log(1.118/1), which equals 0.97 dB. 1 dB is the threshold of audibility.
So if a second noise source is half as strong as the first, or less than that, it contributes negligible additional noise - less than a decibel, and therefore, inaudible.
Okay. For a TL072, the current noise is 10 Femto amps per root Hz. That's 10^(-14) amps, incredibly small.
Let's consider a realistic case, a humbucker-equipped guitar with a 500k volume pot set to 85% rotation; Rout is roughly 125k, which we'll round to 100k. That's 10^5 ohms.
Multiply the TL072 input noise current by the input (source) resistance to calculate the noise voltage it creates: 10^(-14)* 10^5 = 10^(-9) volts = 1 nV/root Hz.
Meantime, the noise voltage from the TL072 itself is 19 nV/ root Hz.
And the Johnson noise voltage from the 100k resistor is about 13 nV/ root Hz.
The noise created by the input current isn't just half as big as the noise voltage; instead, it's 19 times smaller than the op-amp noise voltage, and 13 times smaller than the resistor's own noise voltage. By the time you square that, add, and convert to dB, the noise contribution from the input noise current is nothing flat. Zip. Nada. Nothing.
The OPA1652 has even less input noise current, so the same applies.
Note that this doesn't apply for every possible use-case of these op-amps: if you had a 2 mega ohm source impedance, the TL072 input noise current would begin to matter. If you had a 10 mega ohm source impedance, the OPA1652 input noise would begin to matter.
But at typical guitar source impedances, input noise current doesn't contribute to total noise for either of these two op-amps.
I'm not sure I understand what you're referring to - popcorn/flicker noise? As far as I know, 1/f noise applies to both equivalent input noise and equivalent input current. In good audio devices, total noise stays flat down to reasonably low frequencies before the 1/f noise starts to make the noise floor turn upwards.
For the TL072, that corner is at around 400 Hz, with 1/f noise starting to dominate below that.
Op-amp noise current flowing through any impedance generates noise voltage. But the cap only has a reactance of 41 kilo ohms, worst case (at 83 Hz, the lowest guitar frequency.) That's too low to matter with either of the two op-amps we've been looking at.
The special case is the one PRR brought up: on resonance, an LCR circuit has an impedance that is purely resistive, i.e. there's no reactive component at all. That means it behaves like a big resistor, and that means it also generates thermal noise like a big resistor. That last bit, I didn't know until PRR pointed it out earlier in this thread.
-Gnobuddy
And that's where the problem is when the strings aren't shaking. I read it is unreasonable to calculate noise below .01 Hz. Divide 1 by .01 and we get 100 times whatever the 1/f noise is, whether opamp or resistor?
That may be approaching the computational limit - the numerical integral might be starting to get funny down there.
But guitar speakers have frequency responses that fall like a rock below about 100 Hz. A sealed cab falls off at 12 dB/octave, while an open-back cab falls off slower, but starts to fall off at a higher frequency unless it's a really big cab. The speaker doesn't put out any sound at 0.01 Hz, or at 0.1 Hz, or at 1 Hz, or at 10 Hz...you get the picture!
So I don't think we need to worry about 1/f noise below 100 Hz for guitar. And even then, the ear is very insensitive at 100 Hz compared to 3 kHz. So we usually hear hiss rather than rumble from a guitar amp.
I do remember 1/f noise being audible with several of my DIY cassette playback preamp builds. That was a far, far more challenging signal to work with than an electric guitar. Peaks were usually around 1 mV. This was before the loudness wars erupted, so music had actual dynamic range - a lot of the time, your tape head was only putting out 100 uV or so.
By comparison, a typical humbucker seems to spit out 1-volt peaks if you pick a string with some force, and even a feather-touch will spit out a hundred millivolts (a thousand times more signal than the cassette playbck head.)
Fender seems to have designed their classic amps to reach full output with 20 mV from the guitar, so they must have thought this was a reasonable lower bound - the output from a squeaky-clean Fender single-coil pickup if someone was barely tickling the strings.
Quite by accident, I just stumbled across this: YouTube
Which puts it all in perspective for me. I really don't care whether my guitar amp can be made 2 dB quieter or not; but I care a lot about trying to get even 1% of the way closer to Tim Peirce's awesome musical talents and guitar chops.
What an incredible gift for melody that man has! His solos "sing" better than the song itself does!
-Gnobuddy
OK. 63.5KΩ source impedance, then, for the moment. Thermal noise is 32nV√Hz. Added to the TL072's 18 nV√Hz, it's 36.7, whilst the OPA1652 comes in at 32.3 nV√Hz---that's 1.1 db. You said yourself that 1 db IS audible. SO THE OPA1652 is AUDIBLY QUIETER!!
Then, as long we are "moving goal-posts" (and we don't have any liquid nitrogen available), let's just put another device in front of the fuzz box---say a reverb pedal or an active volume pedal. The we're down to what----50Ω source impedance? Whoa, Nellie, the noise difference between the TL072 and the OPA1652 is back to the original 12 db QUIETER!!! (Hiss, hiss!)
Last edited:
I'm looking at the opamps while they are have no frequency input. They have 1/f noise, and any chart involving noise increases at frequency approaches zero. I'm relating this in terms of idle hiss only.
I do not only hear any hiss, but also no string scraping as he moves up and down the neck. Is it technique, or some of the $50,000 worth of equipment, or a noise gate?
He sounds really good for a graybeard. Gives me hope.....
Enjoy your 1 dB.
Back in post #114 I said:
After 54 more posts, we're back exactly where we started!Gnobuddy said:
And what's in front of the other device? The same guitar?Gnobuddy said:
As we've seen, with a TL072 (or anything else with even lower noise than a TL072), the resistance inside the guitar itself is what sets the S/N ratio.
When you chain multiple linear devices, the combined S/N ratio can't be any better than the signal you started with: the (thermally) noisy guitar signal. In practice, every stage adds some additional amount of noise, so the final S/N ratio will always be at least a little worse than the S/N ratio at the input.
So, the signal coming out of that first device - the reverb or clean gain pedal - has, at best, the same S/R ratio as the guitar.
And the following OPA1652 cannot remove any of that noise - the best it can do is add only a minimal amount of noise, worsen the S/N ratio as little as possible.
So, in your scenario (guitar -> some box -> OPA 1652), the S/N ratio coming out of the OPA 1652 will be no better than the guitar's own S/N ratio.
Which, as you've just calculated, is within 1 dB of the S/N ratio of a TL072 used as input. So, once again, the signal coming out of the OPA 1652 will have a S/N ratio within 1 dB of the TL072...and once again, we find that "You cannot do significantly better than that at any cost - there's only 1 dB of improvement possible, and a 1 dB change is barely detectable by the human ear."
This noise business is slippery stuff. It's hard to wrap one's brain around how it behaves. Thermodynamics in general is like that.
By the way, the OPA 1652 certainly can be significantly quieter than a TL072 in some other applications - if fed from a 100-ohm dynamic mic, say. So it's not that the OPA 1652 can never do better than a TL072. It's just that for a conventional electric guitar, you can't significantly beat the noise performance of a TL072!
-Gnobuddy
I think we're talking about the same thing. When's the last time you heard deep rumbling noises from some audio electronics, rather than just a hiss centred up above a kHz or two? For me, it was back in the 1980s, in the era of audio compact cassettes, before CDs and CD players became affordable.
I was a proverbial starving student at the time, living on a meagre scholarship, most of which went to pay my dorm-room rent. So affordable meant "No more than a hundred bucks." I listened to cassettes well into the 1990s...
I'll bet much of it is technique, but I'd guess he may also be using tape-wound strings on his guitar.
Partway through the video, he plays the slide-guitar solo again, this time on an acoustic guitar. And that is the only clue that he's human and not supernatural: there are a few buzzes and muffled notes in the acoustic solo. (I don't think that guitar was set up for slide playing - the action looks far too low.)
Tim Pierce was a phenomenally good studio / session player. He's been that good for a long, long time.
I'd never heard of him until his name popped up on another diyAudio thread a couple of years ago. He was the studio guitarist on the Monkee's 1966 hit "Last Train To Clarksville", which also hit #1 on the charts.
(There's a video on Pierce's You Tube channel where he talks about that song, too.)
Meanwhile, "Iris" was released in 1998 - twenty two years after "Clarksville". And Pierce was still a top-notch studio musician, and helped create yet another #1 hit.
I will never be anywhere near that good, but he (and others of his level of talent and skill) give me inspiration.
-Gnobuddy
Thank you, I will! And I'll keep in it in my pocket and think warmly of you every time I use my noisy guitar!!Enjoy your 1 dB.
Yes, my WIDE-OPEN quiet guitar. With my 6.6 db LESS of noise. Unless I decide to use my active volume pedal. Then I'm back to my 12 db-less noise than using a noisy old TL072. Of course, you still have your $6.32 and I've spent my $2.97. Enjoy your riches!!!
I was following this, but I got lost when I heard the guitar only had a pickup Z of 4k....
All of mine are >8k, both Humbuckers and single coil.
So far, in all the amps I have owned, I have never had to swap out a TL072 due to noise.
Now perhaps the Ovation armadillo back electro acoustic, with piezo bridge, maybe that could do with a more modern op amp?
No idea. I'm sceptical via a vis audibility dependant on the instrument used.
I can say quite honestly in a line level application, I cannot distinguish between TL072, LF353, NE5532, or LME49720 in terms of white noise.
All of mine are >8k, both Humbuckers and single coil.
So far, in all the amps I have owned, I have never had to swap out a TL072 due to noise.
Now perhaps the Ovation armadillo back electro acoustic, with piezo bridge, maybe that could do with a more modern op amp?
No idea. I'm sceptical via a vis audibility dependant on the instrument used.
I can say quite honestly in a line level application, I cannot distinguish between TL072, LF353, NE5532, or LME49720 in terms of white noise.
I agree---at line level, noise is the LEAST of concern. Even if you consider line level to be -10dbV, all modern opamps are ~ 100 db below that in terms of noise, and thus totally inaudible. Noise only comes into play (in audio use) with very low-level sources, such as phono pickups, microphones, and guitar pickups; of which, only guitars cry for FETs. In those applications, your old TL072s and LF353s are too noisy for me.
As mentioned previously, active pickups reduce the noise to quite acceptable levels. I looked at the EMG ceramic active pickups that claim to be "noise free". This is accomplished by drastically reduced winding turns, a buffer, and low and behold, a 25k volume pot.
I would still use a FET device as buffer or first stage since not everyone has these pickups, or if another guitar needs to be used. The increased performance of high end opamps seems useless considering all the distortion in guitar pickups themselves. I was shocked when I connected my guitar to an oscilloscope to see what the output looked like. It is really bent out of shape, so .000003 THD+N amounts to nothing.
It could be said there are four general realms in music: generating the music, playing it live, recording it, and playing it back through a stereo. These each require special treatments. The "white noise" is always there with high gain, and always was, and is, part of the music. We could also include string scraping, the sound of a whipping guitar cable, etc.....
The ones I've heard don't sound much like familiar passive pickups, and many guitarists won't consider them for that reason.
I've gotten used to having to replace the battery pack or 9V flat battery in my two electro-acoustic guitars, and my three bass guitars, all of which have onboard preamps. But I really don't want to also have to deal with a battery in every electric guitar. I think this is another reason many guitarists won't consider active pickups for their electric guitar.
Fishman is making another attempt to change that with their Fluence line of active pickups. They also chose to use a rechargeable lithium battery, permanently built into the guitar. I have my doubts about this approach - if you find out your onboard lithium battery is out of juice just before you go on stage for your gig, you're up the creek. No quick recharge, no way to quickly swap in a fresh 9V battery. You'd have to borrow someone else's guitar.
This is a fascinating discussion in itself. In a nutshell, the magnetic field falls away with distance (from the pickup pole) in an extremely non-linear way. A guitar string vibrating in this therefore generates an extremely distorted signal, as you noticed when you looked at it on a 'scope.
The pickup output signal is also proportional to the velocity of the guitar string, and not its position, which also changes the waveform quite a bit - it's not the same shape as the physical vibrations of the string.
Anything less than 0.1% THD has always amounted to nothing, except for numbers games and ad copy.
Don't forget the heating / ventilation / air conditioning / cooling fans in the room. In my experience, the HVAC system is usually the largest source of hiss and rumble and white(ish) noise in the room.
-Gnobuddy
Just had an idea. My HBs are medium output. Dimed, it overdrives the vintage Fender. I don't need this; moreover, it doesn't sound good, imo. I recall the HBs are available in increasing outputs over time, obviously for all-tube monsters that produce their own tube distortion. If the distortion is generated between the guitar and amp, the higher output is unnecessary.
The idea is to remove some of the winding and install a buffer and a lower value vol pot. I have a paper on parallel JFETs that says at least four are needed to realize any noise benefit. I will have to look into the current drain, or if it will even make any difference. The 500k pot is the biggest source of noise so it has to go, at the very least.
There might be some tone problems as well messing with this.
Though a brighter sound would be good.Decent commentaries on pickups:
Ultimate Guitar Passive vs Active Pickups
Seymour Duncan Pickups 101 Part 6: How to Find Your Perfect Pickups | Seymour Duncan
One of my guitars in particular - an Ibanez AS73 - really does this. It will easily overdrive the input stage of my (tube) Princeton Reverb, and it turns my Joyo compressor into a fuzz-box.
In an ideal world, the noise drops as the square root of the number of paralleled devices. Four devices in parallel gets you half the noise. It would take sixteen devices in parallel to halve that again, and two hundred and fifty six to halve that one more time.
Obviously that approach quickly becomes completely impractical when discrete devices are involved. But National Semiconductor used to sell an integrated circuit which was basically two NPN "supertransistors", each of which was a vast number of individual BJTs in parallel. The noise specs were really good, and the complexity of vast numbers of paralleled devices was hidden inside the packaging.
I don't know if there is anything like that on the market today, or a JFET equivalent.
...usually followed by the grid stopper or gate stopper resistor between the pickup and preamp. We talked about this early in this thread, but it was subsequently forgotten in the midst of all the subsequent arguments about the merits or otherwise of a 1 dB noise improvement.
DIY guitar amp builds sometimes entirely leave out this resistor, but I have heard an AM radio station come through a DIY guitar amp. Completely omitting an input RF filter of some sort is not a great idea!
For those concerned with every last decibel of noise, perhaps some experiments with Ferrite beads and/or a small inductor might be in order.
-Gnobuddy
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.