How much difference do "low noise" op amps really make?

[trotskyismyniece]

I have an MXR Distortion+ that I like for a nasty punk sound, but it adds so much static/hiss to the signal.  I'm assuming that the culprit is the LM741 which is notoriously lofi.  I've been looking at the datasheets for replacements.  I guess the nV/Hz is how the noise is quantified.  Now, I've looked at some replacements that have a lower nV/Hz. 

Specifically the TL071 which is 15nV/Hz and the OPA2134 which is 8nV/Hz compared to 30nV/Hz for the LM741.  Now, I'm inferring that they will be somewhat quieter, but how much?  Is the Brown Burr going to make the thing whisper quiet so that it's worth costing 6 times more than the TLO71, or is this circuit so inherently noisy that it won't actually make a lot of difference?

I know the cost of ICs isn't enough to break the bank to just socket them, I just hate the idea of wasting any time and money buying into cork sniffing mindset of 'mojo' components that have statistically insignificant differences.  I just figure that the actual noise an op amp makes is something that does make a difference, I just don't know how much?

[merlinb]

I have an MXR Distortion+ that I like for a nasty punk sound, but it adds so much static/hiss to the signal.  I'm assuming that the culprit is the LM741 which is notoriously lofi.

In this case yes, the 741 is particularly noisy. But the feedback resistors are also freaking HUGE (1Meg!) and are also adding a lot of unecessary noise. This could be cured by scaling them down (and scaling up the cap proportionately) without affecting circuit operation.

I guess the nV/Hz is how the noise is quantified.  Now, I've looked at some replacements that have a lower nV/Hz.

In general, stompboxes use pretty large resistances throughout, which means *current noise* is the killer. You want to look for opamps with small input current noise, which means FET opamps like the TL072 (0.01pA/^/Hz). The OPA2134 has even less at 0.003pA/^/Hz, but hardly worth the enormous expense in a stompbox.   (TL0xx are hard to beat for value for money and low noise)

Back of an envelope: If the source impedance of the guitar is 100k, say, then the 741 produces 30nV/Hz voltage noise (sounds optomistic to me, probably more like 60nV for a real 741) plus 8pA*100k = 800nV current noise (I got 8pA/Hz from a data sheet) making 830nV/Hz noise. BUT! the feedback resistors present a source impedance of about 500k to the inverting input, which is an extra 90nV/Hz Johnson noise AND an extra 8pA*500k = 4uV current noise. That's 4.9uV/Hz input noise.
Under the same conditions a TL072 would give 15nV/Hz + 0.01pA*100k = 16nV at the non-inverting input, plus the 90nV Johnson noise  plus 0.01pA*500k making 95nV/Hz at the other input; a total of 111nV/Hz input noise.

Now try scaling the feedback resistors down by 100 times to 10k (and scle the cap up by 100 times) and you will have about 12nV/Hz Johnson noise giving total input noise in either case:
741 = 822nV/Hz
TL072 = 28nV/Hz

So swapping opamps gives -33dB improvement.
Scaling down the resistors to 10k gives -14dB improvement.
Doing both gives -45dB improvment.

[familyortiz]

You're in a position to do some comparison, but getting whisper quiet from a high gain, distortion+ is stretching things a bit. It is still worth the effort to see how much static/hiss you can reduce. The "static" part is interesting. What guitar are you using? The reason I ask is that most Teles have no shielding around the neck pup and it is susceptible to static buildup on the pickguard and Strats can also display this.
As far as chip cost, for a new opamp, you might try getting a few samples from distributers. They're usually more than happy to do that.

[Mark Hammer]

1) As near as I can tell, the Distortion+ depends, somewhat, on using a 741.  I've tried other single op-amps and the tone IS different.

2) Part of the noise (indeed, probably most of it) stems from the lack of any proper filtering, rather than from what the chip produces.  I'm sure you cold stick an ultra-high-end op-amp in there and still get hiss, because it amplifies whatever you feed it.

3) Leave the 741 in there but change the following components:
  a) Change the .001uf cap in parallel with the diodes for something a little higher in value, like .0022uf or even .0033uf.
  b) The 741 uses no high-frequency limiting cap, so stick a 47pf cap in parallel with the 1M feedback resistor.

This should both warm up the tone, and also decrease audible hiss considerably.  And, as always, PLEASE do yourself a favour and read the article starting on page 8 here: http://hammer.ampage.org/files/Device1-8.PDF  where you'll find out that sometimes it's not the op-amps noise properties that matter, but rather where you stick them.

Trust me, you'll like it.

[trotskyismyniece]

I use a Strat with a JB Jr. in the bridge.  I've shielded the pickup cavity with copper tape and star grounded it, getting rid of the grounding loops so it's very very quiet for a strat.  Even with the gain cranked up at high volumes there's not really a lot of static through my Traynor YCV50.  I think I have realistic expectations for how much noise a high gain pedal will add, but the Distortion + adds way more gain than any other pedal I've used, even more than my boss Heavy Metal and Big Muff. It's like standing next to the ocean!

Mark, those are just the kinds of mods I was looking for!  I will be sure to read your article.

[ayayay!]

I think it's pretty noticeable to use a TL071, especially when lowering the FB resistor as merlinb suggested.  I like a 680k/8pf combo there. 

[Paul Marossy]

I use a Strat with a JB Jr. in the bridge.  I've shielded the pickup cavity with copper tape and star grounded it, getting rid of the grounding loops so it's very very quiet for a strat. 

Same result on a Fender Jazz Bass, too. 

[petemoore]

  I never really understood the 741 is super-noisey.
   After re-swapping various singles and duals in the DIST+,
other things seemed more important about noise. Single coil pickup for instance is many many times noisier than DIST+/741.
  Difference between various opamps [built a few DIST+'s and OD250, Blue Clipper] seemed nominal if even detectable, a smaller difference than the change in the effect..'clean' opamps didn't seem suited to Distortion [one place I agree on 'which opamp' discussion is in the D+/+ it's derivitaves.
   D+ seems sort of tame in terms of noise [try BMP] and does the old generic distortion tone I prefer with the 741 in it.
 

[brett]

Hi
+1 for hiss from 741s with a big feedback resistor (matches the theoretical analysis of merlinb)
TL072/FET inputs squish current noise to approx nothing, but emphasize line noise due to high impedance (500k+).

A hybrid approach should work: reduce the input impedance to a TL072 by using smaller-than-normal bias/divider resistors (e.g. 2 x 220k for approx 100k of input impedance). The highest of highs might be lost, but probably only dogs will hear the difference  .
cheers

[merlinb]

 I never really understood the 741 is super-noisey.
  After re-swapping various singles and duals in the DIST+,
other things seemed more important about noise. Single coil pickup for instance is many many times noisier than DIST+/741.

Noise comparisons should be done with the guitar unplugged (dummy plug inserted), otherwise all the hum that the guitar is picking up from the air will mask the stompbox noise that you're trying to improve. And that ain't the stompbox's fault.

"Single coil pickup for instance is many many times noisier than DIST+/741" well, yeah... you have a noisy guitar.
You might as well say "a jumbo jet is a lot noiser than a DIST+" so why bother trying to improve things?  Er...you know... cuz one day someone might plug something in that isn't your noisy single coil... maybe....

[Mark Hammer]

The (now 32 years) old Craig Anderton article Ilinked to earlier makes a pretty compelling point that sometimes ostensibly lower noise op-amps can be outperformed by others if the conditions under which they are intended to minimize noise are not met.  So don't just go by what the specs on the datasheet say.  They don't "lie", as such, but neither do they anticipate the particular circuit you're trying to make less hissy.

The other thing we need to take a step back and think about is that many of the instances where people rush off in search of lower-noise options are high gain circuits; i.e., overdrives, etc..  These are also generally circuits that, themselves, generate a LOT of high frequency content.  The question to ask yourself is how much bandwidth (particularly at the upper end) does the pedal truly need?  I find that many of the favourites here can retain more bandwidth than is necessary to do the job.

I hate to keep harping on it, but the Distortion+ and DOD 250 are excellent textbook examples of this.  The 10k series resistor and .001uf cap to ground provide a lowpass filter function (and only a 6db/octave one at that) that starts rolling off high frequencies at just under 16khz.  If it was a home stereo system and I listened to symphonies or other acoustic music only, and I had both my hearing and top-quality tweeters in my speaker system, then I'd be concerned about keeping all that top end, but who the hell is all that HF content for when it comes to distortion boxes going into speakers that roll off starting around 6khz?  Why NOT insert some caps here and there to constrain bandwidth, and in so doing reduce a lot of audible noise?

[petemoore]

  10 people set an eq to their personal taste [from flat ----------].
  10 people state perception on any given component.
   Hypothesis: the differences in eq settings and perception of component differences will be similar...
  Similar: as in very different .

[Steve Mavronis]

I hate to keep harping on it, but the Distortion+ and DOD 250 are excellent textbook examples of this.  The 10k series resistor and .001uf cap to ground provide a lowpass filter function (and only a 6db/octave one at that) that starts rolling off high frequencies at just under 16khz.  If it was a home stereo system and I listened to symphonies or other acoustic music only, and I had both my hearing and top-quality tweeters in my speaker system, then I'd be concerned about keeping all that top end, but who the hell is all that HF content for when it comes to distortion boxes going into speakers that roll off starting around 6khz?  Why NOT insert some caps here and there to constrain bandwidth, and in so doing reduce a lot of audible noise?

Interesting discussion. What are good bandwidth narrowing values for such a noise reducing mod that still preserves the desired tonal characteristics of the Dist+ and 250 still using a 741 op amp?

[Electron Tornado]

Interesting discussion. What are good bandwidth narrowing values for such a noise reducing mod that still preserves the desired tonal characteristics of the Dist+ and 250 still using a 741 op amp?

The question becomes one of filtering. How do we filter out the stuff we don't want and keep what we do want? Start by taking a look at schematics of the Dist+ and it's variants and see what's being used.

I built a pedal based on both the Dist + and the 250 that I don't consider particularly noisey. I have to agree, if noise is a problem, there are other things to look at first instead of simply blaming the op amp or a couple of resistors. Distortion and overdrive pedals are essentially amplifiers. So, if you let noise in, guess what comes out.

[Steve Mavronis]

That is partially true. I have two guitars, both Strats. One is a Fender Yngwie Malmsteen model and it is very quiet 'compared' to my Fender Standard Strat because of the Dimarzio stacked noise reducing single coil pickups in the YJM. But there is still some noise increase with both the Dist+ and 250 as the pedal gain and volume are turned up - as the gain increases so does high frequencies. With the standard Strat there is much more noticable noise. Still, Mark's premise of cutting unneeded high frequencies and thus less noise throughput is a valid solution to mitigate excess noise. But I'm a purist when cloning any classic vintage pedal. I'll leave mods to new semi-clone versions.

[Mark Hammer]

I hate to keep harping on it, but the Distortion+ and DOD 250 are excellent textbook examples of this.  The 10k series resistor and .001uf cap to ground provide a lowpass filter function (and only a 6db/octave one at that) that starts rolling off high frequencies at just under 16khz.  If it was a home stereo system and I listened to symphonies or other acoustic music only, and I had both my hearing and top-quality tweeters in my speaker system, then I'd be concerned about keeping all that top end, but who the hell is all that HF content for when it comes to distortion boxes going into speakers that roll off starting around 6khz?  Why NOT insert some caps here and there to constrain bandwidth, and in so doing reduce a lot of audible noise?

Interesting discussion. What are good bandwidth narrowing values for such a noise reducing mod that still preserves the desired tonal characteristics of the Dist+ and 250 still using a 741 op amp?

Two part question, really.  One part is "what do you send for clipping?" (pre-clip tone-shaping), and the other part is "what do you retain for amplification?".
I suggested two caps to trotskyismyniece.  The feedback cap is intended to limit the frequency content of what gets sent for clipping.  Remember that the clipping will increase harmonic content over and above that.  So, even if you start to roll off above 2khz or so, clipping will assure that you'll still have plenty of HF content above 2khz on the output.  Remember, as well, that a simple feedback cap in that op-amp, in conjunction with the 1M feedback resistor, only results in a shallow 6db/oct rolloff.  So even if you set the corner frequency for 1061hz, using a 150pf feedback cap, content at 2.1khz is only 6db down, and content 4.2khz only 12db down; there is still more than enough treble being sent for clipping, even though 1061hz seems awfully low.  Most importantly for us, and for this discussion, that means that objectionable hiss above that point is not being amplified nearly as much.

But it IS still being amplified, even if only modestly, which is why we want post-clip treble-taming via that cap to ground.

here's anothe thing.  In the Dist+/250 case, the gain is changed by varying the resistance of the ground leg, which has no impact on the top end of the bandwidth.  In the case of the venerable Tube Screamer, gain is increased by varying the feedback resistance.  And in that case, the treble rolloff is progressively lowered as gain is increased, such that the greater the risk of audible hiss (i.e., max gain), the more treble filtering is applied.  Smart, huh?

[amptramp]

I checked out the Craig Anderton article referenced by Mark Hammer and it is a good starting point for understanding how to design for low noise, but one thing it does not cover is the impedance of the source.  If you have a guitar with a 500 Kohm pot as the output level control, this is added in parallel with the input resistor from the non-inverting op amp input to ground.  This can go from zero ohms at a zero level setting to approximately the pickup impedance (on the order of 7000 ohms) at the high setting, but will be at its worst (highest) near the middle resistance level.  You will have a 250 Kohm path to ground and a 250K + pickup resistance to ground through the pickup at the resistive midpoint, resulting in a parallel resistance of a little over 125 Kohms from the source.  At frequencies where the coupling capacitance is negligible, this is in parallel with the input bias resistor.  The high resistance would only affect noise below the turnover frequency set by the coupling capacitor and the source impedance plus the bias resistor.  The noise level in most devices starts to rise below a certain frequency and this is referred to as 1/f noise because it follows the reciprocal of the frequency.  But if you are connected via DC coupling to the clipping diodes, it will affect the clipping threshold even for normal frequencies.  Try AC coupling to the clipping diodes to eliminate this problem, if in fact you are having such a problem.

I had a job at one time where I designed analog filtering in the audio frequency range for military infrared detectors mounted on a rotating platform and used to spot aircraft.  Achieving a low NETD (noise equivalent temperature difference) was the main design parameter and we took no shortcuts.  It got to the point where an aircraft could come toward the sensor and shut off its engine and the thermal imaging would still spot it due to the gray-body frictional heating of the fuselage and wings.  Great fun back in the '70's.

[derevaun]

I've been curious about why, prevailingly, there's an apparent distaste for "noisy" passives, like ceramic caps, but an apparent loyalty to "noisy" chips like the 741. This thread has shed some light on that; at least regarding potential noise sources in front of high gain amplification. Cool!

[Steve Mavronis]

I think that for some, certain classic pedals used the 741 and reportedly, the desired tone isn't quite the same when substituting a different quieter op amp. Yet the 741 being one of the oldest op amps will still remain in use for some time to come in spite of it's shortcomings. I like it the way it is personally. The challenge is to tame the beast! I would like to make or see someone do a true hybrid or merging of the 79-80 Dist+ and 250 tones as a single combined effect (without switching between the two) but quieter at high gain settings. I don't like old school ceramic caps so I use film caps and tantalum electrolytics instead, unless I'm trying to make a 100% replica.

[merlinb]

I think that for some, certain classic pedals used the 741 and reportedly the desired tone isn't quite the same when substituting a different quieter op amp.

The obvious way to fix that would be to use a FET opamp buffer (or just a single FET source follower I suppose) before the 741 distortion stage. Eventually intend to build a discrete fuzz  effect that is similarly buffered, to avoid the annoying "sounds great with this guitar at these settings, but not with that guitar with those settings" effect.