Opamps Bin! - slew rate/power consumption/noise... (long)

[Branimir]

Hi everyone!

After finishing the last pedal (opamp BMP) i started to play with different opamps in the circuit, more out of curiosity then of mojo. Different types of opamps sounded different, and regarding this fact that is diode clipping distortion circuit i begin to think that there is more going on in there then just diodes.

First term that i recognized was the slew rate. What is slew rate, will be explained later in the post.

I thought this could be sort of a introduction and an overview of opamps, so here we go.

Since there are numerous designs that can have one, two, four or more opamps in a single chip it could be wise to list a few, just to save people trouble of browsing and searching

We have single opamps (one operational amplifier inside the 8 pin chip):
TL071
LM308
741

Dual opamps (two operational amplifiers inside one 8 pin chip):
4558
TL072
TL082
5532


I'm sure there are more in both categories, but i cannot think of more right now, since this is from memory (As there are quad opamps lik TL074 and more that i haven't bothered with typing).

All these opamps are interchangeable for each other, of course in their own group (dual opamps for dual opamps of course, and single opamps for single opamps).
They all have the same pinouts, at least the relevant ones (! inverting and noninverting inputs, output and power supply pinouts).

Some of these have compensation pinouts, where capacitor (between the two compensation pionouts) has influence of the opamp's slew rate. (How excatly they influence, momentary i don't know so i'll leave that fact to somebody other to explain, rather than me heh)

Now, what is slew rate?

Searching the forums i found a really good description of the term:

Slew rate refers to how quickly a device can respond to an instantaneous demand.  This is both a function of the frequency demanded and the amount of gain applied.  So, assuming a gain of 1 (unity) there is essentially no difference whatsoever between devices for anything in the audio range, no matter what their slew rate happens to be.  Just about any device can replicate the input signal with a fairly wide bandwidth.  As the amount of gain applied is increased, the device starts to be a little more hampered in swinging that wide/far at the highest frequencies. 

Note that gain and level are different from each other.  Reproducing a very hot signal at unity gain is a trivial matter and slew rate matters little in such instances.

Of course gain is also one of those multiplicative things, and can be split up several ways.  So, if you have two cascaded gain stages, each set for a gain of 12, then your overall gain at the output is 144.  Though the overall gain is pretty hot, the amount of amplification requested from each stage is quite modest, so the slewing requirements imposed on stage are very modest.  Again, for musical purposes, assuming a bandwidth of maybe 10khz, there is really no functional difference across most commonly used devices, despite substantial differences in their slew rate.

Now, if it were the case that you were asking a single op-amp to provide a gain of 200 or more across the full audio spectrum (e.g., for use as a condensor mic or turntable preamp), with maybe another 10khz for extra measure (so, 20hz-30khz), then slew rate would start to become an issue, and devices that slew at least a couple of volts per uSec would probably be recommended.

If you look at the sorts of chips normally used in everyone's favourite stompboxes, though, the overwhelming majority do not exhibit slew rates of more than 0.5-1V/usec.  Those that do, are generally selected for their input impedance characteristics or something else, and are not selected for their slew rate which is only coincidentally linked with these other attributes  A bit like suggesting that people select girlfriends for their shoe size - in some cases, I suppose yes, but in most cases it is merely a coincidence.

In still other cases, avoiding a high slew rate can be a good strategy to follow.  The MXR Distortion+ and the Proco Rat use a single op-amp to provide large amounts of gain.  Since the upper harmonics lend an unpleasant fizziness to the tone, having a reduced capacity to provide big swings at higher frequencies works in one's favour.  In the case of the Dist+ the capacity is reduced by use of a meagre 741 op-amp.  In the case of the Rat, the reduced capacity to yield high gain of high frequencies is produced by means of the compensation cap between pins 1 and 8.

Topic link

Now that we have been introduced into slew rate, we can also imagine how can impact the sound of your distortion (and non distortion devices), whereas slow slew rate opamps would add their own distortion, some may seem more pleasant to our ear than others. Of course these are the facts that are purely personal taste, so there's no point in discussing about "what opamp sounds better" neither this is a post about sort of thing.

This is more of an orientation towards builders that want to chang opamps in their circuit.

Since changing the devices is far more simple and faster than learning the actual reason why do they sound different, i thought this thread would be helpful in finding out why your XY opamp sounds different than YX in your distortion (for example) pedal.

I'm calling everyone who has more in depth knowledge about opamps to contribute to the topic in order to disect popular opamps to common terms that are valuable to us DIY-ers.

Since i'm no einstein in this category, and i have not made any studies about opamps, i'm limited to search/paste and learn.

Here we go:

741
Topic is from archives, August 06 2003 that i have found out to be too large to copy/paste, since all of the answers in the thread are really valuable, so I suggest reading the complete discussion: Topic link

TL071
Here we have an interesting replies from Mark Hammer and Analoguru from 2003 also...

A quick scan of the datasheets indicates that with a +/-15v supply the TL071 can swing to +/-13.5v, or within 1.5v of each rail.  An NE5532 can do +/-13v with a +/-15v supply.  The datasheet for the NJM/JRC4558 indicates it will swing to within a volt of the rails under similar conditions.  With a +/-20v supply, the LM741 can swing to +/-16v with the same load (10k).

What this indicates is that the 741 has considerably less headroom with a 9v supply than a TL071 or 4558 does.  Even before it hits the diodes, it is clipping, and those of us who had the smart idea to lift the clipping diodes and make ourselves a "clean" boost for nothing were sorely disappointed to find the damn thing never ever really got clean even when the diodes were factored out.  At least part of the "sound" of a Dist+ comes from feeding the clipping diodes a signal that is already distorted due to headroom limitations of that chip.  The BiFet chips or others may well provide lower noise spec and are certainly much better choices for use as general purpose audio devices (e.g., in mixing boards or audio amps), but authentic replication of the Dist+ likely involves use of a 741 with all its inherent weaknesses.  It is by no means a "magic" chip, and I would scarcely recommend for "improving" other pedals.  It just happens to be part of why this pedal sounds the way it does.  Not the critical part, but part.

Another important parameter is the slew-rate, a 741 is very slow compared to the TL stuff.  Harris and Rayethon produced a low-noise quad 741 which Neve used in his consoles.  I don´t remeber the number yet (I believe 4156), but this could be a possible try (use only one or build something around with the rest  Cheesy ).  But with noise will not be much improvement.
That you have full benefit from a low-noise opamp the feedback resistor should not be more than 10k for a NE5532, otherwise the resistor produces more noise than the opamp.

That means in my schematic for a TL071:
R6 = 47k, R5 = 220E, P1 = 25k neg. log., and C4 1µF film cap.

greetings, analogguru

Topic link

TL082
From the archives, here's a good overview of the TL082 chip by transmogrifox:

I compared a TL082 data sheet to a RC4558 data sheet and this is what I found:

1. TL082 has higher unity gain bandwidth (4 MHz vs. 2.5 MHz of 4558).

2. TL082 has a faster slew rate.

3. Both have comparable open-loop gain.

4. TL082 has lower noise voltage, better common mode rejection and better power supply rejection.

5. The TL082 has less harmonic distortion.

From an engineering standpoint, the TL082 is a superior opamp. The 4558's are basically an improved version of the 741 series, which are pretty poor audio amps.

From the stompbox fanatic's point of view, I'm finding it's a matter of personal preference.

My personal conclusion about the difference in the sound is that the TL082 is a better tracking op-amp, in control terms. The output is more likely to follow the input, even with abrupt changes in the feedback, such as diodes. Therefore, the output is more likely to track the diode curve more precisely, leaving the sound more smooth and 'creamy'. The 4558 may not respond as quickly to the 'on' state of the diode, and cause harder clipping to occur.

Check out RG's (or was it Jack Orman's) technology of the tube screamer. I think it was Jack Orman, come to think of it....either way, he has some other thoughts about opamps as well.

Topic link

I'm sure there are more, feel free to contribute!

Thanks!

[TELEFUNKON]

recently:
Op Amp Clipping Theory

and Erik`s comments today.

[Gus]

Remember we are often using 9V supplys

I think a lot of the opamp sound for effects is how it recovers from clipping.  Jack,R.G. myself and other have posted about this in the pass.

Opamps I would never use for audio can work well for effects.

Yes slew rate might have an effect but maybe more in the case of what is going on inside the opamp.  One opamp I like has a slew enhance circuit inside that causes a bit more of a 2nd when being clipped or drive close to clipping.

Some opamps are cheap so buy a bunch and try them. 

I forget who it was but a few years ago someone(s) did work with different external comp opamps in the rat, in one of Aron's forums.

[Sir H C]

Slew rate also has to take into account the load resistive and capacitive.  A capacitive load has to be charged and discharged and often this is the limiting factor on how fast an op-amp can respond to an input.  If the op-amp can deliver 5 mA current (weak one), and you have a 1uF load, you can come up with the slew rate (change in voltage with time):

I = C dV/dT

So

dV/dT = I / C = 5 volts per millisecond slew rate => Slew rate for a capacitive load with a current source driving it.  This is a good way to see if the op-amp will work for your needs or if it will crap out.  Resistive load just makes things worse.

[tubewhimper]

I previously left this in another forum, but it has merit in reply to your question as well...

I began using a MOSFET equivalent chip manufactured by Intersil
called the CA3260E in my circuits that normally call for a JRC4558D.
(Mouser has the CA3260E for $2 or so...) I can't find anyone else
using these and they are drop-in replacements for the JRC4558D
except they are zener-protected MOSFET based. It has excellent
characteristics compared to the other more 'common' substitutes.
The slew rate is 10V/usec as opposed to 1V/usec for the 4558.
The noise figure is lower, & the CA3260E is optimized for single
supply and can be driven with a signal that is NEGATIVE with respect
to the artificial ground (by 1/2 volt!). To top this off, the output signal
can swing right up the supply rails! I have used these successfully
in several of my pedal mods (SD1's and other stuff) so I was curious
to see if anyone else knows anything about using them. 

Nice to have something that responds to VOLTAGE changes like
a tube instead of all of the bi-polar current driven chips. 

BTW: Ibanez used the CA3260E in the MS-10 Mostortion pedal, back
last century sometime.

[WGTP]

That is a nice chip and gets mentioned around here at times.  The OP275 is another one with a little different set up.

Settling Time seems to be a parameter that R.G. was refering to.  I'm not sure if that corralates with the Slew Rate or not.   

[Zero the hero]

Nice analysis, Bran!
Now somebody knows if Slew Rate can be an issue in IC-based envelope follower? I still have not fully understood why some designs are very IC sensitive (for example, Dr. Quack or the BassBalls).
Moreover, some EHX LFOs doesn't work with any TL0x4 chip... People told me that the Bad Stone LFO doesn't work properly with this chip. Could it be a slew rate problem?

[Paul Perry (Frostwave)]

When an LFO design doesn't work with a particular type op amp, in my experience it's been because the input voltages to the op amp have gone out of range. Looking at op amp data sheets, you can see various restrictions. It's possible that particular designs cause 'latchup' in op amps that are prone to it.

Speaking of latchup... I remember one envelope follower proto I made, that suffered from this. It was pretty insiduous, because it only happened during momentary overloads. Then the op amp is out of action until the unit is powered down.

[WGTP]

I don't remember the thread, but Mark Hammer was contending that unless the slew rate was extremely low, it had little effect on the circuits we use most the time around here.   

[Zero the hero]

When an LFO design doesn't work with a particular type op amp, in my experience it's been because the input voltages to the op amp have gone out of range. Looking at op amp data sheets, you can see various restrictions. It's possible that particular designs cause 'latchup' in op amps that are prone to it.

Speaking of latchup... I remember one envelope follower proto I made, that suffered from this. It was pretty insiduous, because it only happened during momentary overloads. Then the op amp is out of action until the unit is powered down.

And the TL074 latches up at its top of oscillation. Power supply could be an issue, thanks for the hint! Maybe the TL074 doesn't feature enough voltage swing compared to LM324 or Sharp IR3702 (which I used on my BS). I'll go and check the datasheets.
I love this thread!