Question about opamps that torments me

[kraal]

Hello,

The following question is the type of questions that drives me crazy until I get an answer that helps me understand how to make wiser (less random) choices.
As I want to stay sane, I decided to ask it here.

Imagine the following situation using 6 opamps: input buffer -> gain stage 1 -> gain stage 2 -> gain stage 3 -> active tone stage -> output buffer

If we state that in the above situation the opamps can be of the same type (for instance TL07X family) and if we exclude obvious reasons such as space saving or keeping tracks short,  is there a rule / are there best practices about choosing

• if one should use single/dual/quad OA,
• how to "chain" them

For instance, in such a context, how can I decide if I should do :

• U1A -> U1B -> U2A  -> U2B -> U3A -> U3B
• U1A -> U2A -> U2B  -> U3A -> U3B -> U1B
• U1 -> U2A -> U2B -> U2C -> U2D -> U3
• any other combination

Thank you for helping me having a better night (with no nightmare about opamps trying to drive me crazy)

[11-90-an]

From what I see with some builds, they seem to use different ICs for main circuit and in/out buffers... but I don't really know why... noise and stuff like that, I believe...so for me I would go for :

U1A - U2A, B, C, D in whatever is convenient in layout of pcb/perf - U1B

So I'm not the only one who goes crazy when I don't get an answer to a specific question.... which can lead to sleep deprivation...

say, its 12:40 midnight here... nah, it's ok... there's no ideas that can't make me slee- NOOOOO 

[Ripthorn]

For me, if they are all the same family/type, I just do whatever makes the most logical sense for layout. For example, on a PT2399 delay, I'll use one dual opamp for the input/output buffers, since one of my signals to the output buffer comes out of the input buffer (dry signal). Sometimes two duals gives more flexibility in layout than one quad. Other times the quad is better.

[Mark Hammer]

And that torments you?    Jeez, if you ever worked in an emergency ward or in politics, you'd never sleep. 

I'm with Ripthorn on this.  "Layout" can include not only how simpler passive components are placed, but also things like pads for pots or switches.  So, sometimes placing a couple of duals "nose to nose" (pointed at each other with pin 8 on each a very short distance from the other) can provide a suitable convenience, and the output of pin 1 on the one unit easily going to pin 6 on the other.

The textbook case for NOT using a quad would be one in which an op-amp or two are used for an LFO.  LFOs are prone to producing audible "ticking" as they draw sudden bursts of current from the power source to produce an initial square wave.  Decoupling the V+ line of the LFO can help to more or less eliminate ticking, as will use of a low-current op-amp.  But the optimal op-amp properties for doing that are different than for an audio signal.  This is why you'll often see modulation pedals using one sort of op-amp (LM358, TL022, etc.) for the LFO and a different kind for the audio path.

  One alternative, not particularly well-exploited by folks here, are SIPs.  Having all 8 pins of a dual op-amp in-line provides the builder/designer a choice of which side they want to run traces to other components.  Of course the choices of op-amp types is somewhat less than in the DIP form factor.

[FiveseveN]

For delays and modulation it's fine but for high-gain designs you probably shouldn't use the same chip for input and output. TL07Xs promise 120 dB crosstalk attenuation, which might be enough, but you're going to make it harder on yourself when it comes to routing.
Do you really need 6 op amps, though? Is gain stage 1 low impedance or is there another reason for an input buffer?

[Kipper4]

R.G. Keen has an excellent book on layouts.

It's worth looking how previous similar ideas to yours are laid out.
There must be something like it already, out there. Study the schematic and layouts.

It's worth thinking about jellybean buffers sometimes, not much saving on a layout footprint.
But worth considering.

It can take a while to improve layout skills. I'm a could do better but it will work.
Good luck and enjoy it mate.

[kraal]

Thank you for all your answers.

@Mark, as long as I understand what I do and why I do it, I do sleep quite well (otherwise I'm like a Border Collie that is unable to gather sheep) Indeed LFO are a case I forgot about in my list, ty
@Kipper4, I'll have a look at it, ty.
@Fiveseven, it is a theoretical example  to illustrate my question
@Ripthorn, 11-90-an: As I wrote, saving space, having shorter tracks are obvious reasons to use dual/quad opamps (I can add cost, less parts to insert, to this list of obvious reasons)

In fact I was wondering about non-straightforward reasons that have an impact on the decision to or not to not and how, in what order use them, such as:

• best practice that consists in separating input traces from output traces
• impact of gain (high gain vs unity)
• Impact of Opamp type (bipolar, JFET, CMOS)
• Impact of Opamp circuit type they are used in (inverting, non intverting, integrator, voltage or output amplifier, etc.)
• impact of voltage (if any)
• Impact of Package
• Crosstalk risks (due to gain, layout, other)
• Power dissipation issues / temperature rise
• Noise risks
• 

[Rob Strand]

In fact I was wondering about non-straightforward reasons that have an impact on the decision to or not to not and how, in what order use them, such as:

A book or tutorial on PCB layouts would address a lot of those, and a general stuff on opamps would address other, like Ron Mancini's book,

https://web.mit.edu/6.101/www/reference/op_amps_everyone.pdf

Higher voltages and faster edges can couple more signal between PCB tracks. It's a good idea to keep those things away from sensitive parts of the circuit.    Also noise couples into high impedance circuits more easily than low impedance circuits.    Opamps that swing to the rails or clip could be considered as that category.

For multiple devices in the one package, the sheer fact the pins are close means you are promoting coupling from one circuit into another so you might consider not using the same package for sensitive circuit and a circuit with more aggressive signals.

A hidden form of coupling is through the power supply.  Using the same IC package for sensitive circuits and something driving high currents (or switching) can couple through the power rails.

These types of things apply generally not just to opamps.    If you want to avoid problems keep the angry stuff away from quiet stuff.

[DIY Bass]

<shrug> It's easier to put 3 dual op amps all the same into a shopping cart than to do 1 quad and 1 dual.