Single or Dual opamp package?

[JojoD818]

Hi,

This may sound stupid but I really can't let it go and I need your opinions.

Let's say I want to build a buffer for my gainclone, leach amp, or whatever amp that I have that has 2 channels (stereo). Would it make a difference if I use 2 single package opamps or just 1 dual opamp provided that my buffer only needs 1 opamp per channel.

I hope this makes sense.

JojoD

 

[PMA]

For the highest requirements (crosstalk etc.) use single chips.

 

[Pedja]

Though they have some shortcomings (for instance you can’t use separate supplies), dual opamps give that interesting possibility for better layout of the feedback loop - practically always their output and inverting input are neighbor pins. This might be important.

Pedja

 

[cadmaniac]

I found that using some singles helped with my PCB layout (4-way active LR2 crossover). Being new at this; I didn't want to crowd it much. I also used some physically larger-than-typical caps, so double opamps would've meant more longer traces.

 

[millwood]

most of dual opamps are fabricated completely separately, and some of them specifically say so in the datesheet. so it would surprise me if you can hear a difference between dual vs. 2x single.

 

[dhaen]

most of dual opamps are fabricated completely separately,

That may well be true, but the heat from one half affects the other. If you're using a dual, don't use each half for different applications.

 

[purplepeople]

What do you mean different applications?

1) Opamp A for left channel and opamp B for right channel.
2) Opamp A is buffer for opamp B in high gain.
3) Other configurations?

ensen.

 

[dhaen]

Jensen,

I was talking abour using different drive currents from each section, so (3).
So let's say use them for the same mean signal level, and with same load.

 

[Geoff]

most of dual opamps are fabricated completely separately, and some of them specifically say so in the datesheet. so it would surprise me if you can hear a difference between dual vs. 2x single.

It may well be the case that the two op-amps are separately fabricated, but they share common supply rail connections so one op-amp will modulate the other unless they are carrying identical signals.

It is surely far better to use single op-amps where the individual supply connections can be suitably decoupled and isolated so that cross-modulation cannot occur.

 

[millwood]

It may well be the case that the two op-amps are separately fabricated, but they share common supply rail connections so one op-amp will modulate the other unless they are carrying identical signals.

even if you use singles, they share the same supply rail, or the same mains connection, or the same distribution networks, or the same power plant, etc.

Plus, look at how impressive PSRRs and cross-talk attentuation modern op amps have, .

 

[Geoff]

even if you use singles, they share the same supply rail, or the same mains connection, or the same distribution networks, or the same power plant, etc.

Plus, look at how impressive PSRRs and cross-talk attentuation modern op amps have, .

Agreed, but at least with single op-amps you can bypass and decouple individual supply rails, have separate R-C or L-C filtering or even dedicated shunt regulators.

PSRR and channel separation figures are indeed impressive at low frequencies, but above 100kHz .............

 

[millwood]

PSRR and channel separation figures are indeed impressive at low frequencies, but above 100kHz .............

in theory, that's right. But let's take a real example. I recall that in the gaincloen PS discussion, there are quite a few of simulated, unregulated ps and FTT on its ripples (I was one of those who started the simulation back then). I think the 50hz and above signals are about -40 to -50db and declining steadily so it should be much lower than that at 100khz. But let's just assume that ripples at 100khz is -40db. that's about 0.01*18v = 0.18v for maximum rail.

Let's take tl082 for example. At 100khz, its psrr is 20db for -rail and 60db for +rail. That translates into about 0.18v*0.1=18mv for the -rail and 0.001*180mv=0.18mv for +rail.

each of us can make a judgement on if we can hear 18mv or 0.18mv 100khz ripples when a crappy opamp like the tl082 is used.

for opa2132, those figures are 2.2mv for -rail and 0.06mv for +rail.

 

[EchoWars]

Would it make a difference if I use 2 single package opamps or just 1 dual opamp provided that my buffer only needs 1 opamp per channel.

As you can see, there is a bit of contention on this issue.

My $0.02, use whatever is convienent to the layout at hand.

 

[millwood]

byw, I have never seen how channel separation figures change with frequency.

 

[Geoff]

byw, I have never seen how channel separation figures change with frequency.

There is a graph in the OPA134/2134/4134 datasheet (and possibly others, I haven't checked). It only goes up to 100kHz but can be extrapolated from there.

 

[millwood]

There is a graph in the OPA134/2134/4134 datasheet (and possibly others, I haven't checked). It only goes up to 100kHz but can be extrapolated from there.

yes, you are right. it looks like 135db through 10khz with 2k load, and goes down from there, to 130db at 20khz, and 95db at 100k.

so if one channel is swinging full 18v (actually 18v-1.5v), the impact on the other channel is 0.003mv (0-10khz), 0.006mv (20khz), and 0.3mv (100khz).

if this thing is used as a preamp, and with 1v peak output, all figures would be 1/18th of what presented above.

That's pretty in-material, isn't it?