Basic Op-Amp theory (Vref needed with dual rail supply?)

[smugtronix]

Hi all:
I'm a long time tinkerer with very little formal knowledge. I have a question about very basic op-amp theory.


As you can see, this is a unity gain buffer with a split-rail supply created by the ICL7660. Do I need to have the 1M/1M Vref network to make this work, or does the dual supply make that unnecessary? I'm a bit confused. I know you need Vref networks like this when dealing with single supply, but what about dual rail?

[Kremmen]

At least you need a bias current path from the inputs to ground to prevent the amp from drifting to one rail. The resistors will do that (supply voltage nodes work as well).
The voltage division will set the ouput operating point at the same time. To get maximum swing you want to set it at 0 volts, but it can be something else as well.

[Mechatrommer]

depending on your purpose, the source circuit and signal's nature at INP, you may remove R1, R2, C1 alltogether.

[fcb]

If your signal is AC coupled in, then you can do a couple of things.

1. If the input is a 'sensible' size, then you wouldn't need dual rails, the MAX1044/7660 thing does give you more headroom if you need it.
2. If you want to keep your basic design, then get rid of R1.

[smugtronix]

2. If you want to keep your basic design, then get rid of R1.

That's what I was thinking. I'd ideally want the greatest swing possible. I'll keep R2 as 1M, as that won't mess with the frequency response of the input signal (guitar).

[fcb]

If you get rid of R1, you'll need to think about what happens to your polarised-electrolytics (although in the real-world not much).

Personally i'd use the 7660 to boost the input voltage up, go with single rail and bias it. Also, perhaps a small resistor in the output line to help the opamp drive longer leads (100-470R). I'd also use a cheaper opamp that's probably as good or better like the NE5534 (current drain is a bit higher).

[smugtronix]

If you get rid of R1, you'll need to think about what happens to your polarised-electrolytics (although in the real-world not much).

Personally i'd use the 7660 to boost the input voltage up, go with single rail and bias it. Also, perhaps a small resistor in the output line to help the opamp drive longer leads (100-470R). I'd also use a cheaper opamp that's probably as good or better like the NE5534 (current drain is a bit higher).

Hmmmm. I'll have to try the 18V charge pump trick. There is a bit of objectivity with audio devices (although not TOO much, before you go into the dangerous realm of audiophoolery!), so I'll have to try both. 18V can make a huge difference in terms of headroom and dynamic response, especially on bass.
Current draw is a non-issue here. The box I'll be building this into won't allow for a battery, and most guitar players use daisy-chain power supplies that put out 1-2A. Good to know that the NE5534 is good enough. My board requires a bit of a redesign to make that work (add the compensation cap), but it shouldn't be too tricky.

[w2aew]

FYI - I posted a video a few days ago that discusses op amp supply considerations, including split supplies, single supplies and virtual ground.  Maybe it will be of interest:

[Mechatrommer]

...including split supplies, single supplies and virtual ground...

cant we just combine the "split" and "single" term as simply just as "supply" or opamp power source" since opamp has only two power terminal Vss and Vee, there is no such thing as gnd, which only another term for 0V. sometime it is confusing to select an opamp with these terms.

[fcb]

Mechatrommer: Nope, it has a huge bearing on the overall design, the opamp is just part of the circuit as a whole.

[w2aew]

...including split supplies, single supplies and virtual ground...

cant we just combine the "split" and "single" term as simply just as "supply" or opamp power source" since opamp has only two power terminal Vss and Vee, there is no such thing as gnd, which only another term for 0V. sometime it is confusing to select an opamp with these terms.

That's why I put this video together - to clear up the confusion.

[Mechatrommer]

That's why I put this video together - to clear up the confusion.

yup what i understand from the video is "opamp doesnt care" ok thats what i have in mind. my question meant was that, why the "industries" still choose to term them separately such as "this opamp can be used for both split and single supply" instead of say "this opamp can take 30V supply source (differential)". about most datasheets in the world and also the graph example you are showing in the video is specd to "absolute values" such as ±10V etc etc. why cant everything is specd around "arbitrary" Vcc and Vee only with no need mentioning of the term "ground". let the engineers decide what the Vcc and Vee will be and figure out if "ground" (0V) or the middle voltage (virtual ground) is in their common mode range ("common mode" is another term that i almost go nuts last time ), is it too much confusing that way? that they are forced to separate the terms "split" and "single". i've seen spec such as Vsupply (Vcc-Vee) = 30V, output swing = (Vcc-3V) to (Vee+3V), everythings are specd relative to this arbitrary Vcc and Vee and no plus minus sign, etc etc. thats more sensible i think? it also will educate newbies not to tie themselve too much with the term ground (our earth) in circuit design. and then they should be aware (learn) that the term "common mode" (something that has to do with potential difference between two circuit/system) is more important than this 0V and an opamp wont give a rat arse on where its standing (on earth or on mars)... imho, no?

[AlfBaz]

Hi Mechatrommer
My understanding is that it has to do with the signals in and out and most of the time you have a signal that is referenced to ground. In a single supply configuration you are biasing the input at a voltage that is above ground so with nothing coming in it will be sitting at the bias voltage.
Similarly with no input the output will be sitting midway between +ve and 0V.
With split supplies the input leg will be at 0V and so to the output with repsect to ground (0V)

With single supplies...
Depending on what's connected, say a microphone input and a speaker output for example, this may be problematic for the signal source and the load.

One side of the mic and speaker are connected to 0V the other to a DC bias voltage. This will cause current to flow through the coil of the mic and speaker causing the mic's diaphram and the speakers cone to shift from its resting point.

What is normally done with circuits that have DC bias voltages is a blocking capacitor is placed on the input and output to prevent the bias voltage reaching the source and load. This the creates bandwidth limitations and non-liniarities depending on the wave forms you are trying to amplify.

w2aew...
Regarding your video and your quick description of the virual earth. I was under the impression that its called a virtual earth in a split supply configuration when the reference input is referenced to ground casuing the output, via the negative feedback loop, to force the input pin to 0V. The input pin is at 0V but is not directly connected to ground (ie the is no direct path to ground) but mimics it by virtue of negative feedback.

In single supply that point is not at 0V but at the "virtual" biasing voltage and given that the input and output signals are most likly to referenced to 0V wouldn't calling that point a virtual ground be a misnomer?

[w2aew]

w2aew...
Regarding your video and your quick description of the virual earth. I was under the impression that its called a virtual earth in a split supply configuration when the reference input is referenced to ground casuing the output, via the negative feedback loop, to force the input pin to 0V. The input pin is at 0V but is not directly connected to ground (ie the is no direct path to ground) but mimics it by virtue of negative feedback.

In single supply that point is not at 0V but at the "virtual" biasing voltage and given that the input and output signals are most likly to referenced to 0V wouldn't calling that point a virtual ground be a misnomer?

In the strictest definition, you are absolutely right.  When the reference point isn't at 0V, it could be a misnomer to call it virtual ground/earth.  But, that being said, this term is used by many/most textbooks, references, appnotes, etc., even in the case of a non-zero virtual reference point.  So, it seems from my 30yrs as a designer, that it has just been accepted.

Mech - I agree, the manufacturers could "clean up their act".  But, I guess the spec sheets are steeped in tradition, where they were spec'd this way when the vast majority of op amp circuits were split supply, and very few units had an input range that included a rail or two, or had an output that could swing close to either rail.  So, when lower operating voltages became more pervasive, and op amps got designed for these applications, manufacturers made a "big deal" out of featuring single supply operation.  Thus, an imaginary line was created between split supply op amps and single supply op amps.  Doesn't make good engineering sense, because the op amp doesn't care.  All we can do is try to clear up the unnecessary confusion.

[AlfBaz]

Mech - I agree, the manufacturers could "clean up their act".  But, I guess the spec sheets are steeped in tradition, where they were spec'd this way when the vast majority of op amp circuits were split supply, and very few units had an input range that included a rail or two, or had an output that could swing close to either rail.  So, when lower operating voltages became more pervasive, and op amps got designed for these applications, manufacturers made a "big deal" out of featuring single supply operation.  Thus, an imaginary line was created between split supply op amps and single supply op amps.  Doesn't make good engineering sense, because the op amp doesn't care.  All we can do is try to clear up the unnecessary confusion.

I agree completely with your logic here but again, semantically, your "typical" split supply configured op-amp has 3 nodes +ve, 0V and -ve, typically sahred throught the entire system where as the single has only 2 and you have to create the 3rd and normally isolate it from the rest of the circuit usually via ac coupling

[Mechatrommer]

thanks w2aew you've reinforced my understanding. its just because of "natural selection" that hasnt "evolved", we just follow the trandition (industries) such confusion is demonstrated in the OP, thats why i brought this thing up. but anyway, i understand why we cant get rid of the term "ground" (0V), its our ultimate reference anyway (earth).

back to OP, the C1 is to pass AC signal component and blocking DC component, hence the signal out of C1 will be pure AC (centered at 0V), the R1 and R2 voltage divider is just to provide another DC biasing for the AC signal so it will be centered at opamp's midpoint, ie Vs and 0V. that is for "single" supply circuit which i believe the OP copied from 100% in the first place. since the OP are now using "split symetric" supply, there is no more need for the AC coupler + bias network, unless your input signal common mode (DC component) is way off, then you'll need to put the C1 to block the common mode voltage (i hope i used the term "common mode" correctly).

I agree completely with your logic here but again, semantically, your "typical" split supply configured op-amp has 3 nodes +ve, 0V and -ve, typically sahred throught the entire system where as the single has only 2 and you have to create the 3rd and normally isolate it from the rest of the circuit usually via ac coupling

from my understanding, the 0V reference is not for the opamp, but for the whole circuit. but anyway i cant disagree, you look at it from different perspective, which is how normally the way people think of (including me most most most of the time).

[AlfBaz]

... because the op amp doesn't care.

please bare in mind that I speak of the entire feedback amplifier cct and not just the opamp chip

P.S.
Ive alway bheen confused about the naming of an op-amp. Is the chip a high gain diff amp and it becomes an operation amplifier when wired as one or is the chip itself an op-amp, ro at least generally regarded as one?

[w2aew]

Ive always been confused about the naming of an op-amp. Is the chip a high gain diff amp and it becomes an operation amplifier when wired as one or is the chip itself an op-amp, ro at least generally regarded as one?

It goes back to the origins of the operational amplifier  - originally conceived as a component for analog computing, the operational amplifier could function as a gain stage, a differentiator, an integrator, a summing block, a difference block, etc., all by proper selection of the surrounding components and feedback.  Of course, it is simply a very high gain differential input amplifier, but it isn't called that mainly due to tradition, and the fact that it is rarely used without some external feedback components.  Calling it an amplifier would imply that it would be used as a standalone amplifier, similar to the LM386 audio amplifier chip.  Splitting hairs, I know...

[smugtronix]

So here's what I've come up with for version 2 of the design. The compensation cap value is to be determined (I'm assuming 33pf should do the trick.) I've also done an EAGLE layout. I'm pretty sure it works, but it is neither elegant nor pretty. The MAX1044 is actually an ICL7660 (has a slightly higher maximum voltage, so accidentally plugging in the wrong supply won't cook it QUITE as fast)

[AlfBaz]

Hi Smugtronix

I'm curious, what's it for and what do you want it to achieve?
So far I'm guessing that the input is a passive bass and the output goes to iether directlty to an amp or more effects

[fcb]

Smugtronix: Looks good, might be worth adding a decent size pair of caps across the +/- rails (470uF/16v?), you won't damage the MAX1044 and you'll get a 'stiffer' supply thats more symmetrical at audio frequencies (at the moment, your battery supply has no added capacitance), but other than that looks OK.

In my experience, the ICL7660A is a pretty quiet chip, and I've used to generate audio opamp rails before.

That 100nF gives you an approx -3dB point on the bottom end of 1.6Hz - which is OK - the same as driving it into a 10K load.  I might bump that 10uF up to a pair of back-back 100uF parts (gets round the polarization).

AlfBaz: I think the OP is using it to buffer electric guitar.

[smugtronix]

Smugtronix: Looks good, might be worth adding a decent size pair of caps across the +/- rails (470uF/16v?), you won't damage the MAX1044 and you'll get a 'stiffer' supply thats more symmetrical at audio frequencies (at the moment, your battery supply has no added capacitance), but other than that looks OK.

In my experience, the ICL7660A is a pretty quiet chip, and I've used to generate audio opamp rails before.

That 100nF gives you an approx -3dB point on the bottom end of 1.6Hz - which is OK - the same as driving it into a 10K load.  I might bump that 10uF up to a pair of back-back 100uF parts (gets round the polarization).

AlfBaz: I think the OP is using it to buffer electric guitar.

Awesome. I always use the ICL7660S (the S designation allows for higher oscillator frequencies by jumping pins 1+8, which helps combat whine in  gain circuits), and I could possibly switch the 10uf electrolytic to a 10uf bipolar (Nichicon MUSE or equivalent).
I'm trying to get this circuit to work with all electric guitars/basses, even those with really hot pickups or active preamps. I'm a bassist who uses low output passive pickups, so I need a buffer on the end of a cable to drive my pedalboard.

[AlfBaz]

Bass player ay...
What do you call 2 fretless bass players playing in unison...

A minor second
Boom dsh