Op-amp as unity gain buffer

[Juddnz]

Hi

Im just working through improved humfree A/B/Y geofex schematic and have come across a part i dont fully understand. Can someone please tell me what the resistor on the feedback loop of a unity gain circuit does? it has no resister to earth so it isnt a voltage divider and im assuming therefore isnt doing any amplification? ive looked through my opamp book and cant work it out.

So signal into Non-Inverting pin
Output loops back onto inverting pin with series 10K resistor.

Thanks


Thanks

[joelindsey]

It sets the gain of the op amp.



in the non inverting configuration, the formula for voltage amplification is 1 + R2 / R1. So for unity gain R1 and R2 should be the same.

[ashcat_lt]

^^^Sorry, but 1+(x/x)=2

For unity gain either R1 must be infinite or R2 must be zero or both.  Normally for unity gain we use a straight wire in place of R2 and omit R1.  In the OP's case there's actually a resistor in R2 and no R1.

The only thing I can think is since the inverting input acts as ground, the 10K would limit the load impedance seen by the opamp to something less than 10K.  Why?  No clue.

[amptramp]

There is sometimes an appreciable DC bias current on the inputs of an op amp.  Designers sometimes add a resistor from the non-inverting input to ground and to avoid having the bias current cause an output voltage shift of (Ib * R), a resistor of the same value is used on the feedback lead.  This leaves the input offset current and voltage to be the source of DC offset.  It is quite common to parallel the feedback resistor with a capacitor to limit resistor noise in the audio band.  But with a TL072, this should not be necessary.

The humfree splitter uses TL072's to drive audio transformers.  Sudden current shutoff from a fast input transient may cause the leakage reactance of the transformer to put out a spike voltage.  The resistor may prevent the voltage from driving the inputs into a range where the outputs change polarity and go to the opposite rail, a common problem with some op amps.

[R.G.]

It's one of those subtleties. Ignore it unless you're really motivated to go understand feedback amplifiers.

[Juddnz]

HAHA fair enough.

i was just trying to combine everything ive read into a practical sense. Have had to order the components from Texas and get them shipped to New Zealand as it is cheaper. Looking forward to making it on some strip board.

I was reading another thread which talked about adding a pot on the transformer output to control tone. Might do some experimenting with that to see what happens.

Im also wondering what would happen if i used the other opamp from the A chipset and set it up as a non-inverting amplifier on the input and put a pot on it for gain control.

[R.G.]

I was only  being a little snide. A feedback resistor with no gain setting resistor in a unity gain amplifier is almost pure side effects. It has some use for equalizing the bias current impedance in bipolar opamps as noted, and for rolling off the high frequency response in amps with gain, although it doesn't work for unity gain setups. It has an effect with the stray capacitance at inputs and outputs, and limits current when spikes trip over the input protection diodes of the opamp (if the opamp has these!).

The bias source impedance has an effect on DC accuracy, which is of almost no importance in many opamp circuits which don't do DC "calculations". So, almost pure side effects in this situation. I put it there for a number of reasons having more to do with experience with sado-masochistic opamps of bygone years than for any specifics of the TL072. It won't hurt, and may help in some relatively bizarre situations. 

A pot on the transformer output for tone control works only so long as there is some impedance in the opamp-output-transformer-pot-tone cap for the tone control to work with. It has a tone effect, but to be able to predict how it will work you have to either (1) know the parasitics of the transformer you use or (2) make the parasitics not matter by swamping them some way. An opamp driven transformer is not as predictably treble-loaded as a guitar output is.

Using the other opamp as an input buffer, followed by a pot for gain control works fine. High impedance to the input, gain control, then driving for the transformers. Good idea.

[amptramp]

Since I don't know the leakage reactance of the transformers in question, I would add diodes to the plus and minus rails from the output of the op amps driving the transformers.  They would be connected with one diode with cathode to the + supply and a diode with the anode to the - supply.  It would not hurt to have some extra resistance in series with the transformer leads to minimize diode current during a spike, although there may be enough already.  This would clamp the op amp outputs to one diode drop outside of the power supply voltages.



Maybe putting some of the snubbing network (C5 - R16 and the corresponding ones for other outputs) on the primary would help.

[joelindsey]

^^^Sorry, but 1+(x/x)=2

For unity gain either R1 must be infinite or R2 must be zero or both.  Normally for unity gain we use a straight wire in place of R2 and omit R1.  In the OP's case there's actually a resistor in R2 and no R1.

The only thing I can think is since the inverting input acts as ground, the 10K would limit the load impedance seen by the opamp to something less than 10K.  Why?  No clue.

Ah, yes! must have been the vodka talking...

[Juddnz]

Hi thanks

That schem isn't the one I'm planning to build. I'd upload a pic but can't figure out how.

This thing is going to be specifically for a bass guitar. Any tweaks or advice needed to archive this or will it be ok?

[jonasx26]

This thing is going to be specifically for a bass guitar. Any tweaks or advice needed to archive this or will it be ok?

I built a slightly modified version of the improved hum free A/B/Y some time ago. Added a gain stage, direct out and active phase flippers.

I gave the build to a bass player for running two amps in parallel.
He could circumvent the low end loss by running his 'clean' SUNN amp from the ABY's direct output and his 'dirty' Bassman from the transformer isolated output.

You will loose some signal starting around about 60Hz with the 42TM018.
It works great with guitars, but with basses there is a noticeable loss in low end.

[Juddnz]

ARGH!!!

any one know of an easy way to overcome this? im guessing better transformer, but another way would be nicer.

Since inve already ordered most of the components i might just remove the transformer from the 'B' channel and use that as a D/O

i think

[Juddnz]

Something such as this.



or

http://docs-asia.electrocomponents.com/webdocs/0d72/0900766b80d726c5.pdf

The frequency response looks pretty wide. But the pri and sec resistances dont quite match.

Would this be an issue?

[Juddnz]

Oh and the basses are wal active basses if that matters at all

[R.G.]

That schem isn't the one I'm planning to build. I'd upload a pic but can't figure out how.

You can't.  This forum doesn't host pictures. You have to upload it somewhere else and insert a link to that other location.

This thing is going to be specifically for a bass guitar. Any tweaks or advice needed to archive this or will it be ok?

As noted by jonas, there will be noticeable loss of low end for bass with those transformers. Edcor makes some 10K:10K's that are specified to perform to much lower frequencies, but I have not been able to test them as of yet.

Something such as this.
...
The frequency response looks pretty wide. But the pri and sec resistances dont quite match.
Would this be an issue?

Good frequency response, but much lower impedance. A couple of issues occur. First, the primary and secondary resistances are of little consequence, as that is the resistance of the wire they're wound with. But the primary and secondary impedances (600 ohms) are much lower than 10K, and that does make a difference.

There are some (ack!) subtleties in the frequency response of transformers. The point of that circuit as drawn is to use an opamp to drive the transformer in a way that extends its low frequency response to lower than it would otherwise be. It does this by driving the primary with a much lower impedance that it would otherwise see.

It's easy to get less than 10K. It's harder to get less than 600 ohms. And it needs a lot of current if you drive it with big signals and ...

Oh and the basses are wal active basses if that matters at all

... you want to use it for active basses, which are almost notorious for producing big signals. Active basses can easily produce signals of several volts each polarity, so the opamp buffer in the front end needs to be able to drive 600 ohms with several volts of signal. The TL072 can do this with 10K, or down to about 2K, but it's gasping for air at 600 ohms.

There are opamps that can do it. The LM833 and NE5532 come to mind. However, they have their own quirks about bias currents, etc. And it's easy to run out of power supply volts with a big, active bass signal. So you will need +/-9V supplies at least. And the power supply currents may no longer be non-trivial. 

And the spec sheets for that transformer says that the maximum signal level at 30Hz is 0dbm, which is about 1V peak. Active basses put out a lot of signal.

The Devil is always in the details.

Like most circuits involving transformers, the biggest problem is availability of a suitable chunk of iron and copper. Always *get the transformers first* and then mod the easy electronics (i.e. everything but the iron and copper) to work with the transformer you can get. In your case, find and obtain transformers. The 600 ohm ones you mention are probably OK-ish for experimenting, and may work OK. But there are enough warning signs that it has to be thought of as cautionary. May be fine, but you may also be looking for other transformers later, as the spec sheet doesn't flatly say "they'll work" in the numbers. Of course, the spec sheets for the 42TM018s also said they flatly would NOT work. So there's room for experimentation. 

[FiveseveN]

And the spec sheets for that transformer says that the maximum signal level at 30Hz is 0dbm, which is about 1V peak. Active basses put out a lot of signal.

One (ugly) way is to attenuate the input signal and boost it at the output. But I have another suggestion: go back to the root of the issue, of which we know little. So, Juddnz, what exactly are you trying to use this for (split between what)? Are you certain you need galvanically isolated outputs? There are other ways to isolate audio signals (wireless, digital or analog light modulation), more complex but sometimes more practical.

[Juddnz]

Thanks for the replies guys.

Back to the design phase.

The application is to split the signal to run a clean amp and a dirty amp. I would also like it to switch A/B or A+B

the chain will go something like this

active bass => Tuner =>tremelo=>Bass Whammy=>SPLITTER THAT IM TRYING TO BUILD=>more pedals=>dirty amp
                                                                                                                                  =>more pedals=>clean amp

Im trying to incorporate a ground loop isolator into the splitter, to allow for ground loops obviously. I may just prototype it without loop isolation and see how much hum there is. If im not using the Tx's thats going to eliminate all of the opamp circuits right? so i can just make it a passive splitter and run some LED's through a TPDT to a batt for indication?

OR will i still use the opamp buffers etc? im concerned about volume losses when halving the signal but if i use an opamp im going to be sending a low impedance signal out which isnt good (i think).

BTW everyone here is about 1000 times better than my electronics tutor was. He knew his stuff but couldnt explain it very well

[ashcat_lt]

The active bass and the pedals all have low out-Z.  This is a good thing in that you can get away with a passive split and not worry about treble loss from cable capacitance.

As long as the in-Z on either side of the split isn't too far off (they match fairly closely one to the other) you shouldn't have any noticeable signal loss.  Plug everything into the same power strip and pray.  You might get away it!

[Juddnz]

a little something like this

[familyortiz]

Not that it helps for the isolated channel, but keep in mind you have to transformer couple only one of the channels. Also, the Xicon 42TM018 transformer, when driven by an opamp can go pretty low... until you test it, it may surprise you. cheers, M.O.