inverting guitar phase with an Op-amp

[Super_Reverb]

Your present circuit is creating a +/- 4.5V bipolar supply (relative to ground) by use of the 10k/10k voltage divider and gives you a DC coupled amp. This design does not require, nor will benefit from input or output coupling amps unless you want to introduce low frequency poles, i.e., high pass filter response. This technique works fine, but in the process, you are using 450uA of bias current (9V/20kOhms) to create a bipolar supply.

There are more current efficient techniques to achieve the same function, especially if the 9V supply is a battery: one way is to cofigure inverting opamp input with 1MOhm to 9v supply and 1MOhm to gnd. This offsets the input to 4.5v, which gives you good headroom with 0v-9v rails. This option would require input and output coupling caps. Another method that won't decrease your input impedance is to offset the non-inverting input to mid-supply.

If you want to design opamp circuits, there are three idealized rules you can use to make your life easier. These rules work in the majority of general purpose design cases.

1) Opamp input impedance is infinite. That is opamp draws no input current.

2) Opamp output impedance is zero.

3) Opamp open loop gain is infinite. That is there is no difference in + and - inputs during normal operation.

So, how do these help you? In the design case you provided, since the + input is connected to ground, opamp will force - input to 0V via the feedback resistor. The + input is called virtual ground here.

Since this is true, your inverting opamp circuit's input resistance will be the 470K resistor value.

Another example using these rules with your circuit is: input=+2V, therefore with + input forced to 0V, input current = 2V/470K= 4.26uA, but using these rules, 100% of input current flows through feedback resistor. Now to calculate output voltage. This current flows in feedback resistor, and given sign, -4.26uA*470K= -2V. Gain = -1.

Sorry for the long winded response, but IMO, these idealized rules will help you design and understand opamps like Bob Pease

cheers,

rob

[Gamedojo]

Your present circuit is creating a +/- 4.5V bipolar supply (relative to ground) by use of the 10k/10k voltage divider and gives you a DC coupled amp. This design does not require, nor will benefit from input or output coupling amps unless you want to introduce low frequency poles, i.e., high pass filter response. This technique works fine, but in the process, you are using 450uA of bias current (9V/20kOhms) to create a bipolar supply.

There are more current efficient techniques to achieve the same function, especially if the 9V supply is a battery: one way is to cofigure inverting opamp input with 1MOhm to 9v supply and 1MOhm to gnd. This offsets the input to 4.5v, which gives you good headroom with 0v-9v rails. This option would require input and output coupling caps. Another method that won't decrease your input impedance is to offset the non-inverting input to mid-supply.

If you want to design opamp circuits, there are three idealized rules you can use to make your life easier. These rules work in the majority of general purpose design cases.

1) Opamp input impedance is infinite. That is opamp draws no input current.

2) Opamp output impedance is zero.

3) Opamp open loop gain is infinite. That is there is no difference in + and - inputs during normal operation.

So, how do these help you? In the design case you provided, since the + input is connected to ground, opamp will force - input to 0V via the feedback resistor. The + input is called virtual ground here.

Since this is true, your inverting opamp circuit's input resistance will be the 470K resistor value.

Another example using these rules with your circuit is: input=+2V, therefore with + input forced to 0V, input current = 2V/470K= 4.26uA, but using these rules, 100% of input current flows through feedback resistor. Now to calculate output voltage. This current flows in feedback resistor, and given sign, -4.26uA*470K= -2V. Gain = -1.

Sorry for the long winded response, but IMO, these idealized rules will help you design and understand opamps like Bob Pease

cheers,

rob

I'm happy you gave a very detailed explaination, but I'm not even sure if you recommended any changes. Are you saying that the Version 2 of my schematic doesn't need coupling caps in the in and out if I keep the power supply the same?

I'm not too concerned about the efficiency of the power supply. I might go through batteries faster, but I figure I'd see if I'm running through battteries too fast first before changing that.

[XgamerGt03]

No need for the input and output coupling capacitors. Since you are using a bipolar supply you have no DC anywhere in the circuit other than the power supply.

As far as the 10uF and .1uF caps, you don't need them either if you will be using a battery most of the time with this unit. You really don't need it anyway at all since supply sag shouldn't be an issue. I really don't think your going to ever try to amplify a 9VP-P signal with this so you shouldn't have to worry about the supply voltage sagging.

You could increase the 10k resistor voltage dividers to higher values to to reduce the current that it takes to get your bipolar supply.

[Cliff Schecht]

The 0.1uF and 10uF bypass caps aren't absolutely necessary, but they aren't just put down for transient conditions. They help the op amp with noise rejection (RF, CMRR, PSRR, etc..) and with stability with some parts. I know the TL08x style devices are NOT very sensitive to power sources but even the datasheet for these recommends a 0.1uF cap for the best operation. I agree with the idea to increase the 10k resistors to reduce current draw, any noise that this creates (Johnson and 1/F noise in the resistor) is swamped out by the 1/2VCC bypass caps.

Also, if you leave the output without a DC blocking cap then you WILL be passing 1/2VCC through the output of the device and into your amp. This is obviously a problem with DC coupled inputs..

[XgamerGt03]

Also, if you leave the output without a DC blocking cap then you WILL be passing 1/2VCC through the output of the device and into your amp. This is obviously a problem with DC coupled inputs..

Are you sure about this? Since there is no elevation anywhere there should be no 1/2 Vcc elevation on the output signal.

I've never needed input and output coupling caps when I use a Bipolar supply rather than elevating the input.

The data sheets tend to suggest those when your going to be working with very high frequency signals as this can cause all sorts of problems. I'm not sure that they are entirely necessary for guitar products, but if he has the caps laying around its not too much to add them.

[Cliff Schecht]

You are misreading his schematic, or rather I should say his schematic is drawn somewhat crudely. He's using the ground symbol to denote the 1/2VCC elevation to the "+" pin and this will be present on the output as well. This circuit is used often for amplifiers that are feeding an ADC with a certain common-mode voltage. Saves you having to do other less accurate forms of DC offsetting.

I can't name a single part that runs better without decoupling caps. It's one of those things that it really is the more the merrier. More caps of different values means that when the impedance of one cap rolls off, the next size up starts to take effect. This isn't just an RF practice, you will see it in lots of high performance analog stuff. If anything it's cheap insurance to prevent his op amps from picking up stray noise in the VCC line. It can happen, even with batteries!

But this brings up another point. IIRC the gain-bandwidth of a TL084 is about 3 MHz and because he's running his amplifier at unity gain, the gain-bandwidth product is also 3 MHz. I recommend that he adds a 10-22pF capacitor in parallel with the feedback resistor (i.e. a cap from the OUT to "-" pins) to reduce his bandwidth to that of audio frequencies only (22pF is about 15 kHz which is fine for guitar methinks). The idea behind adding the input coupling cap is to add an intentional high-pass pole at maybe 10-40 Hz, again to reduce noise. The high-pass filter reduces 1/f noise and the low-pass filter (the 22pF//470k guy) blocks out higher frequency noise and RF.

[XgamerGt03]

I don't believe that is the case. If he were doing that then you would see a ground symbol at the negative side of the battery, which isn't there. The way his schematic is drawn he will get +4.5V and -4.5V on the positive and negative supply of the op amp. The positive input can just be at ground in this case.

If he drew the schematic wrong then my analysis goes out the window, but I don't think its a mistake.

I won't argue that the decoupling caps will increase the performance of the op amp, but in this application it might not make an audible difference.

[Cliff Schecht]

You can't get positive and negative voltage from a single battery without some sort of charge pump device or other method of voltage inverting. What he has there is a 1/2 VCC offset, the negative power supply pin of the op amp is at ground and the plus power pin is at 9 V.

[Merlinb]

Man, that's a lot of bitching over a couple of caps. Just use the caps, they do no harm and it'll allow you to use a mains power adapter too, if you want to.

[Super_Reverb]

You can't get positive and negative voltage from a single battery without some sort of charge pump device or other method of voltage inverting. What he has there is a 1/2 VCC offset, the negative power supply pin of the op amp is at ground and the plus power pin is at 9 V.

Not a bitchin' point, just a technical clarification:

1) The bias circuit we are talking about sets up a 50% divider yielding +/- 4.5V.
2) The "center tap" of the divider is connected to ground, creating a bipolar supply of +/- 4.5V RELATIVE TO GROUND.
The input voltage is referenced to ground (non inv input = 0V) and the output voltage is referenced to ground.
3) The opamp voltage can swing up near 4.5V rail and swing down near -4.5V rail
3.5) If you were introducing a 4.5V offset into the signal path, the output voltage would be pegged to one of the rails.
4) Opamp has no input offset introduced, and doesn't need to due to the bipolar supply voltage.
5) There is no DC offset in signal path.

If there is no DC offset in output voltage, why do you need to use a coupling cap? Do you need to AC couple an AC signal with no offset is the question, no?

Boutique tube amps are partly art: if you feel better using coupling caps or if you perceive the amp sounds better with them, go for it. A good tech would measure output offset voltage and add a coupling cap if necessary. But, understand why you are adding components to a design and that you are introducing additional effects

PSRR isn't really relevant here as the remainder of the tube amp is using an unregulated power supply that varies considerably with line voltage and load current (sag). So, with a fixed input voltage, the amp's output voltage is modulated considerably with supply voltage changes. PSRR is more important in *precision* analog circuits like A/D convertors and instrumentation amps. Guitar amps are more about aural performance and mojo, not precision.

My intention is not to be argumentative, but to clarify this point. Trying to push the art and science toward knowledge and analysis and away from cork-sniffing.

Cheers,

rob

[Merlinb]

A good tech would measure output offset voltage and add a coupling cap if necessary. But, understand why you are adding components to a design and that you are introducing additional effects

As long as the caps are large (low reactance down to ~1Hz), there are unlikely to be any detrimental effects in an application like this. 'Good techs' freak out about components too much.

[Cliff Schecht]

Sorry to be argumentative, I went and did some "sniffing" through some texts and did finally come across the single-battery DC-coupled op amp circuit. Funny that I've been doing analog for years and have never seen this circuit implemented in a real-world situation! I will gladly admit that I was wrong about the 1/2 supply thing and I guess it's good that I learned something new.

I'm not trying to be annoying about the input and output caps, as well as the low-pass pole, but I bet you dollars to donuts that he would notice a decrease in noise by adding the appropriate amount of out-of-band rolloff. Even if this unit works fine with DC-coupled, the bandwidth is too high and modern high gain amps will amplify the excessive noise. But we are splitting hairs at this point because he's not even running into an amp, he's running into a noisy as shit Echoplex!

[XgamerGt03]

I actually had to go back and look and see if it were possible to create the dual supply. I knew I had done it before, but its been awhile since I did that.

I do agree that a cap should be placed across the feedback resistor to decrease the bandwidth though. No need to allow it to amplify signals that he doesn't want.

As far as the coupling caps, I was just saying they aren't necessary for DC. They may be useful in other operations, but as there this circuit specifically doesn't require it.