Correct input current bias resistor for single-supply inverting opamp mixer

[Gordito]

I'm building a 4 input, variable gain (1 to 14), 9V single supply, inverting opamp mixer with a TL074. At the moment i have prototyped one channel with only one signal input. Attached is a quick schematic showing one channel (mono). The real circuit will have dual ganged pots to control the stereo signals. The output will go into a Baxandall tone control opamp stage.

I was having a problem with a DC offset pushing the output signal into the + rail when gain was increased, until i RTFM'd and inserted a series resistor (R8) between the non-inverting input and the 100K || 100K voltage divider. I used 6.8K. I calculated using my one input resistor (7.5K) in parallel with POT5 halfway (50K) + the feedback series resistor (7.5K).
(7.5k * 57.5K) / (7.5K + 57.5K) = 6.63K.

The literature (TI SLOA58 and other) indicates the value for this non-inverting input resistor (in order to correct the input bias current) should be the input resistors (R1 to R4) and the feedback resistor in parallel (R1 || R2 || R3 || R4 || (R5 + POT5) in my case).

I am having difficulty understanding how to determine the best 'one size fits all' value for the input bias current resistor (non-inverting input) because:

A. I can plug from 1 to 4 inputs (so the R2, R3, R4 terms can be optional)

B. The feedback resistor value changes due to the potentiometer.

Do i calculate the minimum and maximum values and do an average (and cross my fingers?) or use the minimum or maximum value calculated? Or does this complicate things enough to justify having separate summing and gain stages?

Any help is appreciated! Thanks.

 

[Roff]

I think your problem is that you have the polarities of C1 thru C4 reversed. Normally no current will flow through R1 - R4. That is the purpose of C1 - C4. They are DC blocking caps (unless you have the polarities wrong).
The bias current of TL074 is basically zero. You don't need R8. If its value causes the op amp's output DC to change, you have other problems besides C1 - C4.

 

[Gordito]

Hi Roff,

Thanks for your suggestions. I have verified the schematic and actual circuit and i am certain the C1 to C4 caps are well oriented: the positive side towards the input jack (tip) and negative side towards the resistor / inverting opamp input. All the schematics I've seen (just verified as well) have the caps this way.

I've attached a pic from the TI SLOA058 doc referring to the input bias current resistor (R8). Note that the document has an error, the Vout formula should refer to R1A, R1B and R1C rather than R1, R2, R3! I presume it's a copy paste error...

This resistor did correct the problem but my understanding of how i need to determine the value when having optional inputs and variable gain is problematic.
View attachment 68139

 

[Roff]

Hi Roff,

Thanks for your suggestions. I have verified the schematic and actual circuit and i am certain the C1 to C4 caps are well oriented: the positive side towards the input jack (tip) and negative side towards the resistor / inverting opamp input. All the schematics I've seen (just verified as well) have the caps this way.

The caps have to have the positive terminals connected to the side of the circuit that has the most positive voltage. In your circuit, due to feedback, the inverting input of the op amp will be at the same voltage as the positive input, which is Vcc/2 (+4.5V). Thus, the positive cap terminals MUST be on the right-hand side. The way you have them, they will pass DC current, screwing up the op amp bias.
I'm not telling you this because I saw it in somebody's schematic. I'm telling you this because I am a retired EE with 50 years of successful analog (and digital) circuit design experience.
Please post links to the schematics you used for verification of your opinion.

And once again - you don't need R8 (assuming you are actually using TL074).

 

[Gordito]

Thank you for taking the time to explain why they are reversed. That totally makes sense (the feedback circuit will keep the 'right side' at a higher voltage). Teach a man to fish...

I will remove R8 and switch the caps around and post a followup.

As for why i thought they were ok (and why they got me confused)
http://www.all-electric.com/schematic/simp_mix.htm

http://ebookbrowse.com/forrest-mims-engineer-s-mini-notebook-op-amp-ic-circuits-radio-shack-electronics-pdf-d355983801 Page 10 - on the 'single supply amplifier' schematic the negative side is towards the opamp BUT the value is 0.47uf so it wouldn't be polarized - just noticed that... But the 'Audio amplifier' has a 1uf cap with the - towards the opamp.

http://www.learnabout-electronics.org/Amplifiers/amplifiers66.php 3. audio mixer: Although there is no explicit '+' marking on the symbol the black band is towards the opamp... this too made me assume the polarity was ok.

**broken link removed**

 

[audioguru]

Your circuit has a single +9V supply so the inputs of your opamp are at +4.5V.
But the circuits you copied have a dual polarity supply so their opamp inputs are at 0V, the same as the volume controls. Then their polarized input capacitors can have any polarity.

I try not to use polarized electrolytic capacitors to couple audio signals when there is no DC voltage, I use non-polarized film capacitors. A few times I have used a 1uF film capacitor which is small and fairly inexpensive nowadays.

 

[Gordito]

I have reversed the cap and removed R8 and everything is fine. Thanks again Roff and Audioguru.

Is higher ESR or leakage the reason you try not to use electrolytic caps for AC coupling?

 

[audioguru]

Low frequencies cause a plus and minus voltage to appear across a coupling capacitor when it has no bias voltage. An electrolytic capacitor conducts (as you saw in your circuit) when its voltage is reversed which rectifies the signal. You do not want audio to be rectified unless you like acid rock "music".

 

[Roff]

Deleted because I had duplicated the content of Audioguru's post.

 

[Gordito]

I have a some 3.3 uf MLCC non polarized caps on hand. Is this equivalent to the film caps suggested for AC coupling?

And i suppose that they would also do for coupling between opamp stages (if required) and decoupling to an audio line out. Is this incorrect?

 

[audioguru]

I design my circuits so that the coupling capacitor value passes low frequencies (20Hz) into a fairly high input impedance (24k ohms). Then the capacitor is 330nF (0.33uF).

Why does your circuit have such a low input impedance that a very high value of 3.3uF is needed?

I don't know what MLCC means.

 

[Gordito]

I have only 100K pots on hand, and for a gain of about 15 i need to have 7.5k for signal input into the summing circuit. I'd rather use these 100k than order others but if need be...

I've heard it said that higher than 100K resistors in the feedback can cause noise. Is this still true in the case of the TL074 or is that some prehistoric (opamp 741) design tip?

In your designs, do you prefer a 24k input impedance solely with regards to how it affects the cap (permits non polarized, smaller value) or is there some other benefit (maybe lower general current for battery life)?

Sorry, MLCC is multi layer ceramic disk. I have these 3.3uf.
(part FK24X7R1E335K - here is the datasheet: https://www.tdk.co.jp/tefe02/e4941_fk.pdf)

 

[audioguru]

The volume controls should have an audio taper (logarithmic). A value that is 100k should be used with a summer input impedance of 100k or more.
You will not hear high value resistor noise on this line-level circuit. On a microphone circuit you should use low input resistor values. The low input resistor value shorts noise to ground from the feedback resistor.

I don't see how the value of an input resistor affects battery life.

Ceramic capacitors should not be used for audio coupling because their value changes as the voltage changes which causes even harmonics distortion at low frequencies.
Also they are microphonic and pickup background sounds and might cause acoustical feedback.

 

[Gordito]

ok, i'll use higher values for input resistors (and pots), which permits smaller values for cap (film). Thanks for the help.

 

[Gordito]

After rereading the last response i have a few questions (rather basic).

Lets suppose i still use a 10k audio pot for signal input attenuation, followed by a 220nf coupling cap (film type), followed by a 33k summing input resistor. The passed frequency is 22hz. The faintest signals i receive from are around 75mv (mp3 player spkr out) up to line level, requiring a variable gain of 1 to 15.

1) Is there a reason 100k would be better than 10K for the input pot? I fail to see one... i do not understand the benefit while at 10K it could short some noise as you say. Does it have something to do with loading the source AC signal?

2)The reason we want an audio taper is because voltage to sound power ratio not linear, or is it because of something else happening in the circuit?

3) The feedback gain pot should also be audio type rather than linear, since the gain also affects the sound power eventually?

Thanks

 

[audioguru]

1) Is there a reason 100k would be better than 10K for the input pot? I fail to see one... i do not understand the benefit while at 10K it could short some noise as you say. Does it have something to do with loading the source AC signal?

Almost any signal source can drive a 10k pot parallel with a 10k summing resistor.
Since you wanted to use a 100k pot then i said the summing resistor should also be 100k (so that level adjustment is smooth).

2)The reason we want an audio taper is because voltage to sound power ratio not linear, or is it because of something else happening in the circuit?

Yes, our hearing's sensitivity to loudness is logarithmic so that we can hear a pin dropped or also hear a nearby jet airplane.
1/10th of the signal level sounds half as loud. 10 times the signal level sounds twice as loud. So when a volume control is turned down to half then the signal level should be 1/10th.

3) The feedback gain pot should also be audio type rather than linear, since the gain also affects the sound power eventually?

If you have a logarithmic volume control then why add a gain pot? Usually the gain is designed into an amplifier or a preset gain pot can be adjusted only one time.

 

[Gordito]

Since you wanted to use a 100k pot then i said the summing resistor should also be 100k (so that level adjustment is smooth).

What exactly makes it smoother?

If you have a logarithmic volume control then why add a gain pot? Usually the gain is designed into an amplifier or a preset gain pot can be adjusted only one time.

Ah. good point. If i understand your meaning i should use a fixed gain to account for the lowest voltage signal and just use the input pots to control all the signals via attenuation.

Thanks for the suggestion, I'll breadboard that and see how that works.

 

[audioguru]

What exactly makes it smoother?

If you load a volume control pot down with a load resistance too low then the volume is too low. You turn up the volume but nothing much happens so you turn it up more and more and more then suddenly the volume jumps up when the pot is turned to maximum.

If i understand your meaning i should use a fixed gain to account for the lowest voltage signal and just use the input pots to control all the signals via attenuation.

Correct.