Op-amp bias problem

[TejfolvonDanone]

Hi everybody!

I built a little tube screamer like overdrive. I've put together a veroboard prototype of the circuit and i ran into a little problem with biasing. Without the diode the transistor and the caps the output is 4V to 2V depanding on the pot's setting. At around 30k the output goes to 2V.
The -IN is at 4.2V the +IN is at 4V and both are independent of the pot's setting. I tried several op-amps but the problem stays.
I tried to calculate the possible output voltages based on the offset and bias currents of the op-amp but it's nowhere near the actual voltages.
I checked the wiring several times but haven't find any mistakes.

Schem: https://drive.google.com/file/d/0BzGRfrmHSjb4MHlCQ0hjam9lbkE/view?usp=sharing
R7 is a 1M and the gain pot is a 100k.

I just have literally no idea what's the problem and i hope someone has.

[slacker]

You probably need a big capacitor across vcc/2.

[TejfolvonDanone]

Why would that help?

[slacker]

Without it your opamp will amplify down to DC, so it's probably amplifying any offset voltage. vcc/2 also needs to appear to be ground for AC signals, a big cap does this, without one for AC signals the resistance from pin 2 is 1k + the parallel resistance of the 10k resistors making up vcc/2, put a cap there and AC signals see 1k to ground through the cap.

[Phoenix]

You'll also want a cap between R1 and Vcc/2, ground, or even Vcc (doesn't matter when it's AC coupled, use whichever is most convenient for your layout, although tying it to Vcc is not best practice as any noise present there can be coupled into the signal) to avoid DC gain from the op amp which will amplify its DC offset. You've got gain of over 100, and the offset voltage of the uA741 can be as high as +/-15mV, which could be amplified to over 1.5V of bias offset with the gain pot at max.
Adding a capacitor here also offers another opportunity for tone shaping - you can cut some bass by selecting a small capacitor if you want to.

[TejfolvonDanone]

100 gain for a 15mV signal for me is 1.5V. 4.2V +/- 1.5V = 5.7V or 2.7V.
I get 2V at around 30 gain. Which implies that the offset is around 70mV. Also the offset would vary between ICs and from the 3 i tried all had the same issue.
(2V is the minimum output for the 741.)

A cap is open circuit for DC signals so my question still remains: why would a cap solve a DC problem?

Just to be sure i tested these tips in the circuit and nothing worked.

[slacker]

Sorry, you're right a cap across vcc/2 won't solve DC problems, one in series with the 1k like Phoenix suggested should do though. It's not just the DC offset of the opamp you need to worry about it will also amplify any difference in voltage on it's inputs, that could be what's happening.

[Phoenix]

A cap is open circuit for DC signals so my question still remains: why would a cap solve a DC problem?

Because the problem could be caused by DC gain. A capacitor is open circuit for DC, so it eliminates DC gain. If the problem IS caused by DC gain, then by eliminating it, the problem is solved!

It's not only the offset voltage that could be causing problems, but also the offset and bias currents. I'll let you do the math, but your results are not impossible with the circuit as shown.

Just to be sure i tested these tips in the circuit and nothing worked.

If that's the case, I would expect that you've missed a wiring error somewhere (or you misunderstood some of our suggestions). If you've been working on this for a while, a break and coming back to it fresh later can really help. Sorry my and Ian's suggestions haven't solved your problem yet.

[karbomusic]

Why is Vcc2 being applied to both the input and the feedback loop for a non-inverting circuit? I've never done that, only need Vref (as I call it) on pin 3. Additionally, I'd have put R4 in place of R7 and let it set the input impedance. Don't worry, I'm not educated in this,  just looking at the schematic made me think it wouldn't work properly. Maybe that is the offset part you guys are discussing? And I don't know the 741 that well so was wondering.

[ashcat_lt]

Why is Vcc2 being applied to both the input and the feedback loop for a non-inverting circuit? I've never done that, only need Vref (as I call it) on pin 3. Additionally, I'd have put R4 in place of R7 and let it set the input impedance. Don't worry, I'm not educated in this,  just looking at the schematic made me think it wouldn't work properly. Maybe that is the offset part you guys are discussing? And I don't know the 741 that well so was wondering.

R4 is a pulldown to keep the cap from popping when the thing is switched in or out.  It must be on that side of the input cap, and of course R7 can't, so you can't combine them.

Connecting that feedback loop to the same reference voltage as the input is supposed to alleviate the need for the cap that would normally be on that leg to ground.  As we've seen, this still leaves the possibility of some DC gain throwing things off center.  Also, as mentioned above, without a cap bypassing the Vref divider, the AC gain of the circuit is significantly less than it otherwise might.

[Rixen]

The 741 can have 80 to 500nA of bias current flowing out the inputs. Into the 500k resistor on the non-inverting input that is 40 to 250mV of DC offset at that input. The other inverting input has 1k to ground so only a small voltage will appear there.

You could try another 500k resistor between the inverting input and the feedback network (junction R1 R2) to balance this, as the bias currents will be similar from both inputs.

good luck !

Julian
www.rixenpedals.com

[Rixen]

..and of course you could substitute an op-amp with lower bias current. At it's input a TL071 with 200pA bias current will only have 0.1mV of offset due to the bias current and 10mV (max) due to the input offset... lower noise as well

[TejfolvonDanone]

..and of course you could substitute an op-amp with lower bias current.

I always worked with TL082 and never had a problem like this. But this time i thought i wouldn't waste a dual op-amp and i have several 741s it seemed like a good idea. Also it isn't pin compatible to the 741 so i'd better get a TL081/TL071.

The 741 can have 80 to 500nA of bias current flowing out the inputs. Into the 500k resistor on the non-inverting input that is 40 to 250mV of DC offset at that input. The other inverting input has 1k to ground so only a small voltage will appear there.

You could try another 500k resistor between the inverting input and the feedback network (junction R1 R2) to balance this, as the bias currents will be similar from both inputs.

I tried to get the maximum DC offset taking all the possible causes into account. Looks like my calculations were really off. 

If that's the case, I would expect that you've missed a wiring error somewhere

I worked on it for 3 days and checked the wiring twice each day. Also i measured the resistances between each junction and everything was well in the tolerance range. I "recut" all the trace cuts even though they appeared as open circuit for the DMM. That's why i'm really sure that the wiring is good.
The Vref bypass cap is good suggestion as you two mentioned and i shall do put it in.

Thanks for all the suggestions.

Do anybody know a a real low voltage Zener? The ones i used are said to be 0.8V Zener but turns out to be a regular silicon diode.

[Rixen]

I tried to get the maximum DC offset taking all the possible causes into account. Looks like my calculations were really off. 

..well the magnitude of the bias currents is easily overlooked as it is so small, but it can become a problem with high and/or differing impedances.. I ran into this exact issue with the TL974 which otherwise is pretty much a perfect pedal op-amp. Also often overlooked is the input current noise, which, when presented to a high impedance at the op-amp inputs can generate a noise voltage far in excess of the op-amps rated voltage noise.

One day I'm going to make a table of some op-amps and their pros and cons for pedal use- we want an op-amp that can run on a flat 9v battery, has low noise, low distortion, low bias current, no latchup, low supply current consumption, output right near the rails for maximum headroom and can be sourced in DIP as well as SMT package in single, dual and quad. Missed anything ? 

[TejfolvonDanone]

Missed anything ? 

Just the unicorns, free beer, and infinite money. 

[duck_arse]

a bit late, but can I ask: why the output buffer on an opamp?

also, +1 for a cap between Vref and the 1k to inverting input.

[TejfolvonDanone]

a bit late, but can I ask: why the output buffer on an opamp?

No particular reason which can't be disproved. It seems a good idea to have an output buffer but it's more like an "I feel like it" thing. Also i was thinking about having a tone control there but i omitted so i have a minimalist design. There's also a "because i can" part there.
This started as a tube screamer clone (and it is almost) which has a transistor buffer after the op-amp.

[TejfolvonDanone]

You could try another 500k resistor between the inverting input and the feedback network (junction R1 R2)

I overlooked it but did you mean 500k between inverting and non-inverting input?

[antonis]

IMHO, you should place a DC blocking cap between IC's out and BJT's base, bias independently BJT and place a current limiting resistor (just in case..) on IC's out...

+++++ for Vref capacitor...
(try to place an IC voltage follower on Vref with a capacitor on it's out to see if Vref is well stabilized..)

[Rixen]

You could try another 500k resistor between the inverting input and the feedback network (junction R1 R2)

I overlooked it but did you mean 500k between inverting and non-inverting input?

no, place it between the inverting input and R1- so that both inputs see ~500k