Summing Mixer Cct self oscillating - ANSWERED

[ZeroApex]

Built a summing mixer using LM358 op-amp at its heart (see attached cct diag) for an audio application. It works as expected until the volume is increased. At roughly half volume the circuit starts to oscillate with sawtooth waves that vary in amplitude in a sinusoidal fashion. Any ideas on what might cause it and what could be done to suppress it?

EDIT - the sawtooth wave that I previously mentioned could very well be a sine wave as my sampling rate is quite low

 

[rjenkinsgb]

Do you have plenty of decoupling on the supply and bias divider?
And is the output well separated physically from the input?

The double inversion means the output is in phase with the input so any coupling between the two, even a current path in the output circuit shared with the input like a thin ground connection, could cause positive feedback.

Edit: The second opamp is unity gain; you could reconnect that as a non-inverting buffer; signal to + and link - to output.

 

[ZeroApex]

Thanks rjenkinsgb

No, I have no decoupling. I omitted it as the power source is a battery.

However, on reading your post, that sounds like a script for an electronics horror movie. :/ Would the addition of caps (see diag) do the trick, or am I over-simplifying things?

PS. The second op-amp isn't really needed, it just puts everything back in phase. With a normal audio mixer this would be absolutely necessary, but since this application is self contained up to the output the second inverting amp can be done away with altogether.

 

[Nigel Goodwin]

Thanks rjenkinsgb

No, I have no decoupling. I omitted it as the power source is a battery.

Decoupling isn't optional, it's an absolutely vital part of almost every circuit - and particularly so when you're trying to generate a split supply from a single one.

You don't 'need' the two shown in your diagram (although the second won't do any harm), but you need one from centre to either of the main rails (and I'd prefer an electrolytic across it as well), and one across the main rails - also an electrolytic across the mains rails (say 100uF).

However, on reading your post, that sounds like a script for an electronics horror movie. :/ Would the addition of caps (see diag) do the trick, or am I over-simplifying things?

PS. The second op-amp isn't really needed, it just puts everything back in phase. With a normal audio mixer this would be absolutely necessary, but since this application is self contained up to the output the second inverting amp can be done away with altogether.

You've got no gain in the circuit, so oscillation shouldn't be an issue, other than leaving out vital capacitors, and possible layout issues?.

 

[rjenkinsgb]

I'd put the caps separately between ground & power, and ground to VE.
And use both a small value film cap and a decent size electrolytic.

Note that audio phase is not normally important, other that when splitting and recombining the same signals somehow, then both or all paths need the same inverted or non-inverted path.

Think about it - the wavelength of eg. a 1KHz note is only about a foot / 300mm in round figures, so the phase inverts every six inches or 150mm distance from the mic or speaker.

 

[Nigel Goodwin]

I'd put the caps separately between ground & power, and ground to VE.
And use both a small value film cap and a decent size electrolytic.

Note that audio phase is not normally important, other that when splitting and recombining the same signals somehow, then both or all paths need the same inverted or non-inverted path.

It is in an audio mixer though (and many even come with phase inverting switches on the channels for that reason), as any phase inverted signals will cancel out in the mixing process.

He's not mentioned what he's trying to do?, but it doesn't look like absolute phase would make any difference?.

I was trying to set up for a gig once - and really struggled with the speakers and leads - I don't know who had been messing with them?. In the end I nicked the PP3 out of an effects pedal to check the phases - turn out left and right were out of phase, and so were various speakers in each stack.

 

[rjenkinsgb]

I was trying to set up for a gig once - and really struggled with the speakers and leads - I don't know who had been messing with them?. In the end I nicked the PP3 out of an effects pedal to check the phases - turn out left and right were out of phase, and so were various speakers in each stack.

Isn't that that's the exact "splitting and recombining" scenario I mentioned, where phase is important?

 

[ZeroApex]

Not to get side-tracked, but phase can be important if an audio signal is split along two or more different paths. If the output of the first op-amp were recombined with one of the other signals at a later stage it would cancel out rather than mix if the phase is not correct. For this application phase is not important, I just included it as part of the learning process. So, no need for any further discussion on phase

I will burn in some decoupling first, hopefully that will improve things.

Layout issues could be an issue as it's an early prototype built on a little piece of veroboard (30mm x 75mm), but decoupling comes first.

Thanks for your assistance guys

 

[Nigel Goodwin]

Not to get side-tracked, but phase can be important if an audio signal is split along two or more different paths. If the output of the first op-amp were recombined with one of the other signals at a later stage it would cancel out rather than mix if the phase is not correct. For this application phase is not important, I just included it as part of the learning process. So, no need for any further discussion on phase

I will burn in some decoupling first, hopefully that will improve things.

Layout issues could be an issue as it's an early prototype built on a little piece of veroboard (30mm x 75mm), but decoupling comes first.

As long as you build it with a little thought veroboard is fine - I built an entire 12 channel mixer for a band on veroboard.

 

[ZeroApex]

Well, the caps are in place as you both mentioned (VE to either rail & +Vcc to -Vcc). Just used closest values available from my little kit box

The result: noiseless beautiful volume that is usable from zero to maximum.

Many thanks. I really appreciate your help. And a lesson learned

The application that I never mentioned? A new design of guitar pickups that I've been working on for quite a while now. The technology is nothing new, its the mechanical side of things and the ways that they can be used that are novel. A few more years of development left, but that's for the engineers. I'm just happy to have a working prototype of phase one of five.

Yay

 

[ZeroApex]

As long as you build it with a little thought veroboard is fine - I built an entire 12 channel mixer for a band on veroboard.

Yes. Physical limitations (like bulky electros) mess with perfection in tight spaces, but once I worked out that a horizontal bus can easily be built across the vertical busses using an upside-down sliver of veroboard everything was so much easier (and neater). Thankfully, I'm not building an entire console.

I cannot get over how quiet the cct it is. Audiophiles bag the hell out of the 358, but it's quiet enough for me. Saves bucks on an OPA that costs 6 times as much.

 

[rjenkinsgb]

You could try a TL072, also very cheap and a pin-for-pin replacement, but specifically a low noise type?

 

[rjenkinsgb]

A new design of guitar pickups that I've been working on for quite a while now. The technology is nothing new, its the mechanical side of things and the ways that they can be used that are novel.

A touch of deja vu there

This one is around 30 years old, though apparently yours is for a different function:

 

[Nigel Goodwin]

I cannot get over how quiet the cct it is. Audiophiles bag the hell out of the 358, but it's quiet enough for me. Saves bucks on an OPA that costs 6 times as much.

It should be quiet, you've got no gain

My band mixer used all 741's, and had to accept microphone inputs - noise levels were perfectly fine - it's all a question of keeping gains and signal levels within acceptable parameters.

 

[audioguru]

Most guitar pickups are designed to drive the 1M ohms input of a vacuum tube or Jfet, but your circuit has an input of only 10k ohms, 100 times less. It causes your output to have a very low level that has its high frequencies muffled.

The TL072 dual audio opamp has low noise, very low distortion and a wide frequency response and slew rate. An electric guitar probably needs to have these nasty sounds.

Price? At Digikey today an LM358 costs $0.60US and a TL072 costs $1.08US.

 

[ZeroApex]

A touch of deja vu there

Early Roland synth-axe?

PS. The intonation on that guitar looks awful!!

 

[rjenkinsgb]

Early Roland synth-axe?

PS. The intonation on that guitar looks awful!!

Almost - Casio MG series.

And the intonation is perfect at that; note that I have tapewound strings on it.

I collect and repair / rebuild guitars as a hobby, though that is the only one I ever bought new.
A few of my others on here:

Guitars

Some of my guitars & gear

www.flickr.com

 

[ZeroApex]

Yeah, intonation is always fine once it's set. It just looks odd sometimes when intuitively the saddle positions should all be pretty close.

Nice collection of axes.

Anyway, we're way off topic here so time to close the thread

 

[rjenkinsgb]

I find thread drift is more often the norm than the exception on this site! But usually in interesting ways.