Simple mixing techniques for 40106/op amp oscillators

Werebear · Joined: Nov 23, 2009 Posts: 63 Location: UK

Not been about for a bit and haven't been up to much but have developed an inability to passive mix strings of oscillators. A long standing project has been the construction of a shruti box synth - a series of droning oscillators to create chords that can be played and dare I say sung against. The problem is that all the mixing techniques I have used for a variety of things fail when mixing simple oscillators such as 40106 and op amps I get unpleasant noise instead of nice chords.

I've tried passive mixes with 100K resistors, caps before and after, resistors to ground...

Am I just unlucky? Do I need to buffer every output and create an active mix for what a million(!) kids on YouTube achieve with a handful of resistors?

Dougster · Joined: Sep 20, 2005 Posts: 272 Location: Tucson, AZ, USA

Would you draw a picture of what you're doing? (You could try XCircuit if you don't have any other drawing tools...)

Are you really opposed to using a few op amps?

Regards,
Doug

JovianPyx · Posted: Thu Dec 13, 2012 9:40 am Post subject:

Ditto what Dougster said about a schematic. In electronics, a picture is worth a million words, not a thousand as the saying goes.

And I agree with the comment about opamps. There are indeed problems with attempting to mix signals passively (and I assume that what you mean by passive is simply connecting the signals together with resistors but without any amplification). Google things like "audio mixer", "DC mixer", "AC mixer" and you'll see some schematics for basic mixers - these circuits are simple, noncritical and are easy to build. OpAmps help you out immensely by converting the high impedance inputs (100K is a fairly high impedance) to a nice low impedance output. That means that while the inputs that feed the mixer don't supply much current, the output of the opamp most certainly can. In an opamp based mixer you also have a virtual ground that prevents signals from one input interfering with others - possibly the source of your distorted sound - and even synching when you don't really want that.

You're description sounds like distortion, but it's hard to imagine where that distortion originates without a schematic.

JingleJoe · Posted: Thu Dec 13, 2012 9:41 am Post subject:

I want to know what this guy is doing too, how can you fuck up a passive mixer? Laughing

Maybe the problem exists further back in the circuit?

Werebear · Joined: Nov 23, 2009 Posts: 63 Location: UK

Well it would appear that passive mixers fuck up more regularly than work.

I have one 40106 making three oscillators the unused gates tied to ground as they should be. Using 100k resistors the oscillators are taken to a single mix point then to a bench amp I've been using for years but also to a hifi amp with the same results - chaos and distortion. Each oscillator sounds fine on its own but the minute I try and mix two or three together crapola. This has also happened with other simple oscillators designs such as op amp square tri ones ala SH09 LFO brought up to audio.

My power supply is a two 9V battery with voltage regs to give me + and -5V with a fairly solid ground.

However my researches has told me I need to give up on the passive mix and use virtual ground mixing.

Will get back to you when I've anything else to show. Thanks

audmix-f4.gif

Description:

Filesize:

3.91 KB

Viewed:

229 Time(s)

This image has been reduced to fit the page. Click on it to enlarge.

JingleJoe · Posted: Fri Dec 14, 2012 12:02 pm Post subject:

JovianPyx · Posted: Fri Dec 14, 2012 12:09 pm Post subject:

Joe, the virtual ground he refers to isn't the bullsh!t thing so many people misuse - he's referring to the mixing node, that is, where all of the input resistors meet with the feedback resistor. In his drawing it is labeled "mix point". Notice that it is marked zero volts - for good reason, because it is, in fact, a virtual ground. Go back to your opamp theory and you'll see that the reason that point is zero volts is because the noninverting input is at ground potential. In an inverting amplifier like the schematic shows, the output will do what it has to in order to force the difference between the + and - inputs to zero.

The bullsh!t thing to which I refer is the use of a voltage divider (usually for 1/2 the single supply voltage) with single supply systems and the misuse is trying to have it sink or source too much current. Truly, this kind of virtual ground does work - under the right conditions - that being don't try to dump current into or out of it - rather, use it as a voltage reference only. Note that this kind of virtual ground can be bolstered by feeding it into a voltage follower opamp circuit - in this case, it can source or sink a few milliamps, but again, don't push it. This kind of virtual ground can be used effectively in small low current circuits. If too much current is required, then you will see the virtual ground flop around and change value - which is a source of noise (and not the good kind).

In the end, I would always recommend a real ground with a dual supply. It is far more stable than virtual grounds I described in the second paragraph above. I know a lot of people like to use single supply with a virtual ground, especially for battery circuits, but they are often more trouble than they are worth and will turn your project into crap unless you are very kind to it.

Also, the gain of the mixer circuit is unity for each input, not 3. Yes, if all the inputs are driven to the same voltage, the output will be 3x one input. However, what happens if only one input is say, 1 volt and the rest are zero? In that case, you can see that the gain is unity.

EDIT ADD:
I should also have written that each input defines it's own gain. If you have 2 inputs, one with a 10K resistor and the other with 100K and a feedback resistor of 10K, the 10K input will have unity voltage gain and the 100K input will have 10x voltage gain. So a mixer doesn't have a gain, it has several, even if they are all unity. The 3 comes from the fact that (in the schematic), there are 3 unity gain inputs that are summed. It's important to understand that the 3 doesn't come from multiplication, it is a sum and as such, it is not considered gain.

Werebear · Joined: Nov 23, 2009 Posts: 63 Location: UK

Hi

the image that I put up is what I'm going to use to mix in future - haven't had chance to actually try it yet but now I'm aware of the topology I suddenly see it every where! My understanding of the virtual earth mixer is that it mixes current as opposed to voltage as the mix point is effectively at ground.

Everything going at the moment is on breadboard but I do have a fantastic history of boarding fabby little things that then turn to crap when soldered up.

Thanks fror the input.
Watch this space.

blue hell · Posted: Sat Dec 15, 2012 5:23 am Post subject:

Werebear · Joined: Nov 23, 2009 Posts: 63 Location: UK

I think the other problem I'm up against is the output from a 40106. I'm using 10V (+5V and -5V), so that means my signal output is going to be pretty close to 10V minus whatever the hysteresis loss is (have I got that right?). My op amps are also on the +/-5V PSU so there's no headroom. I need to drop my output level as well I think using a voltage divider to give the op amps a bit of head room. Or strap my op amps to the +/-9V straight from the batteries with appropriate caps etc.

At some point I will get the time to do this instead of just talking about it...

JovianPyx · Posted: Sun Dec 16, 2012 8:05 am Post subject:

Oh yes, headroom could be an issue.

I believe that you can do this by simply increasing the input resistor values by a factor of at least 3. This will reduce the current into the summing node by a factor of at least 3. Just for grins, try 47K or 33K instead of 10K since these are probably something you have laying around. 33K will probably work nicely and give you about the best amplitude you can expect for standard value resistors.

Tony Deff · Joined: May 25, 2008 Posts: 51 Location: Suffolk, UK

Werebear · Joined: Nov 23, 2009 Posts: 63 Location: UK

Hi Tony

wow thanks. That's a lot of info. I haven't had chance to get back to this since posting - too much real world intrusion from my kids and grandkids - so I've no more info at the mo.

Unused gates are tied to -ve on my +/-5V, but I'm going to use them as buffers as you suggest. At some point I'm going to use 4040's as octave dividers and that will also help with the buffering.

The diagram I posted is what I intend to use as a mixer but haven't had chance to try.

Thanks for the input

Jim

Werebear · Joined: Nov 23, 2009 Posts: 63 Location: UK

Finally had time today to get back to this and it works!

Took everything on board and:

Using +ve and ground to power my 40106 and three 4040's (for octave switching);

Used unused gates on the 40106 as buffers and then straight into the 4040;

Used three 100k resisters to mix and it works wonderfully.

So that's half the voltage and a bit of buffering - from the 4040 as well as I always felt the output of these beasties was quite "hot" compared to straight 40106 - and is still very "loud" as I expect CMOS to be.

Will still investigate the op amp mixer at some stage as I want some control over individual oscillators and I will keep my dual supply for that and also because I want to add a filter to this at some stage.

Thanks to everyone who commented.

Cheers

Jim