Unbalanced Passive Summing Mixer

[Rylan]

I have a friend who wants me to build an unbalanced passive mixer for his modular synth, I told him it is better to go with an active mixer but he is pretty dead set on a passive mixer with a tube gain circuit to make up the gain on the output.

I have done some reading around on other threads and forums on this, and it seems if you have a resistor after each channel fader and the resistor is a good bit higher impedance than the fader there will be less interaction between controls from other channels..

So what I am trying to figure out is if I use a T-Pad resistance calculator to get a value for the parallel resistor from (r1+u1) / 4 would this circuit be about right assuming 2 3 and 4 are the same values, and also any suggestions for where to start with resistance values if it is correct? Not sure what his output impedance tends to be on this thing but I am guessing line level? (PS U1 through U4 are supposed to be potentiometers)

Thanks in advance, I am not %100 sure that the parallel resistor should be there?

 

[Joechris]

I dont think you should wire the pots that way, you will short the input at low level and upsett the conected equipment. Swap 2 and 3.
Look at Ians tube mixer. He use 47k summing resistors and a 100k pot as R5 as master level. Keep the U1-4 level pots low. try 10k or lower, or as low
as your other equipment will drive without starving.

 

[Rylan]

ehhh? in through 3 and out through paddle? interesting I am not familiar with any schematics that connect that way, so you are saying not to give potential to hot or ground from the paddle, but instead give potential across the entire resistor and drain it through the paddle? I suppose ohms law would dictate that the current should flow to ground and not the paddle if it is at 0 resistance setting, but this is just different than I am used to doing this.. I have for about 20 years done guitar electronics and such things which have passive summing and nearly always wire hot to paddle, ground out one side and output to the other. PS I tested the schematic as I posted it and it actually works great, but I will check out your suggestion Joechris, to see if it makes any difference. so far I have only tried summing passive outputs, maybe your advice works more with active outputs? anyway thanks for the feedback

 

[PRR]

> ehhh? in through 3 and out through paddle? interesting I am not familiar with any schematics that connect that way, ... this is just different than I am used to doing this..

Sources like guitar pickups do not burn-up or distort when shorted.

Sources like synth outputs WILL distort if shorted, or if loaded in very-low resistance, and a few older ones "could" burn-up.

> about 20 years done guitar electronics ...and nearly always wire hot to paddle

Compare Fender mixers: the older 5A3 and the later 5F6A. (See below.)

When sources are very high impedance (pickups, pentode, triodes with 250K plate resistors), wiper-in works OK.

When sources are low impedance (triodes with 100K, and essentially all solid-state boxes) then top-in wiper-out works a lot better. (Also avoids the over-size coupling cap the 5A3 needs to maintain some bass at low pot settings.)

 

[PRR]

Either way can work for you but you MUST build-out modern amplifier outputs with at least 10K between source and wiper.

R5 is not needed.

 

[Rylan]

Ah thanks PBR! That is what I was misunderstanding. I spent so many years just reconnecting wires and redoing the same circuits I have to un-learn a few things! Really appreciate the detailed feedback. I also have noticed in some other schematics that R5 is not needed, it seemed to basically just lower output when increased in my test today when I replaced it with a variable (pot) resistor.

 

[Rylan]

Hopefully this thread can help someone else along the way as well. I have been reading a lot of posts from these forums and they are very informative! I will post the revised schematic so this is not a dead end for someone googling it.

 

[ruffrecords]

All the schematics posted so far have some fairly serious short comings. Apart from the wiper drive from low source impedance problems the way they are currently configured will lead to serious interaction between the controls. There a a few basic rules for successful passive summing.

1. All the sources need to present the same, or a guaranteed very low, source impedance.

2. The mixing pot should be several times (preferably at least 10 times) the the input source impedance.

3. The bus mix resistors should be several times the value of the mixing pots.

4. The inputs are connected across the pots and the mix resistors to their wipers.

1 and 2 together ensure that bus impedance variations, and hence control interactions, are minimised. 3 determines the degree to which any bus impedance variation actually affects the output level and the degree to which the bus resistors affect the law of the pots (although you can use this to advantage to deliberately alter the law of linear pots).

As an example, suppose you could guarantee all the source impedances are 1K or less then you could use 10K pots and (say) 47K bus resistors. The with one input fully up, the effect on output level of changing any other control from min to max would be:

Worst case pot source impedance = (10K + 1K)/4 = 2K75

With 10 inputs, nominal bus impedance is 47K/9 = 5K22

Nominal bus attenuation = 1/10 = 20dB

Bus impedance change due 100% change in one control is 47K/8 in parallel with 47K +2K75 = 5K875 // 49.75 = 5K2545

Bus attenuation is 5.2545/(5.2545 + 47) = 19.95dB i.e a change of 0.05dB.

Personally I would not recommend mixing directly from pots unless you can be certain of the source impedances (inside a mixer you can) and you need to make certain the inputs are grounded when no source is plugged in.

Cheers

Ian

 

[Rylan]

Thanks Ian!

Very detailed information.

As for source impedance, all of the inputs will be fed from various parts of a modular synthesizer, have to measure it still so I didn't put in any values yet.

So I should use normalling jacks then for inputs? Also what would be the other option rather than mixing directly from the pots, input traffos?

 

[PRR]

You are quite right, but I'll argue another side of the coin.....

> some fairly serious short comings. ...serious interaction between the controls. There a a few basic rules for successful passive summing.

Remember the design brief was: "passive mixer for his modular synth, ... he is pretty dead set on a passive mixer"

The plans above will "sum sucessfully".

Yes, interaction is an issue.

However there are different demands for a "Console" and "instrument mixer".

In mixdown console, trimming the cowbell should not affect the voice or piano one little bit. In PA mixing we may run mighty close to acoustic feedback, can't stand much unintended consequence.

OTOH: In performance mixing without mikes, interaction may be tolerated, may even be useful (turn-down the tri-wave, the sine comes up a bit, "natural"). If it is a "set and forget" mixer then interaction is just a few more diddles.

In fact that 5A3 DeLuxe is "loved" for the heavy interaction between Vol1 Vol2 and Tone (and bass-cut in coupling-cap). Not by design but by accident.

> Bus ...change due 100% change in one control is.. a change of 0.05dB.

A wonderful spec, but 2db is entirely tolerable for performance work and 6db can be tolerated for many situations.

> unless you can be certain of the source impedances

"Modular synth", to me, implies 1K outputs. "Tube makeup stage" could well be a 1Meg input (although we must allow hundreds of pFd too). 10K or 100K pots and mix-resistors fits the 10:1 rule of thumb on both ends. Using 10K pots and 100K mix resistors makes un-used input loading moot (all for few-db precision, not a $50K mastering desk).

Whether the interaction is benign, euphonic, or annoying, depends on details of design and user attitude. Since the designer here is not broadly experienced (and the client must suspect that), and the user's expectations are not well-defined, and the cost of parts is less than a six of beer, what the heck. Try it.

 

[PRR]

I would try sumthin like this. 2 to 6 inputs (more if gain does not come up short). Mod-synths can put out DC, so input caps are a good idea. Tube is run very hot so it does not suck into modern line inputs.

 

[ruffrecords]

Hi PRR,

I agree, the other side of the coin is just as interesting. I was only trying to set out the basic principles, to give people the tools to decide for themselves where to make compromises, and also to dispel the myth that passive mixing inevitably means unacceptable interaction.

I think your simple tube design would work well. Values look good to me for an assumed 1K source impedance. I would be tempted to change the CC stage to an SRPP to be certain of driving bridging loads without having to run the tube so hot but that's just a personal design choice.

Cheers

Ian