DIY summing mixer schematic help

[gabe real]

ok, so i have been searching around the web. found lots of info.
from what i gathered, i came up with this.

website references-
http://www.gearslutz.com/board/so-much- ... -done.html
http://en.wikipedia.org/wiki/Switch
http://www.prodigy-pro.com/diy/index.php?topic=26482.0
http://www.prodigy-pro.com/diy/index.php?topic=15854.0
http://www.forsselltech.com/schematics.shtml
http://www.prodigy-pro.com/diy/index.php?topic=16731.0



now, there is talk about dropping the mono level 3db or something.
but i think this will be adjusted with a simple decrease of signal sent out
of my DAC.

i want mono channels to send to single channel EQs and compressors.
i would like switches instead of using Y adapters.

using DPDT toggles to switch the mono select on or off.
this would work by having one of the selections on the toggle terminate by soldering the pins together. so when the toggle is up, it sends to only one
buss. when toggle is down, it sends to both buss. from what i read
in some of the reference posts, this works.(if not, let me know please)
this connection is displayed by the red lines on schematic.

i would like to beable to turn a channel off if its not in use. is there
a way to use a 4 way switch? if not a will use dummy plugs to terminate channels
by soldering tip and ring togther then plug it in the channels not in use.
or is there a TRS jack that has a defeat feature when nothing is plugged in?

the schematic is just a small version of the one i want to build.
i would like a 14 channel version.

seems pretty easy.

one other thing, should i use a resistor on the output buss?
so one resistor from tip to ring?
something about crosstalk.
another 5k here too?

also the grounds or sleeve will all be connected together.
would these all go to a chassy ground?


thanks in advance, i know there is alot of controversy about summing boxes.
i want to try it out.

any help on this would be great.
i have the concept and know how to put it together without the switches.
i just need a little guidance in that area.
thanks again

[The Scum]

I clicked on a bunch of the links you gave, and couldn't figure out which one that schem came from. It doesn't make sense to me right off, and I'd like to see what the designer has to say about it. Looking at it, the "channel panned right" actually appears to be panned left (assuming that the red line doesn't make electrical contact with the black line it crosses.).

If I could find the original discussion, I could more easily clarify some of your other questions.

Is there any mention of actually using 2 switches, one to assign left, and one to assign right...with neither switch assigned, the input would be effectively muted. However, with both assigned, the input would be louder in the center than it would on either single side...that's probably the 3 dB they talk about.

(On a normal mixer, when you pan to the center, the input is usually a little quieter when centered than when hard panned. This is because in the center, it'll be coming out of two speakers, each delivering half of the sound, so each should get half as much signal)

Do all of the devices you want to connect have balanced outputs?

[gabe real]

thanks for the response.

yeah, i made the mistake of putting the pan right when it should of been pan left.

i designed the schematic from here-
http://www.forsselltech.com/schematics.shtml
its the first link under, "Summing Buss Schematics"

i modified it.
each channel would have a switch to bridge the signal to the
opposite buss, sending signal to both channels.
with the switch in the off position, it would just send the
signal to the regular path.

so, channel 1 would be sent to the first buss (left channel)
then it would have wires split to the switch to send the signal to the
second buss(right channel). so if you engage the switch it splits
the signal to both outputs. kind of like a internal "Y" adapter.
it took me some time to figure this out.
i hope it makes sense, i could be way off.
but, it seems like it could work.

this is where i got the idea from-
http://www.gearslutz.com/board/so-much- ... -done.html
(the guy, "mitchell" wrote this, i think it was the 9th post)

This works fine for stereo pairs, but if you have something like a vocal, a bass, or drum or anything that will definitely be mono, i.e. panned center, it's dumb to waste a ST pair of ins. Say you have a kick on ch1 in your DAW. You'd pan it hard left in the DAW, and now it's only going to come out the left channel (half of the first pair). But the mono switch on the ch takes the signal and runs it to the the other side of the main bus so it's now mono'd in the mast outputs, and you save an in channel.

The schematic illustrates how to wire it, but basically the important thing is that when you add the extra set of wires for the mono switches, you have to solder those BEFORE the resistors for the ch you're taking the feed off (the mono feed gets it's own set of resistors).

It's also important that when the mono sw for the ch is off, the resistors short each other out. If you use a DPDT toggle, you'd have the in signal on the bottom two terminals, the output to two addt'l resistors on the middle terminals (then to the opposite channel master bus), and then you solder a small piece of wire across the top two terminals which shorts the resistors when the switch is in stereo position. This gives you mono with the switches down and ST with them up, so you could flip the upper and lower connections if you wanted it to work the other way around.

his schematic is not at the links he posted, and the last time he logged into
gearslutz was in 2007. so i didnt even bother to contact him.

all connections from computer DAC are balanced.(lynx aurora 16)
most outboard gear is balanced, some not.(still building this up)
i just figured to make it balanced instead of non.

thanks for your time.

[mrdibs]

You are missing some very important parts in the mixer matrix.

http://www.twin-x.com/groupdiy/albums/u ... etwork.pdf

This is the only schematic you need. NYDave has designed the definitive circuit. I have built this mixer and it works really well.

Yes, you need the resistor across the output. This sets up your output impedance. Notice the formula above Rx on the schematic. You will lose upwards of 30db of gain from this. It is necessary with a passive mixer to provide makeup gain after the fact. A pair of mic preamps are what is traditionally used.

Also notice that after the 4.3K resistors on the pan switches, they go into 10K resistors on the way into the summing bus. If you leave out the panning circuit, you can bypass the 4.3k resistors completely. The 10K resistors are what limit crosstalk. The 4.3k resistors equalize the level and input impedance to both left and right when panned center.

The way I put this together was that I have it built right into my patchbay. My PT outputs normal into my mixer where 10K resistors sum the + and - sides into their respective bus wires. Odd channels sum to the left and even channels sum to the right. The last four channels of my 16 channel mixer are mono. I did this by using 14.3k in place of 10K resistors (incorporating the two resistors into one) to go to both sides.

Bus your grounds together and attach them to pin 1 of your output xlr (or sleeve of your trs connector or whatever).

[gabe real]

thanks for the reply.
the linked schematic is great, just what i was looking for!

i think i got it now. i forgot to mention the preamp part at the
end of the chain.

im a little confused on the ouput impedance formula.
i haven't done math like this since high school, which was
16 years ago. i feel like a complete noob here.

(20K/N)200 / (20K/N) - 200
WHERE N = NUMBER OF INPUTS

i want to use 14 inputs.
so, the first part of the equation would be 20k divided by 14,
then multiply by 200?
(20k/14)200 = 285.7k

second part of equation-
(20k/14) - 200 = 80k

then-
286k/80k = 3.575

so the resister across the outputs would be 3.6k on each side?
from tip to ring on the outputs?

i hope i just didnt go on a complete nood mission with my math skills.

trying to keep this detailed as possible, so the next person
will have the info easily accessible.

thanks again

[The Scum]

second part of equation-
(20k/14) - 200 = 80k

Doublecheck that. I get 1228.

[ballpein]

second part of equation-
(20k/14) - 200 = 80k

Doublecheck that. I get 1228.

ditto. And I think that formula, as it's written, actually works like this:

Rx = [ (20k*N/200) / (20K/N) ] - 200


You're working it like this:

Rx = (20k*N/200) / (20K/N - 200)

[The Scum]

ditto. And I think that formula, as it's written, actually works like this:

Rx = [ (20k*N/200) / (20K/N) ] - 200


You're working it like this:

Rx = (20k*N/200) / (20K/N - 200)

Please double check your typing:
The original is literally

Rx=(20K/N)200 / (20K/N) - 200

The question arises of where to associate that "- 200" term - is it in the demoninator, or to the right of the division?

If it's to the right:
Rx= [(20K/N)200 / (20K/N) ]- 200

We wind up with Rx = 0 for all values of N, because the (20K/N) terms cancel, leaving 200-200 = 0.

And we know that if Rx is 0, we're shorting the outputs together.

If we try leaving it in the denominator, results seem more agreeable:
Rx= [(20K/N)200] / [(20K/N) - 200]

If I get the time, I'll try to walk through the synthesis that yielded that equation...

[gabe real]

ok, i did something wrong here, doh!


ok, im lost now.
oh, do i love math

so i would multiply 20k x 14 ?

i thought "/" meant divide.
i divided 20k by 14, i got 1.43.
then i multiplied 1.43 times 200, i got 286.

so now im here-
(286 / 20k/n) - 200

286 divided by 1.43, got 200
200 - 200 is ZERO!

ok, is this a joke or am i just WAY off here!

[ballpein]

sorry, I should've kept my mouth shut instead of clouding the issue by dickering about mathematical "grammar".

Gabe, your original process was right, you just fudged something in the second part of the formula (the denominator). Like scum said, it should be 1228.

I'll keep my mouth shut now.

[gabe real]

its ok, i welcome all the help i can get.


so, in the 1228 figure.
20,000/14 = 1428.57
1428.57 - 200 = 1228.57.



(20K/14)200 / (20K/14) - 200 = rx

first part-
20,000 divided by 14 = 1428.57
1428.6 x 200 = 285,720

second part-
20,000/14 = 1428.57
1428.57 - 200 = 1228.57.

285,720 / 1228.6 = 232.557

so i would put a 233 ohm resistor on the outputs?
or is 232 the impedance output of the circuit with all 14 channels active?
or maybe just round down to a 200 ohm load.

[gabe real]

after looking at the schematic again, i would put the
230 resistor across the output.

it says below in more text that the output of the mixer is 200 ohms.
so would this include the resistor amount or is this amount without the resistor?

oh wow, did we just do all this math for nothing?

thanks again

[nclayton]

Hi. I have a few comments:

If your inputs to the mixer can drive 10K ohms then you're better off using 5K resistors than you would be using 10K resistors. For the same output impedance you'll get smaller attenuation -> smaller make-up gain -> lower noise.

I didn't fully look at all the equations above to see which versions were right or wrong. What I can tell you for sure is that the output impedance of this type network WITHOUT the final shunt is the mixing R divided by the number of inputs and then multiplied by 2 (because it's balanced, with 2 phases, each with an impedance of Rmix/#inputs). That means that if you're using 14 inputs with 10K ohm resistors the output impedance WITHOUT the final shunt resistor is 2*(10K/14) = 1428 ohms. Obviously a 3.6K shunt resistor won't bring this down to 200 ohms. What you need is (1/1428+1/Rshunt)=1/200. That means you need a 232 ohm shunt resistor.

If you use 5K mixing resistors, the output impedance of the network WITHOUT the shunt resistor will be half of what it is with 10K resistors. This means even without the shunt resistor the output impedance will already be as low as 714 ohms. To get that down to 200 you need a 277 ohm shunt.

With 5K resistors you get about 5dB less attenuation than you do with 10K resistors at the same output impedance. That may not seem like much, but since 10K isn't a challenging input impedance to drive, you might as well use it. Depending on the exact preamp you use for make-up gain, the 5dB can be added almost in its entirety to your S/N.

By the way, in a mixer like this, it's actually best to use TRS instead of XLR jacks. That way you can wire the jacks to switch so that the tip and ring connect when no jack is inserted. That keeps gain and impedance constant no matter how many inputs are being used. If you do it this way you could make a 16 or 24 input mixer out of a patch bay and patch in other stuff like effects as needed.

If you made a 16 input mixer like this with 4.7K ohm resistors and no shunt resistor at all, the output impedance would be less than 600 ohms and attenuation would only be about -24dB. You could use a good quality line amp and the whole affair would be very clean and quiet.

The shunt resistor lowers output impedance, and low source resistance usually equates to low noise, BUT whenever this is done artificially with resistors that drop level you ALWAYS degrade your S/N. You're better off finding an amp that can handle HIGHER input levels (without an input pad), and then NOT using an artificial Z reducing shunt in order to drop the signal down to mic levels.

ned

[nclayton]

Whoops, look like while I was typing that big long POS you figured it out yourself.

Ned

[nclayton]

By the way, in a mixer like this, it's actually best to use TRS instead of XLR jacks. That way you can wire the jacks to switch so that the tip and ring connect when no jack is inserted. That keeps gain and impedance constant no matter how many inputs are being used. If you do it this way you could make a 16 or 24 input mixer out of a patch bay and patch in other stuff like effects as needed.

If you made a 16 input mixer like this with 4.7K ohm resistors and no shunt resistor at all, the output impedance would be less than 600 ohms and attenuation would only be about -24dB. You could use a good quality line amp and the whole affair would be very clean and quiet.

The shunt resistor lowers output impedance, and low source resistance usually equates to low noise, BUT whenever this is done artificially with resistors that drop level you ALWAYS degrade your S/N. You're better off finding an amp that can handle HIGHER input levels (without an input pad), and then NOT using an artificial Z reducing shunt in order to drop the signal down to mic levels.

Also, all of this stuff above only applies to a mixer with inputs hard wired for left or right. If you have the option of centering or switching channels then the channel output impedance and attenuation is going to be changing a bit all the time anyway. In this case the shunt resistor helps things from popping up and down quite so much in level with assigns, so you should use it. Sorry about that.