GGG Mini Mixer - Sucking Tone

[welcomb]

I see a problem with the design without gain compensation, summing the channels to the non-inverting input means the loudest one signal can get with all the other channels turned down is about 1/4 volume!  Imagine that the 100K summing resistors are actually a voltage divider, if one is tied to the signal, and the other 3 are tied to Vref (which to an AC signal will behave the same as ground), that means the three 100Ks in parallel from the non inverting input to Vref look like a 33K resistor. 

Hadn't seen it that way.  The traditional inverting summing stage would be more appropriate.  Or the gain compensation will work.  Mr. Wounded Paw, if you can post your voltages I'd have a better idea what's up with your mixer, I can't see anything in the design that would clip.

Would it not then be better to use an inverting summing stage and the other half of the op-amp to invert it back to phase? In which case we don't use an opamp to hold Vref

[kriista]

Hmm. Sounds over my head, but very sensible.

[Processaurus]

Sure, you could definitely do that.  And it would be the textbook way to make a simple mixer.  Lots of valid options out on how to design a simple utility summing amplifier, and this would be a good place to share them.

This will work just like it's supposed to, though, to make a simple mix with a minimum of parts.  The amount of self noise introduced from the extra 4x gain this circuit uses vs the more regular inverting mixer will be negligible next to the noise of the instruments...

The opamp buffered Vref is good because a resistive divider would need to suck a lot of current to hold itself steady, you want it to run 10x as much current as its load, and the 4 10K pots on the buffer opamps' outputs (a low impedance source) divide to a present Vref with a 2.5K load, so you'd want a resistor divider with two 250 ohm resistors and a big old cap, that's 18mA just for the Vref.  There was another thread recently about other rail splitting techniques that was interesting.

[kriista]

So are there any slightly higher part count schematic/layouts that a not super techy person like myself could just build without having to wrestle with too much?

[welcomb]

The opamp buffered Vref is good because a resistive divider would need to suck a lot of current to hold itself steady, you want it to run 10x as much current as its load, and the 4 10K pots on the buffer opamps' outputs (a low impedance source) divide to a present Vref with a 2.5K load, so you'd want a resistor divider with two 250 ohm resistors and a big old cap, that's 18mA just for the Vref.  There was another thread recently about other rail splitting techniques that was interesting.

I don't quite get this part. Why not use bigger resistors for the divider? Didn't find much in the search.

[Processaurus]

The opamp buffered Vref is good because a resistive divider would need to suck a lot of current to hold itself steady, you want it to run 10x as much current as its load, and the 4 10K pots on the buffer opamps' outputs (a low impedance source) divide to a present Vref with a 2.5K load, so you'd want a resistor divider with two 250 ohm resistors and a big old cap, that's 18mA just for the Vref.  There was another thread recently about other rail splitting techniques that was interesting.

I don't quite get this part. Why not use bigger resistors for the divider? Didn't find much in the search.

Bigger resistors will make the reference voltage softer, and sag when there is a load on it (like when the buffer opamps swing higher or lower).  It stops being a dependable reference voltage when the rest of the circuit can push and pull it around (which here, can cause weird bleed problems or distortion, as the refence voltage would be pulled around at audio frequencies.

http://en.wikipedia.org/wiki/Voltage_divider

See "Loading Effect"

So are there any slightly higher part count schematic/layouts that a not super techy person like myself could just build without having to wrestle with too much?

Kriista, we were just discussing different ways of designing a circuit like this, not be repetitive but the schematic and layout as drawn will work, no design skills necessary.  I'm taking a fair amount of trouble to explain this project to you.  If assembling it is over your head, maybe come back to it after a few other pedal projects.

here's an updated schematic, fixed the resistor values in the last gain stage, and corrected the power pins on the quad opamp (note they were correct on the layout), and made the part numbers match Fixr's layout.  Note R13 should be 47K, not 10K as on that layout:



EDIT: Photo link was weird

[welcomb]

Just a question with using a non-inverting summing opamp, does lowering one channel then affect the other channel? For example with all channels turned to max, and turning down one channel affect the other three?

[Processaurus]

In theory, yes, because there is no virtual ground at the summing point (the main benefit of the inverting summing stage), and the signals would go back through the other channels summing resistors and interact with where the pot is positioned, as far as sinking current into either the low impedance Vref source or the low impedance channel buffer opamp output, or having some added resistance from the pot in the way because the wiper is somewhere in the middle.

In practice, no, short answer is because the summing resistors are big and the other channels pots moving around are a small change in load.

Long answer is because the summing resistors are much larger than the pots, worst case, a pot from Channel B is right in the middle of its resistance, and the Channel A signal sees it through the 100K resistor from its channel to the summing point, then another 100K resistor back to Channel B's pot, as a 5K resistor to the opamp output (Lo-Z, with ground-like current sinkability) and a 5K resistor to Vref (Lo-z, ground-like), so those combined, behave at most like a 2.5K load in the middle of the pots resistance and less as you move to either side, which with the 100K summing resistor from  channel A to the summing point, then through the 100K summing resistor to Channel B's pot wiper, is a small variation in attenuating load on Channel A's signal at the summing point, 102.5K vs 100K. Doing the voltage divider math, at most 1.395% variation is possible in the volume of one channel if you wiggle the all the other channel's volume pots around.  Since to the human ear something has to be twice as loud to be perceived as a noticeable difference, these tiny variations are not an issue.

[kriista]

So are there any slightly higher part count schematic/layouts that a not super techy person like myself could just build without having to wrestle with too much?

Kriista, we were just discussing different ways of designing a circuit like this, not be repetitive but the schematic and layout as drawn will work, no design skills necessary.  I'm taking a fair amount of trouble to explain this project to you.  If assembling it is over your head, maybe come back to it after a few other pedal projects.

I was referring to the mention that the other channels would change when adjusting one channel, making it near useless live (the reason it's being built). That and a couple of mentions previous to that that it isn't working at all.

So the last schematic you posted and the last PnP layout are working/the same?

[Processaurus]

So the last schematic you posted and the last PnP layout are working/the same?

Yours working will confirm that the layout and schematic work.  Go for it!  The offer to help you debug it still stands, if there's a problem.

[kriista]

As in, are those two without mistakes and would presumably work?

So the last schematic you posted and the last PnP layout are working/the same?

Yours working will confirm that the layout and schematic work.  Go for it!  The offer to help you debug it still stands, if there's a problem.

[fixr1984]

PNP has been built/verified twice. I have had no problems, I might go back and change
out R13 to get a little more gain but other wise its fine.

[kriista]

Perfect.

[VPIF]

I have tried to research this a bit lately, but my electronics-knowledge is still to basic to figure it out

I have a Fender Hot Rod DeLuxe:

Pre-amp output impedance:  1,5Kohm max
Power-amp input sensitivity:  200mV for clipping
Power-amp input impedance:  54Kohm

Can I buffer the pre-amp output signal with a simple buffer/line driver (suggestions are welcome), split the buffered signal in four, run it (one line per effect) through a BOSS chorus, Line6 RotoMachine and TC electronic D-two, + direct signal and use the unity gain version of this mixer to blend the signals, thus using this for a parallel effects loop?

Thanks!

[Processaurus]

I have tried to research this a bit lately, but my electronics-knowledge is still to basic to figure it out

I have a Fender Hot Rod DeLuxe:

Pre-amp output impedance:  1,5Kohm max
Power-amp input sensitivity:  200mV for clipping
Power-amp input impedance:  54Kohm

Can I buffer the pre-amp output signal with a simple buffer/line driver (suggestions are welcome), split the buffered signal in four, run it (one line per effect) through a BOSS chorus, Line6 RotoMachine and TC electronic D-two, + direct signal and use the unity gain version of this mixer to blend the signals, thus using this for a parallel effects loop?

Thanks!

Your preamp already has a low output impedance, so buffering it initially before you split it isn't necessary.  You could  probably just make a passive 4 way splitter too.

It will work how you want it to.  The simpler GGG mixer will work too, unless your guitar is getting connected directly to the mixer.

A phase switch is often desirable on a summing amplifier used for parallel processing, you could add one after the buffer opamps with a standard opamp unity gain inverting amplifier:

with R1 and Rf being equal (10K would be fine resistor values), then you'd put a phase switch in, using a SPDT switch with the common wired to the top of the channel volume pot, one lug to the buffer opamp's output, and the other lug to the new inverting amplifiers output, to choose between inverted or non inverted copies of the signal.

The green ground on the diagram would be connected to the 4.5v Vref supply.

[VPIF]

OK. Thanks for the great explanation Processaurus.

Maybe this will be a suitable small and simple circuit for me to try to make my first layout...

Suggestions as to type of inverting amplifier? I have searched the forum, but didn't find very much information (parts numbers etc.) I am not really familiar with the concept of inverting/non-inverting op-amps, but I will try to dig into it when I got some time on my hands!

[kriista]

Ok, finally got the board etched and am going to start building this badboy.

A couple of questions.

First, with the PnP layout, the components, do they go in on the etched side. As in, how they look on the layout version. So I would put the ICs right on top of the cover that's showing and not on the blank side of the board.

Second, would using 'hi-fi' components matter in this kind of circuit? Like using Burr Brown opamps, and nice caps for C1-4 give a better signal/noise/sound?
This little mixer is going to be an 'always on' part of my setup, and the better it sounds, the better it will be.

[Processaurus]

You put components on the blank side of the board.  If you etched the board mirrored from how it's supposed to be, in this design, that's ok, because you'd just rotate the opamps 180 degrees to have the power pins be right, all the other pins are identical if you rotate it, pretty handy.

I can't guess whether you etched it mirrored or not, but you want +9v going to pin 8 of the tl072 dual opamp and pin 4 of the tl074 quad opamp, and ground to pin 4 of the dual and pin 11 of the quad, as is on the schematic.

If a board is done, but backwards, I've seen people solder things on the copper side of the board, a little awkward, but it works.

[kriista]

It's etched exactly as seen in the image. So I take it that's 'backwards' ?

[Processaurus]

The layout with the parts is a top, component side view, so the bottom traces as shown are seen from an xray view, through the top side of the board.  So you want to print the PNP as shown on screen, so that transferring it mirrors it.  Kind of like if you were making a text iron on tshirt that you wanted to be legible from the inside of the shirt, holding it up to the sun, in that case you wouldn't mirror the print.

So if you don't want to waste the board, you can turn the opamps 180 degrees, and keep everything straight with a mirrored component layout, or solder them from the top, using insulated wire for the jumpers, and maybe soldering offboard wires onto the traces rather than through holes.  Kludging is an important part of electronics.

Ha, that would be strange to make iron ons on the inside of shirts, that are legible from the outside; a perfect venue for subconscious messages.  From SATAN!