Yet another summing amp

[bwithwings]

Hi list

Long time reader / builder, first time poster    I've built quite a few gadgets with the help of this list and others, solid state and glass, but this time I would like some advice before I jump in head first!

My band needs a mixer.  We can't really use anything off-the-shelf that I've found so far, as we need it to fit into 1U or 2U and we do things quite a specific way, so I've decided to build up a mixer that will give us the functionality we use.  If anyone could give me any advice, that would be great!

Firstly, apologies for my budget MS Paint schems 

We have about a dozen analog synths and drum machines, currently we use two horrible SM Pro 8-channel DI strips which have mixers built into them:

What I'd like to do instead is:

Inputs are all unbalanced line level, I would probably use simple jack switching to detect mono or stereo and send any mono signals to both inputs.  Pan and level control on each input, and an on-off-on bus selector switch, A/mute/B.  We would probably want to expand to anywhere up to 16 stereo input channels, starting with 10.

Each of the stereo buses would have a level control, balanced outs (to send to the stage box), and an unbalanced insert.  I'd also like to put unbalanced outs on each bus.

Maybe someone can point me in the right direction!

INPUT STAGE

Here's a rough-up of what I was thinking...  but I'm not sure where to put the DC coupling caps.  Right after the input jacks makes sense, but then they would have to be massive, I could use much smaller & better caps just before the voltage follower.  Any thoughts?  I could also use transformers but sadly I am not made of cash  ;-)

Also, maybe it's better to move the volume/pan stuff after the followers?

SUMMING MIXER

I've had a good read here:
http://www.groupdiy.com/index.php?topic=27992.msg337446#msg337446

Radiance seemed to get great performance without using a servo, so I'll give that a go too.  I'm not sure how he went about selecting what opamps to use tho?  Being more of a tube guy myself...  I could just plug & play his summing amp schem for both sides of both buses.

I have used the THAT line driver chips in a Pico compressor and the AD equivalents in Kev's LA-4 clone, so I know how to use them and am happy with that as the balanced & unbalanced output driver. 

UNBALANCED INSERTS

I could use some advice on a good way to add unbalanced inserts into the signal chain though?  My initial thought is like this:

Thanks for any and all comments! 

Darren

 

[PRR]

> synths and drum machines
> we need it to fit into 1U or 2U
> sadly I am not made of cash  ;-)

Then don't get fancy. Space (and cash) is tight. That means SIMPLE. Also, space is small, you may NOT need semi-balanced mixing 'cuz your ground runs are short. And all inputs are high-level, and they all ultimately end up the same place.

Your impedances are awful low. No offense, but your stuff is not soft solo clarinet. Working to a high impedance could be a trace more hiss, but a lot less trouble with ground-garbage, and less strain on the amplifiers.

Your sources "should" be DC-free. (If you have an old leaker, fix it.) Also I suspect you can accept slight pot-scratch. Then DC-block is not needed except where you must break the bias, or have high gain to amplify DC errors.

You have a 10K pot, a 10K series resistor (why?), a buffer, and then another 10K mix resistor. What does that do for you? Eat space, waste power, absorb time which could be better put to music or promotion. Pot, resistor, bus, DONE. The buffer is needed to drive 0-8 sub-buses, and you only show one.

Likewise the post-master buffer is not buying much. And the mix-amp could be overloaded before the master. And while the 990 is a terrific mixer, a NE5532 is plenty good for string quartet and I know it would not harm my ARP or eMu sound. And is physically smaller and many bucks cheaper.

Your insert scheme is also over-complicated. Just insert. Most modern stuff will work fine just patched in.

> more of a tube guy myself...

Then you know every grid needs a grid-leak resistor. Your input buffers have the channel pot, dubious but functional. Your Insert buffers lack any DC "grid"-leak, but need one. (If you need insert buffers; you probably don't.)

Here's my counter-offer. Each 8X2 mix needs one 5532 chip plus however many for main-bal-outs. The job is mostly jacks and knobs.

 

[bwithwings]

Thanks PRR, funny, I had assumed this thread would sink without a trace so didn't bother updating it.  I had done some more reading and thinking, and come to some of the same conclusions.  That's why it's good to think and read a whole lot before you go dropping dollars on parts.

> Your impedances are awful low.

You're quite right, many old synths have 47k or even higher output impedances.  And often in our case, they're feeding into chains of guitar pedals.  I do need higher input impedance.

Originally I had buffered each input so I could tack an LED VU meter onto each input channel at a later stage.  It's very useful to be able to see which channel's arpeggiator is stuck on!!  If I ensure the meters have a high enough input impedance (1M+) they can happily sit post-level pot.

> a 10K series resistor (why?)

Habit...  cookbook circuit with another one tacked on the end.  Jensen stereo passive volume/pan circuit, with stock opamp voltage follower - they always have a resistor there. 

> Then DC-block is not needed except where you must break the bias, or have high gain to amplify DC errors.

Right again...

> Your Insert buffers lack any DC "grid"-leak, but need one.

*laughs*  I saw that too, but as above, assumed this post was headed for Page 2 so didn't bother...  thanks for pointing it out tho!

Also, THANKS for reminding me that gain stages are better with a GAIN control, rather than fixed max gain and an attenuator.  Especially in this application.  Yet another DUH moment.  This is actually one of my pet hates.

Rather than attempting to guess how many stereo and mono inputs we'll need going forward (we swap synths in and out of the set like mad idiots), I'll make 16 identical channels with an L-C-R switch on each.  Incidentally I had skimmed over paragraphs on this forum about L-C-R switches, thinking "I don't need a passive bandpass filter"!!!  Much better than a pan control.  There are better pan controls than mine on this board, too, in case future interested parties reads this.

I think using 22uF/10k between the summing amp and the line driver is pretty excessive though, I tamed it down to 10k/1uF, unless there's something I've missed. 

So here's take two for comment...

BTW, thanks for taking the time to answer PRR,  I found that after spending a few more hours scouring this board that this topic has been covered many more times than I found in my initial search.

 

[steppenwolf]

Hmm, the input impedance of a 1646 should be 5k if I remember correctly so together with the 10k reisitor your 1uF cap sees 3,3k impedance.
This would give me a -3db corner freq at around 50Hz. Is this intended?
Correct me if I'm wrong;-)
Best,
Stefan

 

[bwithwings]

10k for the SSM2142 - thanks!  2u2 instead, -3dB at 15Hz.