Line mixer (active)

[Guest]

PRR, you are the best moderator I ever saw, and Thank you for your post,
but I still don't understand why Svart can't group channels in manageable numbers, instead of doubling everything, and run wires for thyristor regulators of plasma coolers for coffe makers inside of shielding pipes?

 

[the scum]

OKAY, I've actually got parts of one of these things working now.

I've been working from Dave's original schematic. Since the stereo summing buss was the most complex portion of the circuit shown there, I used it for my prototype, omitting the buffer & auxes, as well as any solo logic and the VU driver.

Dave, if you're reading, the original schem fell off the host. It makes discussion of the design a little harder if people can't see it. Did you take it down intentionally? I'd be interested in seeng the updates you'd mentioned, as well.

So I've got two channels feeding a stereo buss, enough to verify that it's mixing properly. So a quick tour.

Here's the front panel, sharpie nearly legible in the photo:

I used toggles for the switching because they're easier to prototype with...the pushbuttons really need a PCB to mount them on. Only the mute is connected right now. Sorting out the solo is going to be the next step in the project.

Here's the proto board:

The opamps are right out of Nelson Pass's article. The sound good, but don't drive loads, and DC offset is a significant problem. But if I can whip their offset, any more civilized opamp shouldn't be a problem.

On the left are the fader buffers. The summing amps are in the middle, and the output drivers on the right.

Here's some detail of the point to point work for one channel:

Mute switch feeding panpot, with the fader on the right.

Finally, here's a scope shot:

I've got a small triangle wave and a larger square wave summed together, both channels panned to the center. The L & R signals are superimposed here.

I also took it into the studio and gave it a listen. It's certainly not objectionable. L/R crosstalk is minimal, the mutes don't thump. Unity gain is at about 2 o'clock, right whre I'd expect it. Pan-law is 3dB down in the center.

I think the next steps are to get the solo sorted out, move to the impedance balanced summing, then get back to the PCB designs.

I'd like the PCB to be sorta a springboard for mixer experinentation. I'd like to hear what people think about the following points:

- mechanically simple: each portion on a card that hangs off the panel, wired together using busses made of IDC and ribbon cable.

-easy to source: Taiwan Alpha pots, standard IDC connectors, ITT/Alps pushbuttons, etc.

-Flexible for experimentation: there are lots of variations on this basic idea that don't require too many changes. For instance, the channel cards could be designed for the "short mix path" of Dave's original, or respun slightly so that the aux buffer is actually a balanced line receiver, and the fader hangs off it instead of the input. Discrete/monolithic opamp footprints aren't too tough to do. Similarly, it's a couple of extra resistors per card to used impedance-balanced summing...or a few extra caps (DC-blocking and unity-stability) to make it so it'll accept a wide range of opamps.

I don't want to do a wholesale redesign of the oiginal, but with a few tweaks to the PCB, it can be much more flexible, and not add much to the overall PCB footprint. Someone could experiment with it, and then they decide they've outgrown it, they'll have gained some insight into the inner workings that will allow them to add the features they desire (left as an exercise for the reader: more channels, auxes, busses, whatever). It'll help remove some of the mysticism about summing. Dave, if you're so willing, we could put it under the an open-source license, so alterations and improvements come back to the community.

 

[NewYorkDave]

The schematic was lost when the Twin-X server crashed. I began work on an updated version with a few refinements; I just haven't had much time to finish it. If I don't feel too crappy tomorrow (I'm sick with a bad cold at the moment), I'll plod away on it some more.

 

[radiance]

Way to go Scum :thumb: It's good to see that someone is actually building this project.

 

[NewYorkDave]

I just looked over that Forssell document that was linked in page one of this thread. I think this must be a typo:

Note: Only One Summing Ground Buss for entire Console

...because it wouldn't work the way it should unless you had a separate build-out resistor and ground buss for each mixing buss (left, right, auxes, groups and so on). Or at least that's how I'm reading it in my present decrepit state. If there were a single buildout resistor for each channel feeding a single ground buss for all the mix busses, consider what happens to the level of the noise on the buss as you strap more mix amplifiers across it. You want the mix amp's effective noise gain (noninverting) to equal its signal gain (inverting) for best cancellation of common-mode noise--and that would be skewed off the target with each mix amp you added on that common ground buss.

I still wonder about the necessity of a balanced or quasi-balanced mix buss in a small console, if good construction and grounding practices have been followed.

BTW, in Steve Dove's epic, the drawings of the master group electronics imply that the signal grounds from the input channels are bussed into the noninverting input of each mix amp (which is connected to its local ground reference not directly, but via 30K). But so far, I haven't been able to find a specific reference to it in the text. Again, I'm not operating at peak capacity right now.

 

[the scum]

Dave,

I can't make any sense of that comment on the Forssell schem either. Also, wouldn't you want to match the buss feed and buildout resistors? Otherwise, wouldn't the buss source impedances be different, and possibly harm the desired CMRR?

Similarly, on the summing side, wouldn't both lines need to receive the same RFI filtering (a la Dove), if any is applied?

As for mix buss topology: if this really is a small mixer, then you could actually move the buss feed resistors to be right next to the summing amp, resulting in a summing node that's maybe an inch long.

When allowing for the impedance balanced mix buss, the cost is a handful of resistors and some extra lines (4 in this case) in the backplace. I don't think that's too prohibitive. If someone really isn't interested, I can provide jumper points on the summing card that go back to the simple inverting amp.

 

[NewYorkDave]

It really depends on the mix amp. Some diff amps care about balanced source impedances, others not so much. I'll include some jumpers which can be replaced by resistors, for those constructors who wish to try "quasi balanced" mixing.

Hold off on making any PCBs. The revision is coming... as soon as I can stay awake and focused for more than two hours at a time.

 

[NewYorkDave]

All right, finally some progress at my end. My freebie version of ORCAD only allows saving schematics of 60 components or less, so I'll have to break this up into sections. The input channel PCB has been uploaded to an album I've created especially for this project:

Link

I've included simple assignment switching that allows a choice of L/R or SUB 1/2. Anything more sophisticated than that, if done right, would require additional buffers--and it's still my goal to use a minimum of amplifiers in this design.

The master section schematic is already in progress.

 

[radiance]

Thanks Dave :thumb:

 

[Guest]

Woah Dave - many thanks
Most of tyhis stuff is way over my head but I enjoy trying to figure it out
One thign I may be able to add is this circuit...
http://www.groupdiy.com/index.php?topic=16647&postdays=0&postorder=asc&start=30

See the big electro cap before the fader - that is there (from my notes from the meeting) to minimise noise as the fader is pulled down
Some console designers allegedlly may have got a wrong cap in there

 

[NewYorkDave]

The input coupling cap (post-fader) shouldn't be necessary unless the amplifier draws significant input bias current.

 

[Baltimore]

I think i'm going to build a little 16 channel guy based on this schematic.

Would it be at all possible to use a 1:1:1 transformer on the input with one secondary driving the fader and pan, and the other secondary driving a pair of aux sends? I realize the auxes will be only pre fader, and I won't have the extra gain provided by the opamp after the fader, but I don't think this should be a problem...I find I usually have to pull the faders down pretty low to avoid overloading the mix bus anyway. If not, I can build with opamps, I'd just prefer the simplist signal path and the easiest build.

Kevin

 

[NewYorkDave]

All that just to save one fifty-cent dual op-amp?

For one thing, the whole gain structure of the master section would have to be reworked, the fader would no longer be buffered, the settings of controls in the channel would become more interactive, S/N would deteriorate, etc. So, do it if you like, but you'll have to work out the details for yourself.

Anyway, my health is improving and I'm becoming productive again, so the master section should be ready to post sometime this week.

 

[Baltimore]

[quote author="NewYorkDave"]
For one thing, the whole gain structure of the master section would have to be reworked, the fader would no longer be buffered, the settings of controls in the channel would become more interactive, S/N would deteriorate, etc. So, do it if you like, but you'll have to work out the details for yourself.
[/quote]

Alright, so I won't be trying that....I was originally planning on using DOAs, but leaving them out would make the build easier, with a much smaller power supply, which is why I asked. However, I do already have a bunch of yamaha 80200 DOAs so I'll use those and a 5532 to buffer the auxes.

Thanks!

Kevin

 

[radiance]

[quote author="NewYorkDave"]
Anyway, my health is improving and I'm becoming productive again, so the master section should be ready to post sometime this week.[/quote]

Take it easy, but OTOH can't wait to see that master section of yours :green:

 

[NewYorkDave]

Well, I haven't finished the master section drawing yet, but I wanted to share a cool tip. If you like, you can build just the channel strips, add suitable shunts to bring the impedance of the mix busses down to 150-200 ohms, and use your favorite mic preamps as program amplifiers.

 

[radiance]

Thanks for the tip. I will be building some channel strips when I've some time left AND when those Lundahl transformers from the black market have arrived......
I will use your chematic as a starting point. Basicly what I need is a 20 channel (10 stereo channel) line summing mixer. Each channel strip will have two aux sends (post fader), a buss select switch, a fader (rotary switch I guess) and a mute button (no pan pot). This way I only need one opamp per channel.

I have one qustion though.....How much buffering can one opamp provide?
In this case it's buffering two aux sends and the master/groups but is it also possible to make the channel assignable to a group AND the master? Or will that be asking to much from one opamp?

 

[NewYorkDave]

That has more to do with the loading the panpot needs to see, which is 10K. If you want to be able to assign to master and subgroups at the same time, the best way would be to add unity-gain buffers after the panpot, and just make sure the panpot is loaded by 10K. I didn't go this route because my goal was to minimize the number of amplifiers.

Another way would be to use 22k mix resistors, change the feedback resistor in the mix amp to ~44k and provide a more complex assign switching arrangement with dummy resistors to ensure a load of about 10k on the panpots regardless of assignment. This would work with one master and one subgroup; the whole thing would have to be scaled up again for additional groups. I didn't go down that path, either, because of the KISS philosophy I've been trying to follow with this design.

You can, however, add a couple more aux sends if you like, without taxing the buffers unduly.

 

[radiance]

I will skip the panning pot since it's all stereo channels that I'm building.
This makes things a little easier I guess.
Also, I don't know if I want to be able to assign to master and subgroups at the same time or not. I would only use such a thing for parallel compression or so, but I can also achive that by feeding the compressor from an aux send....

 

[NewYorkDave]

Ah, grasshopper, if you leave out the panpot, then you have to readjust the level and gain structure throughout. The panpot circuit has 6dB minimum loss, and the rest of the circuit was scaled to harmonize with that.

It's a "simple" design, yes, but there's a reason for everything in it.