Perfect summing frontend?

[Svart]

Even though not many seem interested, I'm still moving along with the summing buss idea. Although I could copy a number of different designs I would hear Ideas on what the ideal design would be and why.

 

[soundguy]

whats ideal for one project is worthless for the next.

you really have one big decision to make, wether to do it passive or active. For a small summer you could easily do it passively. If you want to do it actively you get the benefit of having another pair of opamps in there for color if you so desire to mix and match. Big choice of what amp to use for makeup gain. I think thats really the stuff that is going to make it sound different, the rest is just utility features if built right which shouldnt have an effect on your tone at all. If you want to go way over the top you could have active buffers on each input (dont quote me but I may remember seeing that on the dangerous box) but its kinda not necessary. Good place to add another tone, I think the benefit in most *practical* situations is imagined. Same for transformers on the inputs.

dave

 

[moqtev]

I would spend my time and money on good preamps and microphones (and how to use them). I don't think a summing device is worth the trouble. If you want to color your sound: Sum it with your DAW and send it through your favorite preamp one more time :grin:

Morten

 

[Svart]

I understand your statement Soundguy but worthless is a strong word. Everything is worth something as long as you look at it objectively, thus the reasoning behind trying to get some opinions of those who have used a number of different designs.

The backend is almost complete, DOA w/servo outputs and most of the switching work done. Working on the insert stuff now. Waiting on some input from another thread about using relays for muting or going with Jfets too.

I'm looking for the tradeoff between lower numbers of DOAs and still keeping my S/N ok. I'm thinkng hybrid active/passive. So far the output is active but the Insert is looking to be passive(no buffer before or after) but this would need to be coupled via a cap for safety's sake at least which I don't want to do so i might end up using another buffer.

I'd like to see an active impedance converter, matched BJTs and opamp ala the 4Kssls but I'm keeping myself open to anything right now.

I don't mind trafos either.

how do they do it in your Sphere?

 

[Svart]

moqtev, Summing busses are very important especially in larger format consoles. I was playing around the other day and removed the OPA2134 in the impedance converter of one of my buss channels and replaced it with a ne5532. Although you might have read about my like for the 5532 in regular channel strips, using it here sucked the life out of the signal. this was a room mic/overhead buss and it took the sheen and body from the cymbals when a lot of things were going on. Replacing the opa2134 brought the life back. I was pretty shocked but I soon relaxed remembering that smashing a lot of signals together can take a toll on the quality if the speed of the opamp isn't high enough. I popped in a mc3307(9?) and the lack of speed was apparent again although the coloring of the opamp was different. this was the extent of my testing but later trying with the guitar buss showed the same results.

I feel that this is a very important area that can really make a difference in your audio.

 

[soundguy]

[quote author="Svart"]I understand your statement Soundguy but worthless is a strong word. [/quote]

Its absolutely not. I have the most bitchin studio in the united states if you want to make a rock record that sounds like electric warrior and its pretty much worthless for recording any kind of string quartet or anything else that requires a low distortion, clean signal path. I do (did) a lot of film work and have virtually a complete different set of equipment to use for those jobs with very little, if anything overlapping between both worlds. When you get into designing something that specifically rules you had better believe its going to be specifically worthless as well. Thats the whole compromise with building something incredibly focused.

the summing buss is either passive or virtual ground. Assuming that you are only summing a few channels there are probably little gains over a virtual ground buss beyond what Ive mentioned. Unless Im missing something, having an active output is unavoidable in either layout. The rest of the stuff is bells and whistles. Id concentrate on making the actual summing stage sound good and then add stuff on top of that to suit your needs. Thats the way Id do it at least, many ways to skin a cat.

dave

 

[NewYorkDave]

RE: this and your other "summing" thread, I think these are questions you can really only answer for yourself. It's all in the application and desired result.

For example, I'm planning a small console for use in my modest 8-track home recording setup. (I can hear Soundguy laughing because I've been talking about building this thing at least since 2003). Its architecture will be perfect for what I'm doing, but would probably be fairly annoying to someone who's used to working on a console with the usual modern complement of features.

 

[Guest]

I actually was reading the Steve Dove chapter on Console Design at the weekend - after a year or so of hanging around here I am actually starting to understand some of it
I know what I want to do - more home mixing of 24 tracks to stereo out (with PFL, pan pots, stereo faders - a couple of aux sends - sod the insert points - I will use a patchbay) and no DAW in site - all outboard - I stopped using a PC cos it gets on my tits
My question to add to the thread is - what be the difference between summing and mixing (or is it just that summing nowadays means out of the DAW analogue summing and back to the DAW - whereas mixing means all the bells and whistles)

 

[Svart]

NYD, this will be 24x8x2, 2 aux. Insert, DO, LI, MI, pan, mute and balanced elco I/O will be included and are part of the frame I plan on using. I plan on the signal to the recorder being switchable from direct, after preamp and after fader.

I never use PFL or solo since I don't do any screwing around during tracking, I just track straight through preamps to the recorder. the headphone system is separate in my studio. I will include a pair of separate, buffered output for headphones and studio, but I have my monitor system as the last in the chain after the mixdown computer so i can hear what everything is doing to my audio.

I think the mutes will be relays and not jfets. I haven't decided if I want to incorporate any special devices like touch sensors yet. I also plan on using limited assignment of channels to the bussing/master buss and each channel will have hardware assignment jumpers on the PCB.

 

[babyhead]

[quote author="Svart"]

I think the mutes will be relays and not jfets. I haven't decided if I want to incorporate any special devices like touch sensors yet. I also plan on using limited assignment of channels to the bussing/master buss and each channel will have hardware assignment jumpers on the PCB.[/quote]

I remember a discussion about using optos for mutes, the advantage being silence.

Just throwing it out there.
:guinness:

 

[Guest]

svart
yep - you got me

How about then.. molex connectors so you could add 2 or 4 pole etc.. pusbuttons or go for a Top quality illiminated pushbuttons with your own relay card and then just go in on the molex connectors (ie you choice of pushbutton)

 

[Svart]

That's a good idea. The only thing that concerns me is audio through the element and nonlinearities/noise/distortion it might induce. With relays i know what I am getting into, with Jfets too but I am opting out on those. I wish the silly-con audio switches like what $$L used were easy to find or I would use those.

 

[Guest]

I was actually thinking of the Molex connector for the switching action not the audio feed - so there will be relays on the board

 

[Svart]

have a link?

 

[Guest]

Sorry svart - I may be a bit more simpler than you think
Somethign liek this
http://www.groupdiy.com/index.php?topic=178092&highlight=eao+relay#178092

 

[sodderboy]

[quote author="Svart"]NYD, this will be 24x8x2, 2 aux. Insert, DO, LI, MI, pan, mute and balanced elco I/O will be included and are part of the frame I plan on using. I plan on the signal to the recorder being switchable from direct, after preamp and after fader.[/quote]

You are going to need opamps all over this design, so the word "passive" does not really relate. Transformers can be great, but who wants 5 in the signal path unless they are St. Ives or UTC etc. A standard Neve 80X8 signal chain from mic to volume knob goes through 7 transformers and 4 or 5 opamps. You could lose the trannies, but not the opamps.

The complex changing signal path impedance created by your auxes, the insert, the pan, and the switchable output choices will require buffering (an op amp or transformer) at each position.

Check a block diagram of a simple Maky, Soundcraft or Tapco- there are opamps all deep in them for a reason. You are designing a console- nothing passive about it. What does your block diagram look like? That should be the first drawing after an abstract.

What was the name of Klett's ashram again. . . ?

 

[radiance]

So if I understand this right...everywhere you want to tap off the signal (aux sends) or where you want to distribute the signal to multiple destinations (assigning to multiple groups) you will need a buffer stage right?

Now here's another question: How critical are opamps used for buffering? (By critical I mean: how much do they afect sound quality) Sure, we all want quality but sticking a DOA on every location might be overdone right?
Will there be a noticable difference when u use for example a burr brown OPA2604 insted of a DOA (for example a 990) for buffering?

 

[Svart]

I disagree with needing opamps "all over" a design. In today's output stages there tends to be more than enough oomph to drive the next stage easily especially with something as robust as the FETbloak. In my design, a DOA will drive the output of the impedence converting stage based on MAT04s and two DOA opamps. the insert will be after this and returning from the insert will be the faders then output drivers based on DOA/servo. It is MY personal feeling that since most devices that will be inserted will have their own input buffer and their own output buffer, that why should I put another stage in the way? If during testing I find that the return side of the insert needs some kind of impedence matching, I will figure something out.

OR

I could use a jack that when normally closed (no insert) it bypasses a return buffer and when there is an insert present it engages it..? That way I have the shortest path possible when I need it.

 

[sodderboy]

It is not about oomph, but of constant impedance loading of each stage. Just imagine how a Pultec loads your proposed insert compared to a dbx 160x, compared to nothing. Think about it- the parallel impedance is different in each case, and will effect level.
Have you studied some block diagrams yet? Console makers would not add scores of chips unless they were necessary for function.

Just imagine adding two 25k aux pots to the simplest passive summing buss- how does this change the source impedance feeding the buss? They would decrease the source impedance because they are 50k in parallel with the source resistor. And, it depends on where the pots are set, and what the aux 1 buss impedance is. Bump up ch 2 aux 1 to the spx-90 (damn, im OLD!) and the ch. 2 source impedance feeding the summing buss changes, changing ch 2's level in your mix.
Just imagine that every pot tweak changes all the impedances-read voltage, read level, down the un-buffered line. Buffering makes for many short happy lines. I am not even bringing the possible RC filtering that could also result.
And then add in your selectable Direct Out and the pan pot- it just canna be dunn, Captain Kirk!
Check this out:
http://www.dself.dsl.pipex.com/ampins/mixer/mixerdes.htm
It scratches the surface regarding buffering. Buffers are not really bad!
Mike

 

[Samuel Groner]

Personally I don't think that a discrete frontend does any good to a make-up gain stage based on discrete opamps. First of all, there are many very low noise DOAs out there which you will not beat by many dBs with the discrete frontend. Second, keep in your mind that current-feedback designs such as GainBloak & CO can not be used as an integrator, which is often required with a discrete frontend (haven't seen yours though). Third, DOAs tend to be less stable than their integrated counterparts and an additional stage doesn't help in this respect either.

For my personal use I have a design in my mind which uses a balanced passive summing network followed by a make-up gain stage formed by two DOAs, connected the way we've seen it as the first stage of an instrumentation amplifier. These opamps drive an output transformer with good CMRR.

Samuel