Perfect summing frontend?

[soundguy]

[quote author="sodderboy"] I am not even bringing the possible RC filtering that could also result.[/quote]

level can always be made up but filtering is not something thats easy to fix by turning a knob. As a guy trying to learn mixers, I could certainly learn something from a discussion regarding rc filtering across an unbuffered buss. Are there some basics to touch on there?

thanks

dave

 

[Kit]

Thanks for the link, Mike.

Just the kind of paper i was searching for.
And i agree. Buffers are good, opamps are cheap.

 

[Svart]

Console makers would not add scores of chips unless they were necessary for function.

Sure they would. They do it all the time because it's easier and cheaper than doing this properly sometimes. Sticking a piece of sand between stages is sometimes necessary because it isolates the next stage, BUT we are still talking about sandstate. Impedence is an antiquated term used here. We are really talking about current and voltage drive and the ability of the next stage to sink it. If the output of the stage before the insert is plenty strong to source the current needed to drive the inputs on the inserted device without voltage sag or distortion, then why do we need another buffer stage to do the same thing? Same for the output stage of the inserted device. If the device can drive simple BJTs/FETs in the input of the next stage's opamps(doa or otherwise) then why would we need to add a stage that will also see the output driver of the inserted device doing the exact same thing? What you are talking about is the fine line between being able to support driving insert cable, jack connections, conductors in the console and finally the next opamp itself.

I am simply making the stages strong enough to drive inserted devices directly. this would be the same as the output of the mixer driving a soundcard or recorder's inputs.

On another note though,

I'll compare the ssl4k impedence converter to one from an alesis x2.

Both have BJT frontends, both use 2 opamp stages afterwards in a similar fashion.

the ssl uses caps in a bipolar arrangement and biases them correctly. the alesis uses polarized caps without biasing.

the ssl uses matched BJTs in a parallelled arrangement, the alesis uses two discretes that don't even measure close.

The ssl uses 2 5534s with biasing, the alesis uses a single 5532 without.

The ssl uses inductors where needed, the alesis has none.


I am not simply throwing things together because I think they will work, I am placing things together in an order to see how they work. From there I will determine what is important and what can go.

 

[Svart]

and yes buffers are good.. if used properly and not just thrown in because some other company used them. Remember that most mixers are now IC based and those ICs are very limited in their drive capabilities. To do this properly, you would use an IC with discrete BJTs following as your current/voltage amp, much like they do with headphone drivers.

Opamps are indeed cheap.. usually much cheaper than using inductors and good resistors/caps. Thus the demise of the passive L/C EQ with a following opamp gain. Remember that this used to be the norm and now it's the exception.


And this is a virtual earth design.

 

[Svart]

and..

The board will have a footprint for a DOA and IC opamp of the 5534 footprint besides the servo.

I've not been a big fan of Dself. Some of the things he says are great, some.. get under my skin.

Here is some no BS reading:

http://focus.ti.com/lit/an/sboa092a/sboa092a.pdf

 

[Svart]

an update:

the output driver has been prototyped. I A/B'd a few types that I designed, some i hacked together and one that I took from another project here.

The forssell output driver circuit like the one in the Opto comp won. This servo design works with both the 2520 and the big Fetbloak just fine.

Now to get the front end layed out and proto'd. I came up with a number of partial designs of my own and some that I hacked from available literature. The more I screwed with my own designs, the more they looked like the published SSL4k impedence converter setup.

I guess that makes me feel that I was on the right track at least. I have decided to go ahead and use that design with some of my own hacks and updates to save time. Not quite as DIY as I wanted but I am now running thin on layout time.

I'll get a new prject specific thread up soon when I can post what I have.

 

[sodderboy]

You should not quote the Old Testament to me when I am trying to explain how to get from Atlanta to the North Pole. Keep the sun to your left in the morning, and to your right in the afternoon.

[quote author="Svart"]

I am simply making the stages strong enough to drive inserted devices directly.
[/quote]

Svart, this a great metaphor for my posts to you.
In so many aspects of my life, I have determined that OOMPF is not the answer. I have learned to lift many heavy loads with my brain, and not my braun. I am only one person.

Sorry to be oblique, but the chicken sausage, Coronas, beet salad, potatoes. . . Burrrrp- I should be in the hic! Brewery. . . they have gotten to my brain. Perhaps I need a Bromo- or a BUFFER as you will, to calm me.

Dave- just take the block diagram of the sweet-heart of your studio- and remove all transformers and "DOA's". Nevermind the gain aspects. What kind of Kirchoff laws would you be changing and re-changingting everytime you tweaked a knob? Which RLC circuits would change with every goose of gain? Active electronics are soooo good. . . But CAN be minimized within reason.

The wheels are round, but some have wider white stripes. . . Hic :sam: :sam:
just one tech's opinion. . . :sam: burrrrp
Mike

 

[Svart]

:shock:

strange..

Well, I see where you are coming from. All I am saying that that the output drive has to come from somewhere. It could come from a buffer stage, but why not from the gain stage before it if that stage can handle it without strain? It only seems to reason that getting devices out of the signal path can only help.

Like I said, I'll give it a try via simulations and prototypes. If it turns out that my idea cannot be done I'll send you money for a 12 pack of brew. However, everything that I have done shows it can be done within reason, I just haven't gotten to finish that section of it yet.

The question is.. do I want to use caps after the insert or do I want to servo it? If the answer is servo then I'd have to buffer it...
:thumb:

 

[soundguy]

[quote author="Svart"] All I am saying that that the output drive has to come from somewhere. It could come from a buffer stage, but why not from the gain stage before it if that stage can handle it without strain? [/quote]

not that I know my ass from my elbow, but this probably seems like the place where headroom is going to become an issue. If it can handle it, it can handle it, but a few amps lifting a little may have an overall more pleasant effect than one amp lifting alot. Its a fun theoretical discussion but the sound out the box wins in the end. Look at the sonic differences between one opamp mic pres and dual opamp mic pres, thats a pretty noticable difference on a mic, thats a huge difference on a summing buss. Again, I only have the most marginal of experience testing and listening to different bussing schemes but so far with what Ive played with less gain from more devices has sounded bigger than more gain from less devices, this all within reason of course. Nothing sounds quite as small as the 200 opamps in your signal path in a soundcraft ghost.

dave

 

[Svart]

Yes, it's fun but I also plan on doing some real testing too once I get a template layout where i can move parts around. Heck I might as well lay the board out for both types and just have jumpers for the extra buffers. I'm already going to stick footprints down for 5534s/servos anyway.

 

[sodderboy]

[quote author="Svart"]:shock:

strange..

Well, I see where you are coming from. All I am saying that that the output drive has to come from somewhere. It could come from a buffer stage, but why not from the gain stage before it if that stage can handle it without strain? It only seems to reason that getting devices out of the signal path can only help.

[/quote]
Ok. I slept it off! :green: Certainly the least number of components/ opamps the better, but when you talk auxes or selectable Direct Out sources, you will need more than wire. Just remember that the sections will not be static- equipment of varying impedances will be inserted or not, sometimes sections will be loaded with your DO circuitry, sometimes not. All these variables will need to be controlled somehow as to not affect signal flow from mic to CR monitor out.

You should have a block diagram before you start tracking boards. It is kind of like a business plan or archetectural drawings. It helps you stay on track.
Mike

 

[Svart]

Right now I am still working on the buss channels as I feel these are going to be very important. I'm not planning on having aux outs on these at all, just inserts. the master section will be similar but will have a buffered prefader output for the external headphone/monitor system besides the main outputs.

flow chart:

buss input-->impedence converter-->insert-->fader-->output amp-->assign/mute

 

[analag]

C'mon Svart gimme summ..thing to look at.

analag

 

[Svart]

:green:

patience..

I have to redraw some things now that i decided to use the SSL4k frontend as the basis.

I'll have to figure out how to post it since the schematic is pretty big in paper size.

 

[radiance]

So,..... how's the project goin' Svart?





or did I mis something

 

[Svart]

oh it's slow. I have about 12 schematics of various schemes from extremely simple to full blown nightmare. the only thing in common is the attempt to use as few opamps, IC or DOA, as possible.

I guess it sums up as, I don't know where to go from here.

Someone give me some ideas what you would like to see.

 

[mikep]

[quote author="Samuel Groner"]
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[/quote]

Wow. Ive had that topology in my head for a while as well. Ive done some tests and LL1517 works well for this application. sice all the windings come out to pins you can use just one DC blocking cap at the center tap point. you can do the same thing with 2 inverting amps and the same transformer. by the way, the OTA-1 schematic I posted makes a good discrete virtual earth summer.

 

[radiance]

[quote author="Svart"]Someone give me some ideas what you would like to see.[/quote]

Take it easy, I don't need to see pics or so. You probabely only make a summing mixer once in your DIY carreer, so it better be right!
I'm in the process of building one myself and got a lot of inspiration out of this thread.....

http://www.groupdiy.com/index.php?topic=18007

I'm with you on the case of not buffering inserts since the inserted units have in and output buffers themself. Just leave the cases away and you'll have one big aparatus :grin:
One thing that I'm not sure about is wheter I'll make my inserts balanced or not. Balanced whould mean another trannie or opamp....

Cheers :thumb:

 

[JohnRoberts]

[quote author="radiance"]So,..... how's the project goin' Svart?





or did I mis something [/quote]

This thread looks a little long in the tooth, but is an area I have some interest and familiarity with.

One poster referenced a Steve Dove "chapter". I recall a series of articles he published in Studio Sound magazine back in the '80s that were quite good. Steve is a solid designer who has actually walked the walk (Alice consoles in GB among others).

I would offer a few cautions or suggestions about console design. #1 look at as many schematics as you you can find and try to figure out why they did the little unobvious stuff. Consoles are probably the most difficult "simple" circuit (next to power amps) you will encounter. Just like power amps get progessively more difficult as you scale up power, consoles get dificult with increasing numbers of channels.

My second caution is be very aware that ground cannot be zero volts across a span of several feet even with brute force copper bus bars or whatever. Be prepared to take advantage of differential circuitry to forward and back reference signal grounds (opamps are good for something after all).

A third is try to keep absolute polarity straight if you have inserts and/or sundry opportunities to alternately patch signals in and around your signal chain.

Finally the heavy lifting in mixers is the mic preamps and summing buses. For only 24 channels to L/R, the noise gain of less than 30 dB is probably manageable with modern off the shelf parts. (I once did a console with 112 inputs to L/R but that's a somewhat different animal).

JR

PS: perhaps a minor point but "bus" is spelled with one 's', if you look up "buss" in the dictionary you will find a verb that is unrelated to electronic design (and more fun).

PPS: I am an old fan of 5534 as it was a great part in it's day and still pretty serviceable applied at modest gains. One possible concern for the tweaky types (I'm sure there's none here :roll: ) is the extra wiggle in the open loop transfer function well up into the audible range (look at Bode gain plot on data sheets). This is no doubt related to the technique they used to compensate this older process part to get decent GBW and slew rate. I think they called it "split-pole compensation" or something like that. I'm not making any claims about actual audibility which should be down in the dirt for any credible application, only pointing out that there are better looking (on paper) modern devices. Since I presume for many the time investment is larger than the parts cost this may be a consideration.