Mixer design

[chris p]

console - definition - makes you weak at the knees

mixer - definition - something a drunk guy mixes on at a pub gig

Nice one Jamil!

This Christmas I'm dreamin' not of snow but, in my idle moments, of a DIY mixer, and would be very interested in turtlerockers' views as to what makes a great mixer in terms of functionality (as distinct from the "sound", which is a separate exercise). The exercise is not necessarily to built it (although that may happen), but to better understand concepts like gain structures and how consoles actually do what they do. Here's where I'm at for now:

Concept: a 12 mono channel (1-12) tracking / mixing console (NOT really designed for live work) with 4 stereo busses (A-D), a main LR bus and a Monitor bus. Each channel is also connected to a firewire DA for DAW playback. DAW AD inputs are permanently linked to channels 1&2, buss A though D LR and main LR (12 in total).

Mono channel strips
Inputs: XLR low-Z input for mic and line, 6.5mm socket for Hi-Z DI
TRS 6.5mm unbalanced send and return for channel insert, selectable pre/post fader
XLR transformer balanced out +4 dbu
17 segment LED PPM display
Input controls: Gain, variable impedance, +48V, phase, line (a -28db pad when the XLR input is line level), mute, DAW select (disables inputs and takes signal from DA), pan and a fader. Switchable routing to any combo of the 6 stereo busses.
EQ: in/out switch, 5 frequency HPF, 4 frequency +/- 12db LF shelf, 5 frequency +/- 12db MF bell, +/- 12db HF shelf at 12kHz (yes folks, think the EQ controls of a 1073).

Stereo Bus A through D
LR and Monitor assigns, mute switch, dual 17 segment LED VU display
Single fader for LR vol
EQ: in/out switch, LF cut, boost and freq, HF boost, Q and freq, HF cut and freq (yes folks, think the EQ controls of a pulteq EQ1A but operating in stereo).
VCA compressor: in/out switch, Threshold, Ratio, Attack, Release, Gain, 8 LED gain reduction metering plus something like the RNC's nice switch that implements serial compression
Separate L and R unbalanced send and return 6.5mm sockets, tranformer balanced XLR outs +4dbu.

Master Section
LR: faders, quad 17 segment LED VU display for L, R, Mid and Side, tranny balanced XLR outs +4dbu
Monitor: Vol, 2x 10 segment LED VU display, 6.5mm L and R balanced outs +4dbu
Headphones: 6.5mm socket, vol control, Mon/Main select switch
Talkback: XLR socket, vol control, momentary on/off switch (Talkback goes to Monitor channel)
DAW interface display showing activity on 12 in, 12 out channels

I suppose a picture is worth a thousand words, I have a mockup but its too big to post here easily.

Questions:

How many channels do people actually record simultaneously? I know I'm at the bottom of the food chain, but my max to date has been 8 (as in 8 channels live to DAW at the one time).

Most consoles have additional AUX send and receives, which have vol send controls and maybe a pre-post switch. There is then a series of return vol pots at the master to blend the effect returns with the main LR. While I can see these are really useful live, do they offer any advantage to the stereo busses when mixing/tracking? I've avoided them in my design so far 'cause I can't think of a real use given the channel and bus send/returns.

What else have people found to be really useful on a console?

[graemeh]

Latching talkback button, preferably illuminated!

[Chinagraf]

VCA or channel subgrouping and inserts on the group busses and master buss. A dim button on the monitor feed with adjustable level. A mono button. Some 2 track returns.
The stereo returns, as in the fx returns you described are handy if all your channel are maxed out by splits or parallel processing. I record more than 8 tracks at once but only if I'm doing drums or orchestral based calls. The illuminated talkback is a good one. I've seen a few band fights start because of what someone said in the control room while another member was recording. Ay the place I began we stuck a paper arrow pointing to the red light on the SSL at one point becuase the auto talkback caught a few out.

[michael]

the thing about a DIY console/mixer is the frame and layout as a starting point... so much to consider and understand before you can even start designing internals.

what i've thought would make a cool DIY mixer project (for someone with more of a clue than me), is to start by gutting a small board, something like an old Ta@#$% M35 (8 x 4 x 2), or the bigger model M520? 12 x 4 x 2.

(Those old Tascams were/are actually pretty decent formats, (sturdy frame, fair bit of space inside, sends on all channels, Aux Sends/effect sends, Basic talk back function, modular channels, VU metering, etc), and personally i like the 80's Ta@#$% lo-fi dark sound).

But anyway, a cluey person with more time than sense could start by "simply" looking at the power supply system, or redesigning a great channel card and replacing channels before needing to tackle the summing side of things. Eventually a whole board could be completely DIY'd to your own specifications.

[Linear]

I've got a huge PDF of a Studio Sound article called 'designing a professional recording console'. It was written by Steve Dove, and is both extremely entertaining (in the way it was written) and contains mountains of helpful information. You should see if you can get yourself a copy.

I thought about building a console for a while there - but it's SOOO much work just to get the basics working properly. And I was fortunate to come across a console that basically had everything I had wanted anyway.

I reckon a better way to go would be to buy a fairly ordinary cheap (but reasonably professional) console, like a yamaha, Ta@#$%, soundworkshop or similar. Use the basic existing console as a starting point, and build your new console around it (ie differentiate recording/mix signal path with monitoring/talkback functionality). Rebuild it so it's no longer even a shadow of it's former self - ie not a modded 'X' console, but a new 1-off.

Just my 2c.

Chris

[chris p]

Thanks folks - more comments, please.

The issue of "oops" talkback is why I went with a momentary rather than latching switch. I can see that going with an illuminated (and arrowed!) switch / indicator is a great idea if t/back can be inadvertently left on, but wouldn't a momentary switch avoid the problem in the first place?

-20 dim button on Monitor - good idea, done
Mono button on Monitor - done
2 x 2 balanced stereo inputs mixed onto LR bus - great idea, done
Insert points on LR bus - doable.

China

The groups (A-D) already send/returns and the channels (1-12) do have subgroup assigns. Did you mean something else?

Michael/Chris

I know what you're saying - I've also thought about shortcutting the way you suggest buy buying a cheaper console and refitting it all. Even doing it that way, though, you'd want to know that the console you get has the capacity to achieve the end result, so you need to know what the end result is. I'm also not sure that the Ferrari engine would work in the Daewoo chassis (to make a bad analogy - but retrofitting a great preamp to fader/bus hardware not able to handle its headroom is more what I had in mind).

Chris, my mail server can cope with large PDFs - any chance of sending me that article?

[rob]

the studio sound article that Linear mentions is prerequisite reading

i designed and built a console once.....when young naive and more time on my hands than i ever seem to have these days. It took 18 months and a stupid amount of hours. The best thing about it was learning from all the mistakes i made...oh and the EQ ... it had a great EQ!

R

[wez]

Latching talkback button, preferably illuminated!

every band that has ever been in a studio would have broken up before they finished their first album if talkback buttons were latching. there is a very, very good reason why they're not.

[Damien]

rob where is it now?

[rob]

Damien

that is a damn good question!

the last i saw of it the Adelaide Uni's Music Dept had it....then it seemed to disappear ( along with all sorts of other bits and pieces, such as MCI JH-110 8 track, VCS synthi, Neumann and AKG mics, Moogs, dBx comps etc etc)

for all i know it could be at the tip

[mfdu]

assuming no outboard pre's, are 12 channels going to give you everything you need, even for 8 A>D's? i mean, it doesn't have to be a 1:1 ratio of mics > AD.

i regularly track 10 channels to 'tools (8 console busses > 002 + 2 outboard pre's > s/pdif), and the 8 console busses could easily be -
1 = bass di (phase flipped) + bass mic
2 = kik mic + kik speaka-mic
3/4 = tom hi, tom mid, tom low
5 = snare
6 = guitar 1
7 = guitar 2
8 = scratch vox
(for the sake of example, presume room mics on 9/10 via outboard)

you see what i'm getting at - that's 11 mic inputs to 8 busses. doesn't leave you with much room for m/s jiggery-pokery, or alternate room mics.
not that everyone does it like that, but it is so satisfying to commit, and to get the phase right across multiple mics summing electrically prior to the A>D's.

of course, you 2x2 bal inputs to LR would give you the chance to mix into the LR which could feed A>D 9/10, but what about panning?

looks like you've got a big year ahead.

chris
mfdu

[graemeh]

graemeh wrote:
Latching talkback button, preferably illuminated!


every band that has ever been in a studio would have broken up before they finished their first album if talkback buttons were latching. there is a very, very good reason why they're not.

I should have been more specific - I meant momentary up, latching down, centre off. or similar - so you can choose whether to drop the band in it or not :-) I used to have one like that in the olden days. Now I use a dedicated fader on the desk, and believe me it beats sitting there getting control room folks repeating instructions to the studio (cause they insist on talking just AFTER you release the button) and me with a numb arm from holding the TB button down.
I saw one somewhere that latched if you held it for 2 seconds or therabouts - that'd work...

[jkhuri44]

chris p,

number of channels really depends...

our band tracked drums with 19 channels. drummer's got a real big kit, wanted heaps of detail, and seperation (im waiting for the old school 3 mic technique lovers to slam me here)..but my point being for different sounds, you need more or less channels...

but as you know for most purposes what you are planning is probz enough.

just my 2 c.

[chris p]

Thanks everyone for the feedback about number of channels.

I should probably plan for 16 channels, plus the 2 x stereo inserts gives 20 inputs.

Mockup here: channel strip (1-12), channel strip (13-16), subgroup strip, master strip

[mfdu]

just wondering - what is the design of the mic pre's you're using? and what EQ? the pad/phase/48v looks like a JLM inbetween . . .

and your AD convertors / firewire interface - what is the design?

chris
mfdu

[chris p]

Hi Chris

IF I build it (and "IF" is in capital letters), I'll almost certainly be using the JLM BAD design with 1:4 input tranny and dual JLM 99Vs for the amp sections. For the subgroup and master/monitor sections I'll again use 2 x BADs minus the front end and with JLM Hybrid amps in class A/B mode. The design has a insert loop between the two amp stages, so that is where the EQ/fader sits on the channel channels and the EQ/compression for the subgroup channels. I will likely MacGyver some relay operation for the 3 switches.

The channel EQ is sort of from a Neve 1073, in that it will be inductor based, sequential rather than parallel, and mimic the 1073 controls (renowned for their ease of use). The subgroup EQ will be Gyraf's EQ1A circuit minus the tube makeup gain stage, as I'll have the second section of the BAD for this purpose.

The AD stuff will be a 1U third party box that can just sit in a slot: I'm looking at products that take +4dbu balanced signals without any preamp section, so something like the Echo Audiofire (lower end) or Apogee Rosetta 800 (higher end). Naturally, I'm liking the thought of the Apogee, even though its 8/8 rather than the Echo's 12/12. We'll see.

Regards

[Howard Jones]

Hi Chris,

It would be helpful if the EQ sections had dB scaling on the cut/boost.

Howard

[chris p]

Howard - EQ scales will be printed on the panels - thanks for the suggestion.

Now for a more technical question. Whats a good cap for EQ filters? And is it desirable that all caps in the EQ be of the same type? The reason I ask is that many of the nanofarad EQ cap values can only be readily sourced as ceramics, while others can be MKTs, greencaps and/or metal film polyprops.

Or does not simply not matter provided the value is right (we're talking unpolarised types).

[chris p]

To partly answer my own question, here's an interesting site

http://members.aol.com/sbench102/caps.html

[chris p]

That guy seems to recommend paper and oil as the least distorting capacitors. Does anyone have have thoughts / experience with paper caps?

[rob]

i reckon there is a bigger picture to start scratching your head about.

do you have a schematic for the sections of this mixer?
what about pcb layout? In a large distributed unit like a console grounding is very very important. The whole flow of ground currents and reference points needs to be worked out.
how much space do you have for components...paper/oil caps are big

the low distortion of caps can be important, but in practical terms the differences between good polyester, polypropylene and paper/oil is very small. Is is unlikely that the distortion component of the caps would dominate over the distortion in the active components ( opamps etc )

Factors such as tolerance, tempco, physical size are important considerations in EQ design.

As a rule of thumb chose caps along these lines for audio paths:
Bipolar electrolytics for large values / output / interstage coupling
NPO / COG ceramics for small values ( less than 100pF )
polystyrene if you can afford them
polypropylene if you can fit them

and...remember from a pure analytical point of view all power supply filter and bypass caps are technically in the audio path as well....and who wants to go there!

Rob

[chris p]

Thanks for the sanity check, Rob.

do you have a schematic for the sections of this mixer?

Yep. Far from original, but yep.

what about pcb layout? ... how much space do you have for components...paper/oil caps are big

That's what I'm working on, and hence my query about caps - because paper ones can get very large indeed! I'm working around a 300mm x 150mm standard PCB board. Yet to see if I can get it all together, especially as I tend to work single sided.

In a large distributed unit like a console grounding is very very important. The whole flow of ground currents and reference points needs to be worked out.

I'm working with a ground plane design. I've pretty much got the signal and supply currents and voltage levels worked out, but admit I've not done the same with ground currents. In passing, having brought all inputs through to a nominal 0db (=+4dbu), the signal gets dropped to something like -50db as it passes through pans, faders, mixing networks and so on. Is that normal for a mixer design?

in practical terms the differences between good polyester, polypropylene and paper/oil is very small.

Yep. I see that on the sbench102 pages.

Again, thanks for your thoughts. This is a really informative exercise, even if it doesn't get past the paper stage. I'm learning a lot about how these things actually tick - that my intuitive concept of a pan pot, for example, was all wrong.

[chris p]

Question on bus (stereo) EQ.

Would you prefer one set of knobs that simultaneously controls both channels, or does the flexibility of separate knobs for separate channels pay off?

I'm talking mixing, rather than mastering.

All thoughts welcomed.

[rob]

i guess the first question to ask is do you really want an EQ over the whole stereo buss? This whilst not without uses is unusual.

If they are were ganged you'd want very good tracking, which can be hard to guarantee. Since you probably wouldn't be doing any radical EQ moves, what about stepped switches in 0.5 dB steps

rob

[rick]

my strong advice is to stop ...
and have a think about the metalwork/ size ergonomic format of this thing , long before you worry about the caps and resistors.
when you find out that getting the metal work made to order is going to cost you at least $5k , you will soon decide a better way to go is to buy an appropiate sized mid level console to "butcher for frame parts " ( say $2- 3k) .
that frame will set you up with size constraints about the number of channels , size of the channel strips etc etc.

there is nothing like a few real world limits to get you on track to what you really want/need.

of course if your a metal work wiz then just kep roll along in virtual world ,
just dont get caught out assuming the first part is the last part.

i have well and truly ran the custom console race over the last twenty years, in fact currently Rob is pondering just how to re-configure the tube vintage Pultec mixer i bought last year.
( you think i would have learnt by now !)

[Thirteen]

These guys are obviously dreamers... Each channel strip has 26 tubes.

http://www.vintec-audio.de/vintec/index_E.htm

[mfdu]

ooooh that's a beauty, huh? and forget not all the interstage transformers.
they claim internal operating temp of 38degrees - i find that hard to believe.
and how much fun would it be, re-tubing the bugger?

worth it? well, i don't NEED two kidneys, do i? and teeth are just excessive. i mean, do you really need them? what did god give us blenders for?

chris.

[Howard Jones]

I'm surprised to see that these guys still even have a website up. This product is pure vapourware - take a look at the News page. Hasn't changed in 5 years.

[jkhuri44]

what's vapourware? sounds funny, whatever it is...hehe

[mfdu]

Vapourware. Common IT term. indicates equipment that'll be available when hell freezes over.

Wiki sez :
"Vaporware is a software or hardware product which is announced by a developer well in advance of release, but which then fails to emerge, either with or without a protracted development cycle. The term implies unwarranted optimism, or sometimes even deception; that is, it may imply that the announcer knows that product development is in too early a stage to support responsible statements about its completion date, feature set, or even feasibility."
[url]http://en.wikipedia.org/wiki/Vaporware[/url]

Howard - i got so excited by the impracticality of all those tubes, i forgot i'd seen that console years ago, also.
shame. messed myself over nothin.

chris
mfdu