Building a DIY mixer?

  • Thread starter Thread starter diggy_dude
  • Start date Start date
diggy_dude

diggy_dude

Now 169% diggier!
I've made no bones about the fact that I'm continually dismayed at the poor quality of commercial products and the lack of integrity of the people who sell them. I'm so disgusted, in fact, that I'm tempted to try to build my own mixer. I understand and accept that this may take a long time and involve a lot of trial and error, but I've wasted a lot of time and money already and don't feel I have much to lose by giving it a go. I may even end up with a product that I can sell to others who remember and appreciate quality electronics, but I don't plan to pursue that aspect at this time.

I envision a very simple design with minimal processing. For example, designing good EQ filters and mic preamps is something I think would be quite beyond my capability, so I wouldn't attempt it. What I'm after is a basic unbalanced mixer. Here is what I consider a good basic feature set per channel:

PFL switch
mute switch
solo switch
gain pot
level fader
pan pot
pre- and post-fader inserts
phase inverter switch
one or more pre- and/or post-fader aux sends with send level pots

There would be a number of subgroup channels corresponding to the aux sends, with only a level fader per subgroup. The main section would simply be a pair of level faders. If I wanted a headphone output, I'd tap the main section to connect an external headphone amp. I'm not really concerned about having meters or clip LEDs either. I would mostly play it by ear when adjusting levels. Yes, the lack of these features would negate its marketability as a commercial product, but that's not my primary goal.

What I'm looking for in this thread is guidance in choosing quality components (op amps, pots, etc.) and a proven gain stage design and PCB layout for each channel strip and the main section so as to achieve minimum noise and crosstalk. It could be something as simple as pointing me to a website where I can find such a design. I'm hoping posters with experience in designing and building linear electronic devices will weigh in. I'm definitely not interested in the MBA mindset of cutting corners, reducing costs, etc. I want to know how you guys would approach this if cost were no object. Any advice would be greatly appreciated.
 
frederick started a full-on console design about six years ago . . . I think it may have killed him! :eek: It's buried somewhere in the Studio Building board.

If you want an easily upgradeable IC console, shop for an A&H MixWiz, v2 or v3. The chassis, knobs, connectors, faders, etc. are the most expensive parts, even if you want to toss the PCBs (each channel is on a strip), you'd have a PCB template to start with. Plus that way you could build individual channels while still having a functional mixer in the meantime, do one channel at a time.

It's not so much a matter of reducing cost as time. If you start out to fab your own chassis, you'll be devoting a massive amount of time solely to that task, then doing PCB designs to fit the console.
 
A few other comments:

There really isn't a lot of difference between a mic amp and a following stage amp, except the following stage probably isn't differential and it doesn't have to be super low-noise. But you are still using opamps or something similar. So don't discard the goal of a mic amp, because it makes the mixer far less useful.

An easy way to do mic amps is with low-noise instrumentation amps. These have the advantages of relatively few parts, low noise, and high maximum gain. Their disadvantage is the gain control must either be a stepped switch or a reverse-log pot, and it's very hard to get linear gain change in a pot over a 60dB range. So most inamp-based preamps you see used stepped switches, which are pricey.

The other three ways to get low input noise with a standard opamp input:

- use a really low noise opamp. You'd think this is perfect, but an opamp in differential configuration has series resistors that set input impedance. The thermal noise of those resistors is a function of their resistance. You don't want that to be too high, or input impedance to be too low. If you want super-low noise, it gets problematic.

- use a transformer. Works perfectly, but expensive.

- throw a discrete BJT front-end on the opamp. This is the most common in inexpensive low-noise inputs; it adds a bit of gain that makes the opamp noise not important (as does the transformer approach)

For DIY, I would pick the transformer approach because it works really well and everybody loves transformers. Cost no object, right? Jensen has several that are nice and about $65 each :D But space is a little tight inside the A&H for Jensens . . . Neutrik has a 1:4 transformer that is really small and reasonable for professional mics (150 ohm); if the source impedance gets too high it doesn't have the best bass response (so save the EQ!). You could hack that into the A&H, upgrade its opamps & caps, and have a nice channel strip for relatively little work.

Crosstalk is a function of layout. Use upgraded grounding (fat pipes in star ground) and that should improve.
 
Thanks, mshilarious. I think $65 for a single transformer is a bit pricey, so some compromises will have to be made. I'd prefer FET circuits, but designing a circuit with all discrete components is a bit beyond my skill level. I think low noise op amps would be a good choice because most of the design work has been done by the IC manufacturer.

I would just Google up some schematics, but I'm not competent to tell a good design from a bad one by looking at the schematic. That's where I must defer to professional (or at least experienced) opinion.
 
Well you can't hardly walk across the interwebs without stepping on mic amp schematics, and if you don't mind you can have different circuits on different channels. For example you probably don't need 16 ultralow noise inputs, maybe 4 would do, that lowers the cost of transformers. You could do the basic opamp/cap upgrade on 8 channels to start, then you'd have eight decent channels functional while you worked on the other 8 custom channels.

Discrete FET inputs for a mic amp aren't common, but they do exist. Generally though, it's easier to get very low noise by paralleling BJTs, this is usually what you see in front of opamps.

If you liked you could have a channel or two of single-ended common source FETs in stages (perhaps after a long-tailed pair input), those will kick off quite a bit of desirable even-order harmonics. It surprises me sometimes that people either go straight to tubes from IC opamps and don't consider there are other cheaper, smaller, and more efficient schemes that can be employed for a "color" amp.

You could even do a submini tube design in the A&H, heh . . . that would be slightly tricky because of the need for a heater supply, can't remember what the A&H has for power rails beyond +/-15V . . .
 
I don't foresee ever needing more than 8 channels period unless some of them are subgroups. I'm not really gung ho for tube overdrive either. Not enough to have to replace tubes every time I turn around at any rate. After my recent experiences with gear, I'd have to say reliability is at the top of my list of criteria. That's the main reason I'd bother to build my own vs. buying a commercial unit.

I get the sense from your remarks that the choice of architecture (IC, FET, or tube) isn't overly critical as long as the components are of high quality. All told, I'm thinking op amps would be the most feasible as far as ease of design and assembly goes. Are there any particular brand names of op amps, faders, and pots I should look for? I got me a big fat Mouser catalog right here.
 
I think Mouser carries ALPS faders, if you are going straight DIY then get 100mm if they have them. Pots you would want conductive plastic for maximum reliability.

There's a zillion opamps you could try. Mouser stocks TI; OPA1612 looks interesting to me, but it isn't cheap and it's SOIC only. OPA2134 for anything but the input stage, where you'd need lower noise. If you want to go inamp, INA217, and use a rotary switch instead of a pot.

If I was going to go discrete anywhere, it would be on the headphone output, because no DIP-8 opamp can do the right combination of high current/high voltage, mainly because if they did they'd exceed the package's thermal rating. So an opamp + a discrete AB output is the way to go for crushing volume into any load, 16 ohm and up. Or you can use an IC power amp chip, that works too and is simpler.

There are also some Analog chips I like, but you have to get those from Digikey (or Analog directly for some of them). And Linear has some nice stuff too . . . really, there are many embarassingly good ICs to choose from.

For an EQ I can live with simplicity; 12dB/octave high/low shelving preset at frequencies I like (I would go 125Hz and 8kHz), and fixed notch filters at 300Hz and 3kHz keeps me fairly happy. Or add switches on the midranges for a selection of four frequencies (I would do 300/800 and 1.5K/3K) and you can avoid the complexity of full parametric.
 
Your features are all found in almost any old-school (but solidly built) analog mixer. Why not get one of those, and tweak it out to a fare-the-well? You would have more features, better sound, and more time to do other things.

Just a thought...
 
Your features are all found in almost any old-school (but solidly built) analog mixer. Why not get one of those, and tweak it out to a fare-the-well? You would have more features, better sound, and more time to do other things.

Just a thought...

That's what mshilarious is suggesting with the Allen & Heath mod. I may go down to Horn Trader and see what they have this weekend.
 
mshilarious,

What do you think of the 6021 and other sub-miniature triodes?
 
Sorry to jump in so late, but if you want to go DIY, but don't want to reinvent the wheel quite so much, you can find a ton of projects w/ PCB's or layout guides and directions at Prodigy-pro. Meta-Meta: Look here for overview..

E.g. New Design 1176 Boards, 1108 BACK IN STOCK! + 900 Power Supply Boards etc etc etc. I think Craig Anderton somewehere along the line had a DIY mixer article (the summing network, pan, etc) and e.g. PAIA sells a four channel parametric EQ PAiA - 6760K Parametric Equalizer Kit, Parametric Equalizer, 6760K
 
Thanks for the tip, Jinn. Paia is located right here in OKC, as fate would have it. I don't really need an EQ. I've got a decent Symetrix unit made in 1986. I've bought a 424 since I posted this thread too, so I'm not really hurting for a mixer so much now either. Thanks just the same! :)
 
PAIA is way cool. I've built a bunch of the Electronic Projects for Musicians effects, and the Parametric EQ. Also built the dual spring reverb, I get some hum from it that I need to troubleshoot (I'm sure it's my build) but, it sounds really really good for a spring reverb. I bought one of these too, plan to stick a transformer in front of it and see how it sounds. It's not a PAIA kit per se, but some couple in Hong Kong.

PAiA - K98 Preamplifier, Building Block Kits, K98
 
There was a studio spring reverb unit out around 1979 that claimed not to have all the "boing" and slapback of a regular spring reverb. The schematics are on the Web somewhere.
 
There was a studio spring reverb unit out around 1979 that claimed not to have all the "boing" and slapback of a regular spring reverb. The schematics are on the Web somewhere.

That's the PAIA, (designed by Craig Anderton) it uses two springs, but wires them up out of phase or something, I don't remember exactly the theory. It does sound pretty full without the "sproinginess" for lack of a better word. If I can solve the hum problem (I wired up a funky power jack, so I've been reluctant to get in there and e.g. try it with batteries to see if it is power supply bleed)

you can get the schematic from the PAIA website.

http://www.paia.com/ProdArticles/hotspuse.htm
 
No, this was a different brand. But if the PAIA is just as good, I'll check it out.

I'm not looking for a mixer anymore. Just got a little TEAC M-09 in good condition.
 
Back
Top