12AU7 Mixer

[RicharD]

I have another little project on my plate.  What I have is a rack of metallic object used for industrial percussion that need to be amplified.  The plan is to use Piezo buzzers wired as contact pickups, sum them through a little mixer, and send a singular signal out to the main console.  Yes this could be done more economically using a cheapo Banjo Center mixer but where's the fun in that?  There's no need for tone controls.  I just need a simple 8 to 1 mixer.  Here's what I scratched together.  My plan consists of 5 12AU7.  Each triode is set for 5.5mA.  I'd like to use a singular B+ source and inset a 1k resistor atop each stage (Vdrop = 5.5V).  Each stage has a gain of about 10 or 11.  My concern is the summing.  I may need more recovery gain in which case I'll redesign using a 12AX7 or something.  Does anybody see any glaring mistakes?  Comments and criticism is always greatly appreciated.

Thanks!
-Richard

[Tone Junkie]

Richard my friend you have some of the most interesting projects, always working on something new I love it.
Bill

[firemedic]

Man you must have a bunch of those U7s laying around.
I am no circuit designer but you may want bigger input grid leak resistors?. I think piezo pickups are pretty high-Z, & you may want to match that up.
I have a Kevin O'Connor book, "Tonnes of Tone", which suggests this mod if you're amplifying an acoustic piezo P/U. I believe he used a 10M resistor for the input ground. But I've never tried it.

[bibi]

Well I'll second that I always like looking at your projects, cool to see something different.
This one I'll bite on because I've done similar work albeit without the tubes.  The piezos need
some sort of buffer quite close to them if you want to use them as mics I've found.  Otherwise
all the treble rolls off very quickly.  They don't need a lot of gain after that especially if you're
going to use them for percussion.
Otherwise, I think your summing stage doesn't have near enough gain-- I don't have the math
in front of me at the moment, but I think with 8 stages you'll want that AX7 probably in an
active mixing stage. 

[Backwoods Joe]

bibi, one thing you may want to keep in mind.... back in the day of replacing piezo's in Simmons pads, i used to check each piezo with a scope before installation and i think recall the best piezo's putting out a whopping 5Vp-p! joe

[RicharD]

The piezos will be directly attached to the metallic objects.  This circuit is not going to work do to excessive loading at the summing point.  If you consider the 8 - 100k summing resistors in parallel, that puts a combined load of 12.5k on each input stage.  Oops!  I scratched this thang out while at work and didn't put much thought into it.  Must do some research, or opt out with opamps.

[Fresh_Start]

Richard - There's something else in TUT about an "anode mixer".  Simply tying the plates together.  You can use a common cathode resistor/cap network or separate ones for each triode.  If you combine, don't forget to reduce resistor values accordingly.  Plate resistor value too IIRC.  O'Connor claims that the grids are "perfectly isolated" and that there's no attenuation because the grids "represent extremely high impedance sources to the signal sources."  He also says that there is a noise increase over a passive mixer, so  I have no idea if it would work with 8 triodes combined. 

Maybe you could do 4 + 4, using passive resistor mixing for the two quads?

Just a thought when I really should be asleep...

Chip

[PRR]

> more economically using a cheapo Banjo Center mixer

No; the $49 Bee-ringer has 10K inputs, you want >1Meg for piezos.

> Each triode is set for 5.5mA

Or 55mA. That's more than a whole Champ-amp.

Bogen would do the whole 8-1 mix in 5mA _total_.

Seriously. Plagiarize Bogen's or Altec's 4-input mike mixer, but go to 8-in and use 12AU7.

You sure do NOT need separate decoupling on each input (unless you must meet some absurd leakage spec). All 8 inputs, and the recovery stage, can go on one B+ node. (Your schematic tool may have cut/copy, but your soldering iron doesn't.)

Why 12AU7? Got lots? 12AX7 could be "better" by not needing cathode-caps to get unity gain through the mix network.

> a combined load of 12.5k on each input stage.

Uh, no. Each input 12AU7 has that way-heavy 15K, parallel with 250K||100K, or 71K AC loading.

The mix-amp's grid looks into something like 107K/8 or 13K, which is fine.

Anode mixing by tying plates together sucks: low output for high distortion. Also no individual gain control, unless you put it before the tubes, which is problematic (but perhaps acceptable for this use).

[RicharD]

I have a proverbial bucket of 12AU7's.  Unfortunately that's not the case with 12AX7's.  I do have a few tubes whose quantities can be measured in gallons such as 6FQ7, 12AT7, 6AU6, 6BQ7, & 12SJ7.  The Ampex MX10 uses EF86's and that could easily be adapted to 6AU6's or 12SJ7's.  You are right, I do need to do some homework and plagiarize something to my own end.  Spare time has been sparse this week. 

[RicharD]

So I measured the output of a piezo this morning.  Z is greater than my meter can read in quiescent state, and looks like it might be about a meg when when attached to a beer keg and struck with a drum stick.  Looks like the Vout is less than 1VAC.  This means good ol 12AX7's will be the way to go.  I researched some of the suggested mixer schematics which pointed onto this path.  It's not a straight up copy of the Altec 1567A although I did find a lot of clues.  I don't really need or want a tone stack and that Altec tone circuit is something different.  Kind of a weird Bax I don't really understand.  I'm more comfortable with a cathode follower output stage so I stuck with a direct coupled 12AU7 design.  I've nothing to do today so I'm gonna start bread-boarding this although I'm certain I'm short on 500k audio pots.  I'll work around that with regular resistors for 6 of the channels.  Here's the schematic.

[DummyLoad]

we have a reel each of 10M and 1K5 resistors. if you're short on 220K i have plenty. i probably have the 500K pots as well. i'm out of 470K CC & CF - lately i've been using a stash of 510K for that.

hope did you didn't dismantle the janitor...

no need to breadboard that ckt. just put it in a home - it'll work fine.

[RicharD]

Too late.... it's built & it works pretty darn close to expected.  There's a little more loading in the power supply than I had hoped for.  B+ is 20V lower than I designed.  I adjusted the dropping resistors.  I do have a pair of mo bigger transformers.

Ran short on .47uF caps so I subbed .33's, them big honkin yellow caps from MC Howard.  I had enough pots -n- jacks to complete 2 channels.  The other channels I shorted the inputs and loaded the outputs with 470k resistors.  I grounded what would otherwise be the wipers for these channels too.

I didn't hafta tear down the Janitor.  I have 2 - 5 bottle projects wired on Zilla at the moment.

[PRR]

The _DC_ impedance of a piezo is infinite.

Wire piezo in series between tone generator and 'scope 1Meg input. Sweep up from 2Hz; or sweep down from 1KHz. You get something like flat 1KHz to 50Hz, then falling 50Hz toward zero.

If that corner is acceptable low-cut for your keg and speakers, then 1Meg is fine for grid resistor. If you get a 500Hz low-cut and have a baritone keg, you may want a higher resistor.

Do you really need cathode caps?

Why have a cathode follower then put 500K after it?? You don't drive a transformer from a 500K wiper (125K source at mid-loss). Transformers are rarely over 10K. 100 feet of cable is closer to 3K at the top of the band. I would put the master in the mix-amp grid circuit. I suspect the mix-net's self-noise amplified by one low-Mu stage won't raise the sound-board's hiss.

Mix-net values OK. For personal taste, I would change 12AX7 to 100K plate resistor, use 250K channel pots, 270K mix resistors. That gets sum-node impedance toward 43K. A 100K master pot is OK loading on the mix network. Resistive hiss will be comparable to mix-tube grid hiss.

Gain is, say, 25 through 12AX7, 1/10 in mix-net, 10-15 in 12AU7s, 1:1 iron, about 30. Your sound-guy should tolerate 1V signal. That means inputs as low as 30mV can be full-roar. Seems ample without cathode caps.

[RicharD]

I made your suggested changes and you are dead on, AVmax = 30.  Right now I do not have a 1:1 output transformer connected.  I'm running HiZ right off the wiper.  I'm kinda thinking I don't really even need an OPT.  Hum gets unacceptable with the MV full up but it just fine set at 3 oclock.  With your suggested changes, Vtube is right at 2/3 B+ which I believe is desirable. 

[DummyLoad]

 

i thot this was a 12AU7 mixer.