Tube summing amp

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

stitch-o

Well-known member
Joined
Mar 3, 2007
Messages
629
Location
MA/PA in (not so) good ol USA
Im tinkering with a tube summing amp circuit ala TubeTech.
I found this:
http://users.otenet.gr/~athsam/tube_line_preamplifier_progect_1_English.htm

I think I can hack it apart but does anyone have any dedicated tube summing amp circuits for pro-audio?

Cheers!
 
why that circuit?
I don't see anything special there.
maybe the delay/mute switching.
it is not summing is it?

whats wrong with passive summing with optimized make-up gain stage?

is there a lower noise way to do it with valves? will you lose 'tubeyness' if so?
 
Hello there,

I have a Tube Tech SSA 2A. No cct dgm though I would really like one.
But I love this box. it works really well for my set up.

I use Akai DR16 PRO hardware recorders.
These are modular in design, and contain an internal mixer.
Often I compile Bass and drums on one machine, and vocals and everything else on another, (although this could be further subdivided to more machines if it was of value).
Each stereo output then goes to seperate processors (compression etc) and summed with the SSA 2A.
How hard the valve summing amp is driven affects the final tonal colour.
The output is saved as 88.1 files and completed with software limiting and format conversion to the end product (usually CD).

All the best,
aomahana.


 
> a tube summing amp circuit .... I found this:

That's not a summing circuit.

 
PRR said:
> a tube summing amp circuit .... I found this:

That's not a summing circuit.

Thanks guys. You sure know how to call'em...
I get that that link isn't a tube summing circuit.
I couldn't find a schematic for one anywhere so I thought that
the link for a tube line amp circuit might be a point of departure.

I'll go look elsewhere.

Thanks all.
 
@shabtek
Passive summing just doesn't interest me.
There is a reason big desks don't do it like this.

AFA tubes and why - I like the vacuum tubes ability to deal with complex
audio material. Plus, my idea is to make a switchable to a pair of solid-state summing amps as well as the tube sum amp.
the active summing section will be built after 16 channel input.
Maybe some other tweaks like mute, insert on 2mix and transformer balanced, split stereo outputs, etc...

I just need a starting point, frame of reference to go and research and learn.
I'm not as smart as most of you. I am made of kool-aid for christs sake!


jensenmann said:
go to kubarth.com, look at "GDIY files" and then "vintage german modules" for the TAB V75 schematic

THANK YOU!!
V75 (or REDD 51 was the other schemo) is just what I had been searching for.
Now, I need to learn German...

CHEERS, jensenmann!

edit for questions (if anyone has a minute and more brains than me)
sorry there are so many:

(start with obvious) The V75 is a 4 in x 1 out?
the "1:10 V372", that is referring to a mic to grid, input transformer model from a V372 unit like Haufe?
Could these 'input stages' be bypassed for line level input or is it already line level load input as is?
How much gain for each of the 4 x tube 'line' amps (tubes 1-4)?
How much gain is produced post-summing point (tubes 5-6)? These are wired as inverting amps (I don't think so but I'm often wrong)?
I count: 6 X tubes (ECC88?) and 5 x Audio transformers. Are 104-107 chokes?
I guess by modifying the summing point, more lines could be added...

spinnin' and cookin'

THANK YOU AGAIN, JENSENMANN!

 
that v75 looks like a passive 4:1 bussed together through 200k with make-up gain following.
lots of plates loaded with chokes too. not very budget friendly

would a tube op-amp work as an active combining amp ACA?

http://www.dogstar.dantimax.dk/tubestuf/graphics/phono-s.gif

I have this home etched and half built for about 10 years now.
maybe I'll do something with it now.
 
The V-75 looks to be just what shabtek said - That is passive mixing with 200K resistors.  And a mono-one to boot.

There is a reason big desks don't do it like this

One of the biggest reasons is Loss.  You would get a whopping amount trying to passively mix 56 channels.  The older tube mixers I've seen stay lower on the number of channels and use passive mixing - either resistive or via coils. They all need a makeup gain amp.  Resistors are considered flat through the audio band so that eliminates one problem.

Question is how many channels do you want?  NewYorkDave posted an 8(easily made to 16 if wished) channel low impedance passive stereo mixer, fully pannable.  It's still posted.  You can then use whatever tube amps you wish for input gain.  Like hybrid?  shouldn't be a problem - all the mixer asks is that it be fed with low ~500/600r source which covers most tube rigs and vintage SS.  If your SS choice has lower output impedance it probably won't be that big of a deal.  Make-up gain is your call and could be switched from one amp to another. 

The NYDave design would essentially work like the V-75 but with the addition of fully panable stereo outputs and the flexibility of using whatever input and make-up gain amps you wish - i.e your favorite mic pres padded for line level use. 

To actually have the tube doing the mixing - twin triode mixing works fairly well but you're limited to 2 inputs only.  You can stack several twin triodes together - 2, 4, 6, 8, etc.  but the outputs of each would have to be combined to either mono or stereo - and resistors are usually the best choice.

This page is a good read for understanding the trade offs and compromises with a number of design approaches.

http://www.phoenixcomputerlabs.com/All-About-Mixers/Tube-Mixers1.html




I am made of kool-aid for christs sake!

So was I - 1984 Thanksgiving Day Parade - I sat on a float for 2 hours in the drizzling rain impersonating your avatar.  The suit was ginormous and I had a hard time walking without toppling over. Of course I could only use one hand to hold onto the float rail - every one of the thousand screaming kids without fail were waving & yelling - "HEY KOOL-AID!!!  COME ON DOWN!!!"  My waving arm felt like a limp spaghetti noodle -  :D


 
 
http://www.phoenixcomputerlabs.com/All-About-Mixers/Tube-Mixers1.html
appears to be lifted from John Broskie.

what about 'virtual earth'?
 
what about 'virtual earth'?

Haven't tried it.  Seems like there were some mixer design forum threads discussing this.

This article touches on it briefly - some stuff from Steve Dove

http://www.customtubeconsoles.com/design.html
 
I figured out what was bothering me.

> tube summing amp
> vacuum tubes ability to deal with complex audio


The alleged "complex ability" happens with low feedback tube amps.

A summing amp worth the wiring must be a HIGH feedback amp.

The linearity, overload, and clipping behavior won't be much different than a high-feedback silicon amp.

Plus the gain and impedances in modern audio consoles/mixers are scaled for transistor hiss. Tube hiss Voltage is very much higher. (Tube hiss current is lower, but we don't get any advantage from that fact.) To get mix-hiss similar to a 5534 summer you want about gain-of-4 in front of the summing network. That's a lot of booster amps. Sticking with conventional ~~10K summing, they must be pretty beefy tubes.

Much depends on how many channels and how bad you "need" the main advantage of active combining. With active you may disconnect summing resistors with "no" change of gain.

Gain really changes; if you have 10:1 of excess gain then change is like 9/10 or 1dB which is marginally tolerable.

All assuming nominal unity gain from one input to output.

For 4 inputs you need gain of 4*10= 40, which is not too hard. one stage 12AX7 does that. It also needs to drive the ~~10K feedback resistor, and whatever load, which a 12AX7 won't do; you want at least one unit 12AU7 and may be looking at a power tube follower. (Pentodes have more gain and more hiss, a bad deal.)

For 56 inputs and 0.5dB error you need gain over 1,000. No single tube stage does that. To come close you must trade-off bandwidth (a bad deal). Two stages would give LF gain of 1,000+, but a double-rolloff, 180 deg phase shift which will find a way to oscillate at the top of the audio band. I won't say the Philbrick K2-W tube op-amp is the greatest, but it's pretty good. It was unity-gain stable and 300Kcps gain-bandwidth. Which means it would meet the spec for 56 inputs and 0.5dB error.... only up to 300Hz.

Here's what you do. You limit yourself to NOT disconnect summing inputs mid-track. Now there is very little difference between passive and active summing. Run a high gain volt-amp and buffer. Apply just enough NFB ("active summing") to bring the gain down 3dB-15dB. Work inputs hot to overwhelm tube hiss. Be sure your buffer stage is working easy, nothere near clipping.

 
aomahana said:
I use Akai DR16 PRO hardware recorders.
These are modular in design, and contain an internal mixer.
Often I compile Bass and drums on one machine, and vocals and everything else on another, (although this could be further subdivided to more machines if it was of value).
Each stereo output then goes to seperate processors (compression etc) and summed with the SSA 2A.
How hard the valve summing amp is driven affects the final tonal colour.
The output is saved as 88.1 files and completed with software limiting and format conversion to the end product (usually CD).

A bit off-topic, but this is a very interesting workflow.  Buss mixing out of necessity, love it!
 
I use this "zero-field"-ish summing amp in the G16 mixbox - transformer is a miniature 10k:10k type, feedback helps to keep levels low across trafo. Works well in real-life, though it's far from theoretically correct...

Mix resistors are 10K.

Yes, amplifier stage is the same as for the G-pultec.

Jakob E.
 

Attachments

  • G16_mixamp.JPG
    G16_mixamp.JPG
    31.8 KB · Views: 351
I love you guys.  ;D

I also am digesting this:
http://www.tubecad.com/april_may2001/page23.html
and this
http://recording.org/tech-talk/44410-virtual-earth-tube-mixer.html

I'm knocking up a couple of ideas now and readingreadingreading...

CHEERS!

edit - @ Gyraf:
So in your application, the tube summing circuit is configured as non-inverting or a virtual ground type amplifier?Im going to take a stab and say it isn't inverting as I believe the amp in the G-pultec is of the non-inverting sort...just a tube line amplifier.

Cheers!

edit 2 - I've also found some NewYorkDave sketches for study.
That guys is really onto something...
 
For summing of many channels there´s always the option to group a number of channels together into one summing amp and later sum the groups together in a second summing stage. This helps keeping many tradeoffs small but obviously add complexity and noise.
 
jensenmann said:
For summing of many channels there´s always the option to group a number of channels together into one summing amp and later sum the groups together in a second summing stage. This helps keeping many tradeoffs small but obviously add complexity and noise.

I did an inspection of the sundry approaches to signal combining back in the '80s and noise is kind of a red herring in that analysis. One microphone amplified to nominal level will define a noise floor significantly higher than even typical electronics of a few decades ago. So it isn't a large concern to worry about adding noise X dB below some other dominant noise source.

That said combining topology still matters.. inadequate loop gain margin in high closed loop gain approaches degrades final linearity, and introduces phase shift if stability compensation is not relaxed in consideration of the high closed loop gain.

Using sub groups and a later combining stage is not a bad approach. Just like any topology, it ultimately depends on the execution and details.

JR

Note: all things equal, the noise of 4 sub groups, later combined into one, is "not" higher noise than combining all the sources in one amp, since the 4 sub groups individual input noise is incoherent and adds incoherently. 16 channels in one virtual earth sum amp will deliver 17x that ein. Four sub groups with 4 inputs each will combine to SQRT (4x 5^2)= only 10x . While these are both far below that one microphone noise floor so academic. Distortion and phase shift will not be incoherent but the sub groups have effectively better loop gain margin, so improvement there too. Of course more parts and complexity, and cost. 
 
jensenmann said:
For summing of many channels there´s always the option to group a number of channels together into one summing amp and later sum the groups together in a second summing stage. This helps keeping many tradeoffs small but obviously add complexity and noise.
Actually, cascaded mixing improves the noise performance because separate mix amps combine noise quadratically (3dB per doubling number of stages) when a single mixing stage sees its noise increasing linearly with the number of inputs (6dB per doubling of inputs). Worth considering when the number of inputs exceed 64 (36dB noise gain).
 
stitch-o said:
RE: cascading summing stages:
At how many summed tracks does this really become an issue/solution?
32? 64?

cheers!

The answer is it depends...

Ignoring the noise for now, since as I've mentioned that is dominated elsewhere in the signal chain, there are two other deviations from signal purity driven by noise gain (number of inputs). For simplicity we can combine these into one and just call it  error. To make a long story short, negative feedback makes an imperfect open loop amplifier with lots of spare gain into a more perfect closed loop gain amplifier by trading off excess open loop gain to linearize the transfer function.  The purity of this path using NF is the open loop error reduced by the ratio of open loop to closed loop gain, aka loop gain margin.

The problem simply stated is a virtual earth sum amp with tens of inputs directly reduces loop gain margin by that noise gain. If the closed loop gain consumes all or most of the available open loop gain, there is little left to force the signal to follow the negative feedback. Since we routinely operate solid state discrete mic preamps at similar gains, this isn't impossible, it just requires some design attention.

To keep from writing the full book on this,,, just like a mic preamp gain block, a large sum amp can be decompensated in anticipation of the high noise gain, while buses with inputs that can be added or all removed, we will need to be unity gain stable for that situation.

To answer your literal question, it depends on the topology, rather than a hard number we approach this by how many channels do we need to combine, then solve that problem.  The largest 2-bus I ever did accepted over 100 feeds (36x24 split with 3 feeds to L/R from each of the 24 monitor strips).

When using standard ICs look for decompensated of perhaps even consider repurposing mic preamp chips. When rolling you own DOA, consider adding more open loop gain, and less compensation lag (if channels hard assigned and you don't need to be unity gain stable).

Sorry that there isn't a simple "do this" answer... but it's not a simple question. Or as that old saying goes for every difficult question there's a simple answer, but it's usually wrong. 

JR 
 
Back
Top