DIY Summing Box

[radiationroom]

Has anyone here rolled their own DIY summing box? I am considering ditching my Mackie boards but have a ton of analog signals originating from "outside the box" that need to be summed along with DAW tracks. My budget does not allow for a 32 input summing system, so I'm considering rolling my own. Any ideas would be appreciated.

[synthetic]

I've been thinking about this too. There's really nothing to it. A bunch of inputs into some resistors, then a gain stage. Use a 1272 for a Nevey sound, or some other preamp to get it back to line level. Here are some schematics:

http://www.forsselltech.com/schematics.shtml

[synthetic]

NOOOooo, please don't get me back into DIY. Months wasted troubleshooting broken stuff....

[radiationroom]

NOOOooo, please don't get me back into DIY. Months wasted troubleshooting broken stuff....

Oh but YES we are going to get you back into DIY!!!

FYI - I used to be a board builder at Honeywell/IPC. Part of the job was to replace surface-mount components that were soldered onto multi-layer boards when something flunked one of the testing proceedures. Building stuff on Vectorboard is child's-play in comparason! :Coolio: :grin: :Coolio: :grin: :Razz:

[Bob Olhsson]

Passive summing is pretty easy although the build-out resistors will drop the signal quite a bit. Old consoles summed a group of 4 into a gain stage and four groups of those into another gain stage. The object is to keep the signal level well above mike level in order to avoid noise and crosstalk.

API used active summing in groups of eight inputs. Active summing calculates the gain but lots of folks don't realize you still need to Christmas tree the gain structure in order to keep the summing point at a reasonable line level.

[malice]

Passive summing is pretty easy although the build-out resistors will drop the signal quite a bit. Old consoles summed a group of 4 into a gain stage and four groups of those into another gain stage. The object is to keep the signal level well above mike level in order to avoid noise and crosstalk.

API used active summing in groups of eight inputs. Active summing calculates the gain but lots of folks don't realize you still need to Christmas tree the gain structure in order to keep the summing point at a reasonable line level.

I'm in the process of building a suming box faithfull to that concept.

I was first in the verge of doing passive summing every 8 voices, then I noticed you were mentioning an amazing custom console doing this every 4 tracks.

This made me wonder about doing both and make test in order to compare crosstalk vs signal to noise ratio in order to chose the best solution.

btw, very large console are often summing every eight tracks. At some point, there is no way to handdle the noise floor by buffering every tracks one by one.

I'm talking VERY large console.

Crosstalk vs noise floor. Eternal problem..

Anyway, I let you all know the results of my tests.

malice

[MacGregor]

I've been thinking about this too. There's really nothing to it. A bunch of inputs into some resistors, then a gain stage. Use a 1272 for a Nevey sound, or some other preamp to get it back to line level. Here are some schematics:

http://www.forsselltech.com/schematics.shtml

Yep, this (passive) design is pretty simple and works and doesn't cost much.
As Bob mentioned, if you need more channels proper gain staging is a must,
so you have to add a couple of decent pre-amps to the cost calculation.

Mac
.

[malice]

btw,

my guess is that I wouldn't be surprised that summing every 4 tracks is about right when panpots are involved and no less than 16 tracks is cool if you sum by left/right pairs.

i'm not saying I have made necessary test yet, but lest say it's an educated guess.

More to come

malice

[malice]

one ore thing.

A good class a opamp costs about 50$, minimum. (except if you doin it discrete, by yourself, but make sure to buy a good bunch of transistor to pair them right)

for 16 tracks, you would need, if you are using gainstaging every 4 tracks : 10 of them.

that is 500$. plus the panpots and the faders. like 160$ easy.

Plus a couple of trannys for the output. Like lundahls (you do wanna something better than a Mackie, right ?)

So add 160$

Case 80 $

switches : you might wanna mute unused channels, right ? Let's say 30 bucks.

damn, you will need a PSU. Let's say you are a lazy SOB and buy a +/- 24 volts from Power one at 120$ ...

Are you starting to get my point ?

What do you want for master fader in your fancy DIY summing box with no solo bus ?

Bottom line is : DIY is for the guys that prefer soldering that going fishing during sundays.

It is not necessary a good bargain, especially if you're taking in account the time spent soldering and designing.

That said, pick you poison, it might revealed to be fun ;)

malice

[radiationroom]

API used active summing in groups of eight inputs. Active summing calculates the gain but lots of folks don't realize you still need to Christmas tree the gain structure in order to keep the summing point at a reasonable line level.

Does the API circuits use "virtual ground" summing or "common collector load" summing?

[malice]

Does the API circuits use "virtual ground" summing or "common collector load" summing?

If by Virtual ground summing you meant: "negative feedback", yes, I would bet this was what API used.

"common collector load" provides no voltage gain wich would be strange if you need more than 0dB gain.

Damn, I'm becoming a Geek

HELP :D

malice :D

[Bob Olhsson]

I never heard of the latter so I'd assume the former. Haven't looked into this stuff in 30 years.

[Kenny Gioia]

Why don't you take my summing test to see if it's even worth it?

http://thewombforums.com/showthread.php?t=7264

[DaveC]

reading about this christmas tree architecture, analogue summing sounds like a really nasty thing to do to your audio!!!

I guess the extra 'space' of an analogue mix is precisely due to the audio channels being grouped ad-hoc, and these groups of 4 or 8 being independently distorted (using 'distorted' in its generic meaning of 'changed') before being summed, and then the further groups combined and again processed by their gain stages...

Presumably this type of stereo bus architecture could easily be modelled ITB if anyone was bothered?

[Bob Olhsson]

Properly done analog summing sounds wonderful.

What hasn't been modeled, are the subtile differences between analog channels. I think these help create a little more contrast that speeds up getting a good balance. It you match everything up to a tenth of a dB., you won't hear more than a very subtile difference however putting a mix together through analog summing seems to get better results for lots of folks.

[Kenny Gioia]

Properly done analog summing sounds wonderful.

What hasn't been modeled, are the subtile differences between analog channels. I think these help create a little more contrast that speeds up getting a good balance. It you match everything up to a tenth of a dB., you won't hear more than a very subtile difference however putting a mix together through analog summing seems to get better results for lots of folks.

That's why I can't get into "Tape" or "Saturation" or "Channel" plugins.

These are things I never "applied" before. They were just there. I couldn't get "rid" of them.

They were way too subtle (for me) to find usable.

[groovenagl]

hi there,

unfortunately the files are not available anymore. so - since the test is more than 2 years old and kenny does not accept PN's anymore, would someone be so kind and post the answers?

i've been playing around with a "novice/trial" setup by having 4 stems summed into an SP828 and found it to have depth, esp. for the fx-return as well as have a more accurate imaging compared to the ITB mix (logic)

currently comtemplating to have someone do the DIY box for me (passive) + some nice makeup-gain hardware i'd be very interested get to know if those who picked the ITB mix were right..

thanks!
tim