Summer (NOT the season)

[chris p]

Hi everyone

My DIY brain has now turned to a summing box, thanks to Wez's Astor distribution amps that means I have 1:1 trannies coming out my ears. Basic idea is to sum 4 stereo stems, with an active makeup stage, send it out to an external FX loop (eg eq and compression across the summed stereo bus) before returning it for some metering and then out.



Big question (with a special nod to Rob, 'cause its his VU buffer design) - how should I get the L+R and L-R wired going in to the buffer PCB. Do I need a summing resistor in there?

Any help appreciated.

[astrovic]

Chris - this is the second time now you've embarked on a project that I've been about to start!! I have very similar plans in the pipeline - nowhere near as advanced as yours though.

Do you visit Prodigy Pro often? If so, you might have seen this recent post :http://www.prodigy-pro.com/forum/viewtopic.php?t=18595&postdays=0&postorder=asc&start=0

There's some discussion there on the 2nd page about the need to use some 1k resistors as part of a "mono" switch. I'm far too noob to be able to fully think through how that discussion might impact on your project, but it might lead you in the right direction. Thought I'd mention it, anyway.

While I'm at it, I do have a couple of questions and comments:

1. Why are you using the 1:1 trannies? Is it a bit of "cos I can"?
2. Do you know what you intend to use for the make up gain amp?
3. I like the send/return concept. As Krusty would say, "I gotta steal that bit"
4. What software do you use to lay out your designs?

Sorry to bug you with questions when you're looking for answers. Best of luck with it and thanks for putting it up there.

Chris (L)

[chris p]

Thanks for that thread, Astro - it seems there's some difference in views over at the lab as to whether a summing resistor is actually needed. I might need to do a little suck it and see on this one.

(1) 10 trannies in a box does seem excessive, doesn't it? The thing is going to weigh a ton. But yeah, pretty much I can do it so I will do it. It also provides electrical isolation for the signal (that's what they were used for originally) which helps with noise. I'm also conscious that the jury is still out on analogue summing boxes, and so want to try to make something that will also add some element to the mix at the same time - in effect having two mix "flavours" I can use. Hmm, I'm not usually into ironing.

(2) Until I bought Wez's Astors, I thought I'd use a cut down JLM Baby Animal (no switches or input tranny) using his Hybrid opamp. Now I might just use the discrete transistor amp stage that came in those amps.

(3) In most mix boxes, you don't need an external loop because there's no VU metering in them, but if I want to add the metering into the summing box, I need the loop if I want to meter the processed mix (which of course IS what I want). It also allows me to just patch a stereo signal in for metering if that's all I need to do. The downside is a couple of patch cords if I have no external loop.

(4) I use the layout function in Filemaker Pro, which I already use for other work. Its essentially just a basic drawing tool, nothing too flash.

Regards

[rob]

you will need to mix the L and R signals via resistors for the sum and difference meters otherwise you'll just be dead shorting different signal lines together and that won't be good.

the 2K resistors are maybe a tad low in your mixing network, depends on how hard you want to make whatever is driving the summing box work.

maybe you should also have the summing point pre amps available as an output. This will allow you to try different makeup amplifiers readily.
You'll only be loosing about 12dB on level through the summing box, easily made up with a line level device.

Rob

[chris p]

Thanks Rob

you will need to mix the L and R signals via resistors for the sum and difference meters otherwise you'll just be dead shorting different signal lines together and that won't be good.

Duh - its always obvious once its pointed out. Why didn't I see that? Thanks for the info.

the 2K resistors are maybe a tad low in your mixing network, depends on how hard you want to make whatever is driving the summing box work.

Have I got this right? Resistance = (N-1)/(N+1) * impedance (N is number of inputs, 4 in this case - not 8, for the folks at home, because they are going to separate L and R buses). My value of 2k would give an impedance of just 3.3k, which is fairly low - I see what you mean. So for a 10k impedance, the resistors should be about 6k [edited when I realised my maths error], double that for 20K and half that for 5K. What's a good impedance value for this sort of thing?

I assume the same applies to the metering mix resistors? They could be a different value to the summing bus ones, but is there any reason to do so?

You'll only be loosing about 12dB on level through the summing box, easily made up with a line level device.

For those following at home, db loss is 20*log(1/N), so N=4 means db loss = 12.04db

maybe you should also have the summing point pre amps available as an output. This will allow you to try different makeup amplifiers readily.

I hear ya. In fact, I was chatting to Wez just yesterday with the thought that I might do just that, and build the Astor preamps into a separate box. The other enhancement I've added is a front panel switch for each stereo pair. This will be a DPDT per pair to short each L and R + to ground (or should it be to its negative) pre-transformer, in effect taking it out of the mix. Here's v.2

[astrovic]

Hey Chrisp - thanks for sharing your thinkings with us. I'm definitely playing along at home!

Another link to a thread on Prodigy Pro - this is a new one that is developing, and might have some thoughts that add to yours (or just confuse them!)

http://www.prodigy-pro.com/forum/viewtopic.php?t=21032

Note NewYorkDave has posted a design as well, which I'm going to study and see what it adds to the project.

Cheers

ChrisL

[harry]

what about this??......

http://www.prodigy-pro.com/forum/viewtopic.php?t=19426

i'm thinking of building one in a 16:2 configuration to go with my creamware a16 convertor box. I have bought some marinair carnhill transformers sp i will build a pair of 1272 / 1290 ish pres to use on the master output for make up gain, will probably make em patchable so i can use em for mic and di duties when they aren't in summing mode.

so chris, are you specifically after a summing box? or a monitor controller?

either way i'm curious to see what the difference will be between summing in the box, summing and mixing on a mackie 8 bus or sending a stereo bus outs to this summing box....

either way it seems to be acheap build and i believe there is a custom group order for screen printed and punched rack cases for this thing

[astrovic]

Which Chris was that question to? :D

If me - I'm looking at summing either 8 or 16 channels down to 2 (not sure which yet), via a make up gain amp (not sure what yet - the 1272 route is one option), then via an attenuator (a la a "Big Knob") an on to my monitors, with some Vu meters along the way.

Pretty simple, huh :roll: My current thought is to do it in stages, basically to suck it and see in order to learn, develop and improve. I have a 2 channel Paia mic pre that might become my "test" make-up gain amp - so I'll built a summing network around it, and once I have it operational look at tinkering with it to improve it. Something like that.

That PCB you linked to is probably overkill for my needs, but it does look pretty good.

[chris p]

For me, its definitely a mix thing rather than a monitor thing. With just 4 stereo stems I'll still be doing some basic mixing in the box, but the idea is to have the option of adding some "heavy metal" transformer to the final mix if that works for the track.

Harry, check out Rob's passive mixer thing (I'm sure he can post a link) as well, it seems very close to the NYDave summthing without the DIY hassle.

[rob]

chris, thanks for the mention

you'll find our unit here : http://www.broadcastproaudio.com/new.htm

regarding your unit
i'd punt on 5K1 resistors, this will reflect back an input impedance > 10K, which is a nice undemanding line input impedance

however, your design is complicated by the transformers. The attenuation through the summing unit is determined by the summing resistors and the effective termination ( or parallel load ) that the other channels place across the summing buss. With transformers the termination load is not a simple resistor but resistors in series with transformer secondaries. These transformer secondaries present a complex impedance and whilst over most of the audio band may be consistant, at the freq extremes may well deviate. The upshot of this is that through the summing network you may not get a flat freqency response. This will also be affected by the source impedance of what is connected to the inputs of each channel. Especially if you leave inputs unconnected or you dead short them, levels and possibly freq response will move around quite a bit.

It would seem to me that if you want to explore the world of passive summing that you'd keep things as simple and straight as possible.

Could you be better off just building a straight summer with resistors, no trannies, no metering, no make up amps. Then build a transformer box using your 1:1 transformers as a separate item, likewise with metering and makeup amps.

I know this is more boxes, connectors, dicking around, but it leaves open all options and you can explore what happens with transformers --- use them like a piece of outboard processing!

Certainly look at doing some tests on your proposed design to see the effects of the transformers before you commit

Or...just go for it! Bearing in mind that your subjective impressions of the sound of passive mixing over mixing in the box using your design may in fact be a result of a 2dB boost at 16kHz or something like, all that due to the effect of the transformers on frequency response. Not to mention their contribution to distortion etc

Rob

[chris p]

Things are moving along on this project, but there's still one question I have, if anyone can assist.

What should happen if nothing is plugged in to one or more of the summing inputs? If I do nothing, then an empty socket will just be an open circuit, but there seems to be some confusion on the Prodigy site as to whether the input + should be shorted to the input - in such cases (or is it that input + and input - BOTH be shorted to ground?) to ensure a consistent db drop.

For those following at home, my final design allows inputs to be plugged either into a socket that goes through the 1:1 trannies before the summing resistor or a normal socket that just connects to the summing resistor, so I can play with the effect that the transistors give (if any). The metering can be switched between the sum output or the external inputs, and I'm strongly considering a pair of 1:4 trannies (rather than an active stage) on the outputs to restore the 12db loss. I'll post it once I've got a solution for the empty socket issue.

So far as build is concerned, the power supply and meters side of it is about 80% done, the transformer part about 60% and the mixbus side about 20%. I'll just keep plugging away, and post some pics once its more complete.

[rob]

i think you've answered your own question

if you leave inputs open circuit, then there is less attenuation applied to the other inputs. If you short the inputs when nothing is plugged in then you get close to maintaining a consistant attenuation.

Rob

[chris p]

Thanks Rob.

What I've done is get some of those DPDT switching sockets that switch when a plug is inserted.



So if there is anything plugged into the transformer input, that's what goes to the mix buses. If nothing is plugged in to the transformer input, the signal comes from the normal input. If something is plugged into the normal input, that's what goes to the mix busses, otherwise the + and - are simply shorted together (bearing in mind there is still a 5k1 resistor between each of them and the mix bus. Fun Fun Fun.

[rob]

neato

R

[chris p]

Here's about where I'm at -



The VU part is now complete: play the famous 1kHz -14dbFS tones and you get

STEREO (Note L and R cancel out in the L-R meter)



LEFT, and RIGHT (Yes, the L+R and L-R meters are SUPPOSED to read about -6db: think about it)





I then ran a commercial track through to watch the levels - amazing how LITTLE side (L-R) there was on it! Might as well have played the thing in mono.

[astrovic]

Looking good Chris!

The -6 question is a curly one for me - don't tell me, I'll get it!

And thanks for following up on this - I'm still in the throes of my planning, and your build has been extremely helpful and informative.

Cheers!

[astrovic]

Ok, I think I got (I'm here on my own, aren't I?)

a drop of -6dB means the voltage has halved (yep, I've read Rob's article in AT). If Ohm's law has taught me anything (hmmm...), at a constant current the resistance must have doubled - which it has, because whilst we only have the left (or right) channel of audio, we have the 8k2 resistors from both left and right presenting a load on the voltage travelling to the meter.

How'd I go, teach?

PS This is great stuff. After spending most of the day churning through a public company's constitution, it's nice to use my brain for good instead of evil. Thanks again.

[chris p]

Hey Astro

Well done. You're pretty much there, but I'd express it a little differently. It has 2 aspects, acoustic and electrical.

The electrical one is the one you've identified: the L+R and L-R meters are separated by mixing resistors from the inputs (see Rob's post above), and so the db loss is 20*log(1/N) where N=2 in this case, which works out to -6db. The actual value of the resistors doesn't affect this db loss (but it does affect impedance) - FWIW I'm using 5k1s.

So as a simple matter of electronics, a 0db signal at either the L or R input (and the other with no signal) will show -6db on the two summed meters. The question is whether this is right and meaningful, or whether we need some gain stage after summing to correct the drop.

As it turns out, the -6db drop (or half the power) is exactly what we want for this sort of metering if we want the L+R meter to show 0db when there is a 0db signal at BOTH the left and right inputs - which probably makes the most sense in terms of representing the balance of the mix. I understand that the BBC standard is to use L+R-6 for its famous VI metering for this reason - the meters show what the brain is hearing. So its a nice happy coincidence that the most useful mixing design is actually the easiest to build - how often does that happen in DIY audio?

I can't let this go without a big nod and THANK YOU to Rob for his VU buffer design and Joe Malone for implementing it. There's a JLM VU Buffer behind each of those meters, the only addition being the 100uf capacitor across the meter terminals to more closely reflect the correct VU ballistics. Thanks guys.

Oh yeah, and that front panel switch (4PDT) selects between the output from the mix busses or an external signal for metering. The rear panel has outs from the mix bus, an external L and R input and external L and R thru. This means I can put the mix through some outboard, bring it back into the meter box and then straight back out to my AD converter, monitoring either the dry mix and the processed mix with a flick of the switch.

[astrovic]

Cheers Chris - that was a very polite way of saying "wrong, but good try"!!

Glad to hear this project is working nicely for you.

Chris

[Martin]

just stumbled across this thread, about 30% of the technospeak is lost on me but its now my mission for the day to understand some more of it...

having a 4 VU meter external box with one dead meter is giving me ideas about ripping the guts out and following a similar route

great work! very informative

[chris p]

Project almost complete!

After fiddling around with some designs, I decided on an active virtual earth mixer with LOTS of iron - I call this my heavy metal mixer, for good reason. It takes four stereo stems into a LR output, with the mixing stage being a virtual earth design based around the JLM Hybrid opamp. A BIG thank you to Joe Malone for his help in the mixer board design - I had a lot of unnecessary components in my design which he pruned to keep the whole thing simple.

It doesn't explode when switched on and passes a 1KHz test tone at unity gain on each channel. I'll now fine tune the calibration and run some mixing texts - I'll post some audio if anyone's interested.

[astrovic]

I'll post some audio if anyone's interested.

Very interested, Chris - that is awesome. Well done!

[mylesgm]

Hey thats great! I've a large collection of 600:600 tx's and this looks like a cool project indeed. How did you go about making the opamp pcb?
Myles

[chris p]

How did you go about making the opamp pcb?

Once the design was settled I made up a one channel prototype using a breadboard to make sure it worked.

I then used Osmond Quartz PCB design software to turn the schematic into a layout. I use the iron-on approach (laser print the design in reverse onto dedicated paper, or even glossy magazine paper - I prefer the former) to get the design onto a standard Jaycar PCB board and etched it (again, Jaycar for the chemicals). Drill the holes, stuff the board and spray with sealant (yes, Jaycar).

This was my first "ground plane" board (where you leave most of the copper intact and just etch channels either side of "wires", rather than etch everything apart from the traces). It worked reasonably well but I did have to use a touch up etch-resistor pen (guess where from) in a couple of places before etching to get the design just the way I wanted.

I've really enjoyed this degree of "hands on" in this project, rather than just stuffing a 3rd party board, but like everything it takes time to learn it.

Astro, I'm getting the samples together. Basically 4 files for ITB mix to disk, the ITB mix out and back through the converter and then class A and class A/B mixes. Big question is whether I label the files upfront or get you all to try and guess .....

[chris p]

OK folks. Play time. Software is latest Logic Pro, converter is Apogee Ensemble. The four stereo tracks were each set to peak at around -10dbFS, rather than as an actual attempt at a mix. The song is (intentionally) dreadful - I put it together quickly more to get something across most of the audio spectrum rather than to sound good.

Can you link these files to the following methods?

A: ITB mix to disk
B: ITB out to converter and straight back in to disk
C: External mix of the 4 stereo channels, class A
D: External mix of the 4 stereo channels, class AB

Bonus points if you can identify the Wiggles tunes that underlies the chords.

http://www.legalfinesse.com.au/forum/1.aif
http://www.legalfinesse.com.au/forum/2.aif
http://www.legalfinesse.com.au/forum/3.aif
http://www.legalfinesse.com.au/forum/4.aif

I will advise as soon as someone gets it right, or failing that in about a week.

[chris p]

No players? OK

1 is class A
2 is ITB bounce
3 is class AB
4 is ITB via converter

[mylesgm]

so a year on, are you using the summing amp? finding the class A/B switch useful?

Thinking about building myself one and want some real use feedback...

M

[chris p]

Myles

It does get use, but perhaps not as much as it deserves. I'm in a rebuilding phase this year, so haven;t done much recording or mixing. So far as the A/AB switch, it makes a subtle difference (or perhaps I just think it does) - but I'm generally happy with the switch in either position and given my time again I'd probably just go for the faster class AB.

One thing not often talked about in the ITB OTB debate is cables. Do not overlook the potential for noise and signal loss caused by cabling, and be prepared to spend $$$$ to get it right. The perceived benefits of OTB over ITB mixing can easily be lost by one dodgy cable or connector in the entire rats nest you need to implement it.