Summing and Splitting, same bat time, same bat channel

Hey there.

I'm thinking about this, and thought I'd collect some wisdom :wink:

If I had twelve channels coming in, and I wanted to passively sum them, but also wanted to split each channel to feed the summing network and another device ( recorder), how would I do this in such a way that would not allow audio from one channel to bleed into another ( on the recorder feeds) :?:

I'm thinking this is a job for a transormer with dual secondary, but am wondering if there is a simpler passive way I'm not thinking about :?:

Regards

ju

ju,

What sort of levels?

Balanced or unbalanced?

If unbalanced and line level, an op amp buffer/driver would allow a split off as a "direct out" and then a send to a passive summing network followed by gain (or virtual earth/feedback summing)

Any use?

Mark

Currently, this is all balanced at +4, but even in the unbalanced world, it still looks like I'd be in need of some active circuitry to accomplish this, no?

Not that I'm opposed to it =), just exploring options

Maybe I should describe what I'm after, as that may help with things.

I use an HD192 for recording. 12 in, 12 out balanced interface.

I've got racks of outboard equipment( all balanced). Mic pre's, compressors, etc.

The box I'm envisioning (and have built, just not completely) serves two purposes.

When tracking, the box will sum all the outboard equipment into a monitoring output, so that I get around the latency problems with monitoring through the computer. This is where those direct outs come into play. the signal has to feed the HD192, and the summing network at the same time.

When mixing, things are bussed to the outputs of the HD192 in stereo pairs, to outboard gear ( compressors, eq, etc), and then to the summing network, the output of the summing network then goes to master(tape, computer, whatever).

What I started with was NYDaves passive summing network, with the LCR switching arrangement he subsequently posted. on the back, I have just thru'd the input, so that I can feed the HD192.

What I get with this arrangment is the summing I so desire, but the directs on the back of course are getting some of the other tracks bleeding in. Not desirable :grin:

So hence the question. In order to feed the HD192 without bleed, what needs to be done. I've got lots o room in the box, can totally redo the electronics if necessary

Here's snap:

http://shinybox.com/diy/summer.jpg
http://shinybox.com/diy/summer2.jpg

regards

ju

[quote author="fum"]When tracking, the box will sum all the outboard equipment into a monitoring output, so that I get around the latency problems with monitoring through the computer. This is where those direct outs come into play. the signal has to feed the HD192, and the summing network at the same time.
[/quote]

maybe you should think of it like this: tape send = protools inputs. so if you mult tape send to the inputs of your summing box via half normalled points on your patchbay [or build a mini patchbay for the box], durring tracking you get 0 latency monitoring of the exact signal path that's being recorded, as opposed to idle outboard gear [if i understood you correctly]. in mix mode, you patch tape return [protools outputs] into your outbaord gear, and the outboard gear outputs get patched into the summing box inputs breaking the normals from tape send.

so for the box you would have 2 rows of patchbay. tape send, and monitor input. patching into mon in breaks half normal.

or maybe this isn't what you are looking for...

do what al says and half normal the bay and leave the box be.

Ive gotta go dig out those bays...

dave

[quote author="al_p"]
maybe you should think of it like this: tape send = protools inputs. so if you mult tape send to the inputs of your summing box via half normalled points on your patchbay [or build a mini patchbay for the box], durring tracking you get 0 latency monitoring of the exact signal path that's being recorded, as opposed to idle outboard gear [if i understood you correctly][/quote]

I guess I should also state that at present, I don't use a console at all. Everything is outboard, and wired into the bays.

At present, tracking a 3 piece rock band, there isn't much idle outboard gear =). What's going into the recorder is the exact signal chain only, for each track.

I don't see how half-normalling is much different then what I have currently ( other than moving the connection point to the bay rather than in the summing network box), and it may eliminate the outputs of the summing network by using the half-normalling of the bay.

What isn't shown is the back side of the summing network, which has 12 TRS outputs ( which go to the patchbay).

Input currently feeds the summing network, and an output which goes to the recorder. Moving it to a patchbay is only moving the point of connection, but doesn't change how the connection flow is functioning. I'll still be getting junk coming back up from the summing network( it's at a low level, but is still there).

It appears to be a side effect of how the passive summing is working.

Or am I totally missing something here?? thanks for the feedback :grin:

ju

You're always going to get some degree of crosstalk in a passive mixer. This becomes less of a problem as the number of channels is increased (and the mixing buss impedance is decreased), although this comes at the cost of greater loss. What resistance values are you using in your mixer? What's the output impedance of the sources connected to the mixer? The greatest isolation is achieved when the source impedance is low compared to the value of the mixing resistors.

Unsure of the output impedence of the HD192, but assumed that it was somewhat low ( can dig to see if I can find more info on this).

edit: This was statement, should have been question :grin:
I used a Z of 10K in building the summing network. So if I lowered the Z to say 5K, I'd see less crosstalk. :?:

ju

No, don't do that; if it changes anything, it'll give you more crosstalk, since you'll be lowering the ratio of mix buss input resistance to source impedance. When considering a passive mix buss from the viewpoint of crosstalk, it's useful to simplify it as two voltage dividers in series. Say you want to figure out how much signal from channel 1 can leak through to channel 2. The first voltage divider is formed by channel 1's buss input resistor and the composite impedance of the mix buss. The second voltage divider is formed by channel 2's buss input resistor "looking into" the source impedance connected to channel 2.

The attenuation of the first voltage divider will be roughly 1/N (where N is the number of channels). In the case of your 16-channel mixer, it's -24dB. Then add to this the attenuation of the second voltage divider, which is completely dependent on how low your source impedance is compared to the mix buss input resistor.

Suppose, for the sake of argument, that you're using 10K buss resistors, and your source equipment is "prosumer" type with a 1K source impedance. That's only -20dB of additional attenuation, for a total of -44dB, which ain't fantastic by modern standards. Reducing your buss input resistors to 5k would, in this case, degrade this even further to -35dB total. On the other hand, if your source is a piece of "pro" gear with a source Z of 100 ohms or so, your total crosstalk (with 10K buss resistors) is -60dB, which is an order of magnitude better. All of this represents "typical" case and will vary somewhat depending on the settings of controls, etc. Nobody said passive mix buss design was easy! As you can see, its performance is fairly critical as regards circuit impedances, which is why it worked well in the days of "constant impedance" controls, but can be somewhat trickier to implement with lower-cost topologies.

If you can tolerate additional loss through the mixing network, one solution would be to shunt additional resistance across the buss to lower its impedance. The lower the impedance of the buss, the lower the crosstalk, but the greater the loss. You'll never eliminate crosstalk entirely. Even active summing has some crosstalk since an amplifier with infinite open-loop gain, and a perfectly conducting ground system, has not been developed yet :wink:

In this case, I can tolerate additional loss in the network, as more often than not, the output of the mix network is taken into makeup gain (either a buss compressor, or a gainstage ala API, NEVE, QE, etc.).

When it's used for monitoring, there is plenty of room for loss as well ( the amps are running about as low as you get them, and are still extremely loud for near field monitoring).

I obviously missed the mark in interpreting what you last said correctly, so when you speak shunting additional resistance across the buss, if I were to ask really nicely ( with sugar on top), could I get a sketch on a napkin to establish the visual of what I'm to be messing with? Or even a concrete example would help.

The fact that even active mixers still have some crosstalk never crossed my mind, I assumed that "perfect" was obtainable :grin:

I'm willing to spend some time experimenting with this to get a balance between the two that will work.

Just to give the details:

I implemented a balanced 12 channel LCR network, with a Z of 10K, so the resistor values I'm using are R/2 = 4.2K, and R/5=1.69K


Regards

ju

maybe i'm off on this, but take a look at this and see if this helps at all. you will need a summing amp for the 2 mix out...i think.

http://kimaguresound.com/FUMbay.jpg

I'm with ya on that part Al. :grin:

But what I followed up with is that even if I do it exactly as you've shown, there will still be crosstalk between the channels going into the HD192, meaning that because track 1 and track 2 are both feeding the summing network, some amount of track 2 will be seen on the track 1, and vice versa, which is what Dave and I are talking about minimizing.

Regards

ju

...right. just looked at that again. i'm with you now. :thumb:

Fum,

Put a 330-ohm resistor across each of the two outputs of your mix network (in other words, between tip and ring on the 1/4" output jacks). This will increase the loss of the buss by about 10dB and will improve crosstalk by an equal amount. As a nice bonus, it will bring the buss output impedance down to a value that will match a typical mic preamp input pretty nicely.

I'll give this a whirl, and report back =)

Thanks Dave :!:

ju

had one more thought on the drive in to my veal fattening pen. Since this is balanced, do I need to do anything with the ring of the output as well?

If not, how does adding attenuation at just the tip keep the signal balanced still? How do it know? :grin:

ju

Re-read what I wrote. I said to put the resistor between tip and ring.

oops, sorry bout that

Dave,

I think that's the ticket. Just tried it out, and the crosstalk is much smaller.

I'll get this in the rack, and do some listening tests now.

Thanks so much for your help :!:

ju