2:1 Passive Bridging Summer

[JShaffer]

Good day folks! First post so please take it easy on me ;-)

I'm trying to implement a technique I learned from Allen Sides where you split a signal, say a snare drum, to two busses and print two identical signals to two channels of tape in order to gain 3dB of SN.

On the way out of the tape machine, I would love to then be able to transparently recombine those signals before hitting the line input of one channel on the board since the point is a gain in headroom and noise, not any kind of parallel processing, etc. Would any of you be willing any able to point me in the right direction? Am I missing something?

The line inputs on the Helios are unbalanced but I have 10k:10k transformers to work with

 

[JohnRoberts]

It could be as easy as a passive resistive sum ( say 2k ohm resistors, one in series with each output multed together).

Companding tape NR (like dbx) could provide far more headroom and S/N benefit, but there is no free lunch (dynamic tracking in compandors is not perfect).

Summing adjacent tracks from a multitrack tape machine "should" be reasonably coherent.

JR

PS: I would lower my expectations for how much transparency 3dB delivers. You didn't share what kind of tape machine you are using.

 

[Bo Deadly]

This could be done many ways.

The simplest would be to use balanced cable with a Y at the tape end and then implant two 1K resistors on the console end to make an unbalanced summed signal. Try to match the resistors carefully.

A fancier method would be to record one of the otherwise identical signals inverted or invert it on the way out (if the out is balanced just swap pins) of the tape machine and then again make a Y cable for the tape end but make the console end a standard balanced connection into your transformer so that it automatically debalances and mixes the two signals. This method might have superior performance because you will be leveraging the debal / mix capability of the transformer and, if you record inverted, the debal / mix step could cancel non-linearity of the recorder and actually reduce THD.

 

[JShaffer]

Thank you very much gentlemen!

Sorry I could have given more specifics. It's a 3M M56 16 track from Wally Heiders at 15ips running SM900. I'm converting it to IEC from NAB with Ian's help.

John, I'd like to try to avoid NR. Trying to keep a pretty simple setup. I have exactly 13 Helios channels, so it's kind of a win-win right now for my very simple setup. Doing recordings that could probably be best described as somewhat in-between a natural approach and a circa 1970 rock approach with my own ideas sprinkled in. The idea is to implement this strategy on 2-3 of the loudest yet most dynamic and sparse sources such as kick, snare, and lead vox.

Bo, I really like your "fancy" method. Same thing... just swap pins if I'm inverting on the way in?

Thanks again

 

[Bo Deadly]

Bo, I really like your "fancy" method. Same thing... just swap pins if I'm inverting on the way in?

Actually no because the two signals will just cancel each other when mixed. And the transformer would have to have a split winding which is kind of an important detail I failed to mention. Do your transforms happen to have a split winding? If not, there's actually no "fancy" option.

Also, are the M56 outputs balanced? Electrically or transformer?

If you just have unbalanced to unbalanced, then your best option at the moment would be to just make a Y cable from some balanced cable with each signal wire on a + output and the - outputs connected together to the shield. Than at the other end use 2K mix resistors to your one + input and shield to - of input. At least that would be the first thing to try.

You could insert your transformer between said mix resistors and input of the console but you should test to see how well that works because in practice transformers sometimes need to be driven from a low impedance.

If your transformer has a split winding then you could put each on the two tape outputs and use it to mix the signals together and get balanced isolation. Then if you phase invert one of your tracks you could phase invert that one winding which might improve THD a little.

But this is all just thinking-out-loud kind of stuff. It's the sort of thing I would fiddle for days with testing and taking spectrums and being OCD over before deciding on an actual concrete implementation.

And in the end you probably wouldn't really notice much of a difference. The 3dB is only if the signals are perfectly correlated as JR already mentioned.

You might be better off using a simple passive crossover to split the signal into LF and HF and then record those separately. That would minimize intermodulation distortion (which is huge on a tape machine) and improve SNR because the bandwidth of each channel is much less. But again, I literally just thought this up typing so ...

 

[JShaffer]

Yeah that makes sense.

The M56 is transformer balanced out at 600 ohms. I do now know if the transformers are 1:1 or other.

The 10k:10k (Gardners MU7530) has a center tap BUT, something else just dawned on me. I also have a pair of Gardners MU7525. These are actually 600:600:600 transformers. If the transformers on the tape machine are indeed 600:600 (like Helios, Trident A Range, etc bus outputs were) then I could conceivably bypass them and replace the pair with a single Gardners MU7525 (the 600:600:600 transformer) and mix the two channels back together that way right at the tape machine.

Does that make sense?

 

[Bo Deadly]

Ehhh, I wouldn't do that because another thing I failed to mention is that you have to be careful about driving outputs against each other. It's probably assuming too much that the two tracks will have signals that do not drive the outputs against together. For example, if one track was not recorded and you didn't realize it one winding would be driven short. An output that is 600 ohms doesn't mean the output impedance is 600 ohms. It means it can drive 600 ohms. The actual output impedance could be much lower. If it's too low and you load it too much and burn out driver transistors. Thinking about it more, even with 10K transformers I'm getting an itchy feeling.

But it doesn't matter because if the outputs of the M56 are transformers, then the Y cable described previously would work really well. Because the outs are transformers, the common (the shield) would be floating and so you get the full benefit of isolation. The grounds of the devices are not connected together and so there would be no ground loop currents. And the 2K resistors isolate the outputs from one another so it's super safe from driving outputs against each other.

However, I might consider lowering the 2K a little. If the signals are perfectly matched, the load on each output would look like 4k+10k I think. I might reduce the 2K add a load resistor. Old 600 ohm outputs like to be loaded. Note sure about the values but the cable would be like this:

 

[JShaffer]

That is very very helpful. I'll try it out and report back! Thanks Bo!

Meanwhile if anyone else has an idea or notices a chink in the armor, feel free to give ya boy a holla