Noise and signal leakage in a Speck M72 console

[yosh]

Hi, I'm trying to figure out how best to wire the inputs to my Speck M72 console. The inputs are unbalanced. I have attached a diagram of how I have it wired now. The ground from the input cable is floating on the console side with pins + and - of the external balanced signal going to + and ground of the console channel card.  This scenario is low noise, but unfortunately has signal leakage. (even if I have fader down, or mute engaged, I can still faintly hear the signal). With loud signals like electric guitars or drums this is loud enough to be a problem.

The other way I've tried wiring it is to have pin 3 shorted to ground on the console input jack. This gets rid of signal leakage...yay! But unfortunately now I have a noisier, humming channel. It's not super loud, but if I have all the channels wired this way, it adds up and becomes way too loud when I have a whole mix going.

Any advice for me in how to wire things for minimal noise AND minimal signal leakage? I feel like I'm missing something obvious...

 

[ruffrecords]

Hi, I'm trying to figure out how best to wire the inputs to my Speck M72 console. The inputs are unbalanced. I have attached a diagram of how I have it wired now. The ground from the input cable is floating on the console side with pins + and - of the external balanced signal going to + and ground of the console channel card.  This scenario is low noise, but unfortunately has signal leakage. (even if I have fader down, or mute engaged, I can still faintly hear the signal). With loud signals like electric guitars or drums this is loud enough to be a problem.

I just downloaded the manual from Speck and according to the schematics the mic inputs are balanced and the line inputs are unbalanced - so, first question,  are you using the line inputs or the mic inputs?

Cheers

Ian

 

[yosh]

Hi Ian, thanks for your response. The Speck M72 is a line input console, the mic pres are optional and I just have unbalanced line inputs that I'm using. I found a post by John Klett on the R/E/P forum about this issue http://repforums.prosoundweb.com/index.php?topic=6754.0:

...Gear with bipolar +/- 15 volt power supplies and active balanced outputs can develop nearly of +28dBu differentially into no load.  That is 6dB more than an unbalanced output on the same supply voltages.  There are really two output amplifiers driving a single output, each driving a signal pin on an XLR or TRS jack, with one amplifier being 180 degrees phase reversed from the other to proved a complimentary or "differential" output.  The peak to peak figure is going to be double because while the "pin2" amplifier output is sitting at +14.5Vdc, the other "pin3" amplifier will be sitting at -14.5Vdc... so that is a differential "positive" peak voltage of 29 volts and, in the next half-cycle, the polarity has reverse so the differential "negative" peak voltage is -29 volts...  the differential peak to peak is now double - skip the math - it's 6dB more...  UNLESS you are trying to drive INTO and unbalanced input.

When you drive a differential balanced output into an unbalanced input the driving amplifier is seeing one of it's amplifiers shorting to ground.  These days the output circuits are generally designed to sense this and correct the level at the surviving un-shorted output so there is no 6dB loss in level (because half the output is shorted)...  but we are back to one amplifier driving off a set of 15 volt (or whatever) supplies and the max level is dropped back down to what it would have been with an unbalanced output...  but, in addition, it's not as though that low side amplifier shorting to ground has actually shut itself off.  It may be current-limited but in order for a differential balanced output to "sense" that one half has been shorted to ground that amplifier has to keep trying to output signal to GROUND...  and this can cause problems with crosstalk.  Crosstalk can make itself apparent both as signal bleed, leakage and an increase in overall distortion...  because the amplifier driving ground is often distorting pretty badly so the leakage signal is distorted and mixing back in with the rest of the audio you are trying to keep clean.  There are some differential outputs that are better than others in this regard (dumping audio to ground when unbalanced) but by and large those circuits cost a few cents more to produce so many manufacturers opt for saving the money.

This is why - in a topic running in this forum "Anyone Else Constructing Their Own Console?" - a solution to a crosstalk problem was to drop the low side pin of the differential balanced output that was driving in to an unbalanced line input - drop the wire from the low side pin and tie it to ground.  If you can deal with loosing 6dB in maximum level from the source then this can work.  It's not the very best way to deal with it but it can solve the problem.

Other fixes would include

-> as above (drop the wire from the low side pin and tie it to source ground) but jumper the low side pin to ground at the source device and hope that the audio current driving into ground there takes does not find it desirable to go to ground through the low side wire to your console - a good chassis earth for the source equipment can help.  The amplifier will sense the unbalance and maintain level - not dropping by 6dB... and sometimes a 10 ohm decoupling resistor inserted in series with the low side of the common wire between source and destination can help by making the path to the console high impedance than going to ground another way...  but this can be problematic for other reasons - 10 to 100 ohm earth decoupling resistor can work great in closed designs like inside consoles and such but in the less controlled and chaotic environment outside the the gear where connection and ground paths change...  it can be problematic.  I am already out on a big tangent so I won't explain to much more about that here except to say that this is one reason among many that I design and view studios as complete systems - like one big piece of equipment - instead of just a bunch of disparate gear piled in a room and wired together.

more fixes, continued

-> balancing every input to every unbalanced input device that could be driven from a differential output - you can do this with transformers or by adding little circuit boards with differential input chips on them.

-> transformer isolating all differential output devices so they drive a transformer primary differentially and stay happy while the transformer secondary can be unbalanced and not care - oh yes - transformered outputs really don't care if they are driving a balanced load or an unbalanced load as long as both ends of the transformer winding are connected to the destination (one end to ground and the other to hot in the case of an unbalanced input).  Transformers don't drive audio current to ground or loose headroom when connected to an unbalanced load.

-> you can wire some 1:1 transformers into a balanced patchbay and patch from differential balanced outputs through the transformer to unbalanced inputs on a case by case basis...

This seems to make sense to me. Though it also makes it sound like I have fairly limited options for solving both my problems. I guess I need to balance the inputs to the Speck using Burr Brown INA134 or the like. Does this sound right? I'd love to find a solution that does not require balancing all 72 inputs, but maybe i just need to...

 

[ruffrecords]

OK, thanks for clarifying that. Nest question is what are you driving the inputs with?

Cheers

Ian

 

[yosh]

Ah yes, sorry, I left out important details. I am driving the console from 2 Lynx Aurora 16 converters.

 

[eemarty]

This thread is old so you have probably found a solution already, but I have some ideas.

I suspect the source of the crosstalk that you were experiencing when you connected the '-' from the aurora to the gnd of your speck input is the current that the '-' output was sinking into the ground  of the speck. Theoretically current flowing into the ground of the board wouldn't produce noise, but since there is some small resistance in the ground wiring, the large currents from loud drums could create small voltages in the ground of the speck which are amplified on the other channels.

I suspect that the hum your were getting when you connected the grounds together and left the '-' floating was a ground loop between the aurora and speck.

I have a few suggestions:

1. Did you try simply not connecting the '-' or ground from the aurora? So only the '+' of the aurora would be connected to the '+' input of the speck. The '-' and gnd could be connected at the aurora side of the cable and floating at the speck side. If this works it would be the simplest solution.

2. If that doesn't work, you could add a resistor in series with the '-' aurora output and the gnd of the speck to reduce the current flowing through that path. This should reduce the crosstalk significantly without connecting the grounds together and adding hum. Somewhere around 1k should work.  Or even better, you could use the same value resistor as the input impedance of the '+' input of the speck so the aurora sees a equal impedance load on the '+' and '-'.

 

[scott2000]

This thread is old so you have probably found a solution already, but I have some ideas.

I suspect the source of the crosstalk that you were experiencing when you connected the '-' from the aurora to the gnd of your speck input is the current that the '-' output was sinking into the ground  of the speck. Theoretically current flowing into the ground of the board wouldn't produce noise, but since there is some small resistance in the ground wiring, the large currents from loud drums could create small voltages in the ground of the speck which are amplified on the other channels.

I suspect that the hum your were getting when you connected the grounds together and left the '-' floating was a ground loop between the aurora and speck.

I have a few suggestions:

1. Did you try simply not connecting the '-' or ground from the aurora? So only the '+' of the aurora would be connected to the '+' input of the speck. The '-' and gnd could be connected at the aurora side of the cable and floating at the speck side. If this works it would be the simplest solution.

2. If that doesn't work, you could add a resistor in series with the '-' aurora output and the gnd of the speck to reduce the current flowing through that path. This should reduce the crosstalk significantly without connecting the grounds together and adding hum. Somewhere around 1k should work.  Or even better, you could use the same value resistor as the input impedance of the '+' input of the speck so the aurora sees a equal impedance load on the '+' and '-'.

Interesting...

Thanks for bringing this up!

 

[abbey road d enfer]

Indeed it really looks like the Aurora's outputs are not really floating and they're injecting current in the system's ground.
This was a common problem, known as the Tascam problem, where the negative leg of the output is an opamp's output carrying a mirror-image of the positive leg. Connecting it to ground injects distorted currents into the sytem's ground and distorts the other outputs. One of the solution is to use just one leg and let the other float, the other, much preferrable IMO is to convert the inputs to balanced types, using a THAT 120x chip.