DIY line level summing

[ChrissA]

Hi there!

I'm an electronics noob and want to verify I have done my research correctly before I destroy something. I need to sum analog line level signals together out of a mixing desk. The output is 50 ohms, max +21dB(U), balanced line level signal. After summing they go into a line level input. What I understood from my research is I have to put a 10k resistor (0.25w ok?) in series with both the + and - of both signals before joining the + of both and the - of both and the ground without anything in between. The loss should be -9.5dB, but if they are identical that should add 6dB so the total loss should in theory be between -9.5dB and -3.5dB.

Are these assertions correct?

** EDIT ** Also quite important, am I also correct if I say there is no (or barely any) loss of audio quality?

Thanks,
Chris

 

[ruffrecords]

ut
Are these assertions correct?

More or less. The exact loss depends also on the input impedance of the line input. If the two outputs are uncorrelated the level will increase by 3dB. If they are 100% correlated (i.e. identical) the level will increase by 6dB

** EDIT ** Also quite important, am I also correct if I say there is no (or barely any) loss of audio quality?

You are correct.

Cheers

Ian

 

[ChrissA]

More or less. The exact loss depends also on the input impedance of the line input. If the two outputs are uncorrelated the level will increase by 3dB. If they are 100% correlated (i.e. identical) the level will increase by 6dB

You are correct.

Cheers

Ian

Okay so after testing there seems to be a lowpassfilter that has been introduced. It's a lowpass with the -3db point at 12khz, but somehow -4.3db at 20kh (even though the capacitance theoretically would be like 1300pf for a -3db point at 12khz while in reality it should not surpass 30pf since it's 55pf/m and it's less than half a meter of xlr wire). Can I safely lower the resistor to 1khz, or even 100 ohm? The output is 50 ohms, the input 10kohm. This should increase the filter beyond the audible range is my thinking

 

[ruffrecords]

How low you can reduce the value of the resistors depends on the drive capability of the outputs, not on their output impedance. If the spec for the mixer outputs says they can drive a load less than 2K then yes you can reduce the resistors to 1K each.

I am surprised you have a low pass filter effect at such a low frequency. The length of cable you are using in insufficient to cause this. However, it may be that the line input you are feeding the mixed signals to has some internal capacitance across its inputs (for RFI purposes). Although it has a 10K input impedance, it probably assumes the source driving it will be 600 ohms or less. This would mean a higher source impedance could suffer HF loss.

Cheers

Ian

 

[ChrissA]

How low you can reduce the value of the resistors depends on the drive capability of the outputs, not on their output impedance. If the spec for the mixer outputs says they can drive a load less than 2K then yes you can reduce the resistors to 1K each.

I am surprised you have a low pass filter effect at such a low frequency. The length of cable you are using in insufficient to cause this. However, it may be that the line input you are feeding the mixed signals to has some internal capacitance across its inputs (for RFI purposes). Although it has a 10K input impedance, it probably assumes the source driving it will be 600 ohms or less. This would mean a higher source impedance could suffer HF loss.

Cheers

Ian

I am as surprised as you are. The source driving it is the mixer right? Because the outputs are 50 ohms according to the spec sheet. I have no idea why I am missing so much HF. I'll see if I can find what it can drive according to the spec sheet and check back in. Thanks!

 

[Newmarket]

I am as surprised as you are. The source driving it is the mixer right? Because the outputs are 50 ohms according to the spec sheet. I have no idea why I am missing so much HF. I'll see if I can find what it can drive according to the spec sheet and check back in. Thanks!

Without driving into the maths atm (too much like the day job ) something doesn't seem right. Can you measure the actual capacitance ?

 

[ChrissA]

How low you can reduce the value of the resistors depends on the drive capability of the outputs, not on their output impedance. If the spec for the mixer outputs says they can drive a load less than 2K then yes you can reduce the resistors to 1K each.

I am surprised you have a low pass filter effect at such a low frequency. The length of cable you are using in insufficient to cause this. However, it may be that the line input you are feeding the mixed signals to has some internal capacitance across its inputs (for RFI purposes). Although it has a 10K input impedance, it probably assumes the source driving it will be 600 ohms or less. This would mean a higher source impedance could suffer HF loss.

Cheers

Ian

I can't find what is says what load it can drive, it's an M32

 

[ChrissA]

Without driving into the maths atm (too much like the day job ) something doesn't seem right. Can you measure the actual capacitance ?

Good question! I have done the maths, the capacitance according to the math should be like 1300 pf and some change, which I am unable to explain. The cable should be 55pf a meter and there is no more than a meter of cable in the entire thing. I will measure this saturday. Just to be sure, I measure this by probing the + and - on the summed output with the connectors plugged into the mixer, right?

 

[Newmarket]

Good question! I have done the maths, the capacitance according to the math should be like 1300 pf and some change, which I am unable to explain. The cable should be 55pf a meter and there is no more than a meter of cable in the entire thing. I will measure this saturday. Just to be sure, I measure this by probing the + and - on the summed output with the connectors plugged into the mixer, right?

Measure + conductor to - conductor and also each conductor to screen/shield/ 0V.

 

[ChrissA]

Okay so I've measured. When the cable is NOT plugged into the mixer the measurements reflect what I'd expect; between + en - around 65pF en between both + and ground and - and ground a bit higher, around 95pF. But when it's plugged in and the mixer is turned on the jumps an incredible amount. Between + and - to around 36nF and between both + and ground and - and ground 70nF. With that sort of capacitance it doesn't matter what kind of resistor I put in instead of the 10k, I'll make a low pass filter. Any ideas on how this is possible?

*EDIT* It wasn't uF it was nF. Still just under a 1000x increase. Also measured a line out without the summing cable with 10k resistors in place, that was in the multiple uF range

 

[Newmarket]

Make measurements with the mixer unpowered.

 

[ruffrecords]

I am as surprised as you are. The source driving it is the mixer right? Because the outputs are 50 ohms according to the spec sheet. I have no idea why I am missing so much HF. I'll see if I can find what it can drive according to the spec sheet and check back in. Thanks!

No, as far as the line input is concerned the driving source is the 10K resistors. Even so, the turnover frequency of 56pF and 10K ohms is over 280KHz so there has the be some extra capacitance on the line input circuit itself. So try 2K instead of 10K and see if the HF turnover point increases.

Which mixing console are we talking about?

Cheers

Ian

 

[ChrissA]

No, as far as the line input is concerned the driving source is the 10K resistors. Even so, the turnover frequency of 56pF and 10K ohms is over 280KHz so there has the be some extra capacitance on the line input circuit itself. So try 2K instead of 10K and see if the HF turnover point increases.

Which mixing console are we talking about?

Cheers

Ian

It's an M32r. Have not measured it with the desk turned off and left already, will do that tomorrow. What is odd to me is that with the 36nF capacitance and the 10k resistor it should be a 440hz lowpass, it's a mystery to me why it's at 12kHz. I'll measure the capacitance with the desk turned off tomorrow and I'll also do a test run with 2k instead of 10k resistors and see what happens. Very curious what is going on

 

[ruffrecords]

It's an M32r. Have not measured it with the desk turned off and left already, will do that tomorrow. What is odd to me is that with the 36nF capacitance and the 10k resistor it should be a 440hz lowpass, it's a mystery to me why it's at 12kHz. I'll measure the capacitance with the desk turned off tomorrow and I'll also do a test run with 2k instead of 10k resistors and see what happens. Very curious what is going on

Where did the 36nF come from. I thought we were talking about 56pF of cable capacitance.

Cheers

Ian

 

[ChrissA]

Where did the 36nF come from. I thought we were talking about 56pF of cable capacitance.

Cheers

Ian

56pF is what it should be, but I have measured what it actually was. When the cable is NOT plugged into the mixer the measurements reflect what I'd expect; between + en - around 65pF en between both + and ground and - and ground a bit higher, around 95pF. But when it's plugged in and the mixer is turned on the jumps an incredible amount. Between + and - to around 36nF and between both + and ground and - and ground 70nF.

 

[ruffrecords]

56pF is what it should be, but I have measured what it actually was. When the cable is NOT plugged into the mixer the measurements reflect what I'd expect; between + en - around 65pF en between both + and ground and - and ground a bit higher, around 95pF. But when it's plugged in and the mixer is turned on the jumps an incredible amount. Between + and - to around 36nF and between both + and ground and - and ground 70nF.

That is definitely very odd. I downloaded the Midis M32R schematics and there is nothing like that value connected across the input. Two or three nF at the most.

What was the result of the experiment with 2K resistors?

Cheers

Ian

 

[ChrissA]

That is definitely very odd. I downloaded the Midis M32R schematics and there is nothing like that value connected across the input. Two or three nF at the most.

What was the result of the experiment with 2K resistors?

Cheers

Ian

I was hoping you could explain it. It gets weirder (at least I think it's weirder). The output straight without the 10k summing resistors (so just a cable plugged into the mixer) shows a 7 uF (not pF, not even nF, uF) capacitance. With the new setup and 2k resistors in place the capacitance still gives back the same measurements. With 2k resistors in place the low pass filter increase but it still is in the audible range, -1dB at 17kHz. I'm going to check with 1k resistors now.

 

[ChrissA]

With 1k resistors it's 0.4dB at 20k. I'm just going to solder 1k resistors into the plugs I made when I expected 10k to work instantly and call it a day. I still am extremely curious what happened here, because nothing adds up in my head. Where the 36nF comes from in the summing cable, where the 7uF in the output from the console comes from and why it's so much less with the summing cable, why with 36nF of capacitance and 10k resistors the -3db point is at 12kHz (while the theory says it should be 440hz), why I have to drop all the way down to a 1kOhm resistor (while the theory says a 2k should have been plenty, with -3dB at 12kHz with a 10kOhm resistor it should have theoretically be a 1326pF capacitance (which I didn't and can't measure) and with a 2k resistor the -3dB point should have been at 60kHz so with a 1kOhm at 120kHz and there should not have been a dropoff in the audible spectrum), and much more curious things.

I am still eager to find these things out but also want to finish this project up and it's usable now. Unless someone can tell me it's a bad idea to leave it at this of course.

 

[ChrissA]

Someone told me today I might want to try a shunt (is that how you write it?) resistor of 50ohm between the + and - of the summed output. The level drops significantly, but even with 20kOhm summing resistors it's flat as it can be. I am going to hold off on redoing the plugs I made and wait for someone to tell me if this is a serieus option (and I am not going to damage things) because if it is I might just do that. The level drop is annoying but I can manage that, I have to put the input where the sum will go into at -36 anyways, so if I'll put that at 0 we should be good.

 

[ruffrecords]

Sounds to me like there is something wrong with your capacitance measurements. First the values are far larger that you would ever expect to find across a console input and second, the turnover frequencies bear no resemblance to the calculated ones.

Cheers

Ian