DIY line level summing

[ChrissA]

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

I'm not sure what I could have done wrong. I have to TRS plugs with 10k resistors soldered between the + and - pins and the cable, both go into an XLR plug and both + strands are soldered to pin 2, both - strands to pin 3 and both ground strands to pin 1. The resistance between pin 2 and the tip of either TRS is 10kOhm, and between pin 3 and the ring same thing. I measured it by placing the probes of my multimeter on pin 2 and 3 on the XLR, then on pin 1 and 3 and then pin 1 and 2. To reiterate, these were my findings.

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 the cable not plugged in the results seem normal, it changes when it's plugged into the mixer while it's turned on.

Should I redo measurements and if so what should I change?

 

[ruffrecords]

I think the measurements with the mixer turned on are not valid. You cannot make valid measurements of capacitance with the equipment switched on. At the moment you are taking two XLR outputs from the Midas, combining them and feeding them back into the Midas. Do you have another interface or piece of equipment with a line input? If you do then try plugging the mixed signals in it instead o back to the Midas.

Cheers

Ian

 

[ChrissA]

I think the measurements with the mixer turned on are not valid. You cannot make valid measurements of capacitance with the equipment switched on. At the moment you are taking two XLR outputs from the Midas, combining them and feeding them back into the Midas. Do you have another interface or piece of equipment with a line input? If you do then try plugging the mixed signals in it instead o back to the Midas.

Cheers

Ian

I am not explaining myself correctly, my apologies. I have plugged the summing cable into the outputs but the xlr with the sum of the two outputs is not plugged into anything and I was measuring on that XLR plug. I will measure again tomorrow with the XLR plugged into the input it'll go into (a wireless transmitter) and let you know what the results are.

 

[ChrissA]

I've redone all measurements. In the meanwhile I have replaced the 10k resistors with 1k resistors, so the cable now are 2 TRS jacks that plug into the output of the M32 with 1kOhm resistors soldered to the + and - of the jack. The cables meet inside an XLR male connector that plugs into the input of a wireless transmitter, so it's essentially a Y cable from 2x TRS jack to 1x XLR male with 1kOhm resistors soldered into the TRS housing. These are all the tests with measurements:

The cable disconnected from any device, measured at the XLR side
74pF between + and -
100-115pF between + and gnd, and - and gnd (varies a bit between the two but close enough I suppose)

TRS connected to M32, XLR disconnected, M32 turned off and measured at XLR
136nF between + and -
257nF between gnd and -
267 between gnd and +

TRS connected to M32, XLR disconnected, M32 turned on and measured at XLR
363nF between all pins, + en -, gnd and + and gnd and -

XLR connected to wireless transmitter, TRS disconnected, transmitter turned both on and off (results were the same) and measured at TRS (both, results were the same)
1.93nF between + en -
3.94nF between gnd and +, and gnd and -

TRS connected to M32, XLR connected to transmitter, both turned off measured in the middle of the cable (cable cut open and copper exposed for measurement)
150nF between + and -
274 between gnd and -, and gnd and +

Same setup as before but with the transmitter turned on yielded the same results.

Same setup as before but with the transmitter turned off and the M32 turned on
365nF between + and -
377nF between gnd and +, and gnd and -

Same setup as before but with both the transmitter and M32 turned on yielded the same results as only the M32 turned on.

 

[rh001]

Hi!

You can't really sum into a medium, bridging or high impedance input circuit. You have to sum into a low impedance circuit like a mic preamp. 10kohm resistors should work. 10kohm summing resistors feeding a 150ohm micpre will drop the level so much that there will be only the tiniest level at the summing point and so it will not back up into the sending circuits and that's a good thing. think of it as drain not clogged. The high gain of the mic pre will make up for signal level loss.

What you're attemnpting is sort of cheating and trying to get away with s. All this talk of capacitance....way off base. The console outputs should feed a balanced input summing mixer. The mixer can have active differential balanced inputs, the output of which is unbalanced and feeds via summing resistors (5kohm to 10 kohm) into summing node, and from there to a summing amp, an inverting summing amp whose feedback resistor value is equal to the summing resistor value.

Instead of active differential inputs line input transformers can be used, secondaries of those transformers wired singled ended (one lead at gnd, the other will have signal). Transformers should have secondaries terminated per their datasheet. Summing resistors will probably be 5k to 10k ohm....try different values to find what sounds best. From summing node feed into an active summing amp as in preceedig paragraph.

BTW you will probably fail if you try putting summing resistors in the connector at the console end even if you then feed into a mic pre only 3 feet of cable later. For one thing the capacitance post the 10k resistor will be the capacitace of all the 3 foot cables....but I'm not going to say more cuz you just can't get away with s.

If I had time to screw around I might try using a smallish metal chassis box just big enough for all the needed female xlrs or 1/4" jacks, then add the summing ressitors to + and - of those connegtors, and make the output male xlr be the + and - summing nodes and feed from that male xlr via 1 or 2 foot cable into a mic pre.....might work.

Doing stuff the right way is costly and painful but it pays off...here is an essay in that:

 

[gyraf]

I'll raise you with the mix amplifier from the G16 - simple, tube-based zero-field / virtual-ground amplifier, giving good performance up to quite a few channels



Switches are for stereo (even numbers to L, uneven to R bus) or mono (to both busses). Input transformer surprisingly non-critical because of no voltage across it...

/Jakob E.

 

[ChrissA]

Hi!

You can't really sum into a medium, bridging or high impedance input circuit. You have to sum into a low impedance circuit like a mic preamp. 10kohm resistors should work. 10kohm summing resistors feeding a 150ohm micpre will drop the level so much that there will be only the tiniest level at the summing point and so it will not back up into the sending circuits and that's a good thing. think of it as drain not clogged. The high gain of the mic pre will make up for signal level loss.

What you're attemnpting is sort of cheating and trying to get away with s. All this talk of capacitance....way off base. The console outputs should feed a balanced input summing mixer. The mixer can have active differential balanced inputs, the output of which is unbalanced and feeds via summing resistors (5kohm to 10 kohm) into summing node, and from there to a summing amp, an inverting summing amp whose feedback resistor value is equal to the summing resistor value.

Instead of active differential inputs line input transformers can be used, secondaries of those transformers wired singled ended (one lead at gnd, the other will have signal). Transformers should have secondaries terminated per their datasheet. Summing resistors will probably be 5k to 10k ohm....try different values to find what sounds best. From summing node feed into an active summing amp as in preceedig paragraph.

BTW you will probably fail if you try putting summing resistors in the connector at the console end even if you then feed into a mic pre only 3 feet of cable later. For one thing the capacitance post the 10k resistor will be the capacitace of all the 3 foot cables....but I'm not going to say more cuz you just can't get away with s.

If I had time to screw around I might try using a smallish metal chassis box just big enough for all the needed female xlrs or 1/4" jacks, then add the summing ressitors to + and - of those connegtors, and make the output male xlr be the + and - summing nodes and feed from that male xlr via 1 or 2 foot cable into a mic pre.....might work.

Doing stuff the right way is costly and painful but it pays off...here is an essay in that:

Okay so I'm not following. What I'm doing now is summing line level into a 10kOhm line level input. The level drop is super managable, it's less than 12dB. It should be possible to do passive summing without active summing amps, there are commercial products being sold that do exactly that. Also if you would have read the past conversations you'd know that I have put summing resistors in the connector successfully. I did get away with it, we were looking into capacitance because it seemed like I had accidentally made a lowpassfilter (which is negligible with the 1k summing resistor, 0.5dB loss at 20kHz). It works now, just curious why with 10k resistors I had a -3db point at 12khz.

 

[ruffrecords]

I've redone all measurements. In the meanwhile I have replaced the 10k resistors with 1k resistors, so the cable now are 2 TRS jacks that plug into the output of the M32 with 1kOhm resistors soldered to the + and - of the jack. The cables meet inside an XLR male connector that plugs into the input of a wireless transmitter, so it's essentially a Y cable from 2x TRS jack to 1x XLR male with 1kOhm resistors soldered into the TRS housing. These are all the tests with measurements:

The cable disconnected from any device, measured at the XLR side
74pF between + and -
100-115pF between + and gnd, and - and gnd (varies a bit between the two but close enough I suppose)

OK, that makes sense.

TRS connected to M32, XLR disconnected, M32 turned off and measured at XLR
136nF between + and -
257nF between gnd and -
267 between gnd and +

TRS connected to M32, XLR disconnected, M32 turned on and measured at XLR
363nF between all pins, + en -, gnd and + and gnd and -

So, just to be absolutely clear, before we go any further, this measurement was made with the pair of TRS just plugged into the output o f the M32.

If that is the case, what you are measuring is the output capacitance of the M32 which is not really relevant to the discussion

And again, before we go any further can you make up a lead with a TRS at one end and an XLR at the other and use it to connect the M32 output to a line input and measure the frequency response.

And then can you modify this lead to have 1K resistors in the hot and cold legs of the TRS and repeat the experiment.

We need to get some firm stakes in the ground.

Cheers

Ian

 

[ruffrecords]

Hi!

You can't really sum into a medium, bridging or high impedance input circuit.

Actually you can. People do all the time in monitor sections to create mono signals from stereo ones

You have to sum into a low impedance circuit like a mic preamp.

You can, but you don't have to. If you sum into a low impedance the advantage is minimising interaction between simple passive level controls connected to the bus feed resistors. If you are not bothered by this interaction or you don't have pots (as in the OPs case) you do not need to sum ito. As long as the bus impedance is comfortably less than the input impedance of the following amplifier than it should be OK.

10kohm resistors should work. 10kohm summing resistors feeding a 150ohm micpre will drop the level so much that there will be only the tiniest level at the summing point and so it will not back up into the sending circuits and that's a good thing. think of it as drain not clogged. The high gain of the mic pre will make up for signal level loss.

Depends. If the driving outputs are of low source impedance (like 75 ohms or so) the amount of signal from the bus reaching back to an individual output is very tiny. It is not going to cause the output amplifier any problem at all.

Cheers

Ian

 

[ChrissA]

OK, that makes sense.


So, just to be absolutely clear, before we go any further, this measurement was made with the pair of TRS just plugged into the output o f the M32.

If that is the case, what you are measuring is the output capacitance of the M32 which is not really relevant to the discussion

And again, before we go any further can you make up a lead with a TRS at one end and an XLR at the other and use it to connect the M32 output to a line input and measure the frequency response.

And then can you modify this lead to have 1K resistors in the hot and cold legs of the TRS and repeat the experiment.

We need to get some firm stakes in the ground.

Cheers

Ian

Hi Ian!

You are absolutely correct, the TRS cables were plugged into the output of the M32.

The measurements I have taken were done by comparing the summing cable (both TRS plugs plugged into aux outputs but audio over just one of the two, the other wasn't assigned to anything) to a simple TRS to XLR cable, after gainmatching of course. So I had both TRS jacks of the summing cable plugged into aux outputs of the M32, XLR plugged into an XLR input on the M32, another simple TRS to XLR cable plugged into an aux output on the M32 and XLR input on the M32, assigned the same source to both aux outputs (the one side of the summing cable and the comparative output), gainmatched the inputs and then run comparative tests using pink noise and sine sweeps. I'd think what you're asking essentially the testing I have done, or does having the second TRS plugged in change enough to warrant redoing tests? Since if I'm correct in my assessment using the summing cable with one TRS plugged in, the other one disconnected (or perhaps even connected but without audio coming out), and the XLR plugged in essentially does the same as having a single TRS to XLR cable with resistors in place.

 

[Newmarket]

I think we are seeing why diagrams are so useful

 

[rh001]

Okay so I'm not following. What I'm doing now is summing line level into a 10kOhm line level input. The level drop is super managable, it's less than 12dB. It should be possible to do passive summing without active summing amps, there are commercial products being sold that do exactly that. Also if you would have read the past conversations you'd know that I have put summing resistors in the connector successfully. I did get away with it, we were looking into capacitance because it seemed like I had accidentally made a lowpassfilter (which is negligible with the 1k summing resistor, 0.5dB loss at 20kHz). It works now, just curious why with 10k resistors I had a -3db point at 12khz.

True, you're not following. You and some others. There are various sorts of janky cheap gadgets made for the unwashed masses that sum some signals together into a line level (bridging or medium or high impedance) which will have one or more downsides, such as loss of high frequency, and the unwashed will find it acceptable cuz the end use is undemanding and there's no money for more better anyway. Compared to the right way of doing things, the Yuzuru Hanyu way, you're expecting great results from this:



BTW that same accompyaning music is put to better use here::



Doing things the right way is usually a little more expensive in terms of material but less expensive in terms of lost time. Also, besides just getting the frequency response right there is the issue of how does it sound? And have you considered.....if summing into bridging (etc) impedance, the output circuit of each source sees the output of the other sources, which it will attempt to negate via negative feedback but negative feedback is not free nor perfect and it is not intended to sink or negate other signals...it is meant only for itself, it's own signal. Summing with only 1kohm resistors you're likely having increased distortion as the sources are trying to sink what seems to them as distortion. When using a virtual earth summing point there is no signal for the sources to sink away or if using a a mic pre transformer as summing notde the signal to be sinked away will be very tiny and rarely if ever a problem for the sources. Neve consoles from the 1970's used mic pre's for summing amps.

 

[JohnRoberts]

4. You will find that the members of this community are courteous and respectful of each other, so please reciprocate those gestures. Leave the flame-war mentality at another forum. Personal attacks and generally hateful comments (regarding race, religion, gender, sex, politics, etc...) will not be tolerated.

Come on guys let's be nicer.

JR

 

[Voyager10]

It works now, just curious why with 10k resistors I had a -3db point at 12khz.

If you're curious and have time on your hands, build a single TRS to XLR cable with a 10K resistor in, and measure the response.

Put the resistor at the XLR end of the cable, so the cable capacitance is before the resistor.

If it still gives a lowpass effect, try different resistor values. If the -3dB frequency varies inversely with resistor value, there must be capacitance on the input, that's all it can be.

If you don't get the same result with the single cable, you'll have to figure out the physical differences between the two.

(BTW, you often find a capacitance meter will give silly answers when measuring things which aren't a pure capacitor, like the mixer input).

 

[Newmarket]

(BTW, you often find a capacitance meter will give silly answers when measuring things which aren't a pure capacitor, like the mixer input).

Yes. There's more to it eg Cp or Cs modeI. I think the OP may be using a DMM making things more doubtful especially if not simply measuring an isolated capacitor.

 

[abbey road d enfer]

I've redone all measurements. In the meanwhile I have replaced the 10k resistors with 1k resistors, so the cable now are 2 TRS jacks that plug into the output of the M32 with 1kOhm resistors soldered to the + and - of the jack.

If this is correct, these resistors don't help anything. They just unduly load teheoutput, resulting in loss of level and potentially higher distortion and/or loss of LF.
Mixing resistors should be connected from a signal source to a receiver.
EG for a balanced signal, one resistor from Pin 2 of the source to pin 2 of the receiver, and one from pin3 of the source to pin 3 of the receiver.

 

[abbey road d enfer]

What is there to learn from it?

W if using a a mic pre transformer as summing node the signal to be sinked away will be very tiny

This is very debatable. True only if the reflected impedance at the primary is significantly smaller than the mixing resistors.
In the absence of relevant data this is open to speculation.

 

[ChrissA]

If you're curious and have time on your hands, build a single TRS to XLR cable with a 10K resistor in, and measure the response.

Put the resistor at the XLR end of the cable, so the cable capacitance is before the resistor.

If it still gives a lowpass effect, try different resistor values. If the -3dB frequency varies inversely with resistor value, there must be capacitance on the input, that's all it can be.

If you don't get the same result with the single cable, you'll have to figure out the physical differences between the two.

(BTW, you often find a capacitance meter will give silly answers when measuring things which aren't a pure capacitor, like the mixer input).

I definitely will do that tomorrow. I've been away for the weekend but continue the search. Does it matter a lot if I instead of solder them into the plug make it more modular and put alligator clips in between? The -3dB point did vary with the summing cable resistor value but not in a way I would have expected. Going from 10k (with -3dB at 12k) to 2k should have put the -3dB point way out of human hearing range but it didn't.

Good to know the capacitance meter will give silly answers because it does not make sense to me in the slightest. I tested everything, thought it was weird and then redid all the tests but the results were virtually identical. It still all is a mystery to me lol

 

[ChrissA]

Yes. There's more to it eg Cp or Cs modeI. I think the OP may be using a DMM making things more doubtful especially if not simply measuring an isolated capacitor.

It don't know what Cp and Cs models are, but I am using a DMM (digital multi meter, right?)

 

[ChrissA]

If this is correct, these resistors don't help anything. They just unduly load teheoutput, resulting in loss of level and potentially higher distortion and/or loss of LF.
Mixing resistors should be connected from a signal source to a receiver.
EG for a balanced signal, one resistor from Pin 2 of the source to pin 2 of the receiver, and one from pin3 of the source to pin 3 of the receiver.

Its late here so I might understand in the morning, but I'm not following. To prevent confusion like there has been before I have made some (extremely janky) diagrams. This is how I have soldered everything up now:


From what I understood this prevents connecting an output directly to an output, and that's why it's needed. If I have understood you correctly you say I should solder it up like this, but doesn't this connect output 1 directly to output 2? I'm under the impression that is bad for the hardware.