Geekslutz DIY Summing Mixer

[Tim Farrant]

If the output transformer had a grounded centre tap on the primary, would you say the 100uF capacitors are still in series? I think not. Assuming the signals on the output of the opamps are 180 degrees out of phase and equal in level, the centre point of the transformer primary winding has no signal present. Each opamp "sees" half the load impedance presented on the secondary. With a 10k load, that's 5k, which works out to a -3dB point of around 0.3Hz.

[Matt Syson]

Hi
The situation of having a grounded centre tap is entirely different compared to a non grounded centre tap. When it is not grounded you have a series circuit where the voltage the primary 'sees' is simply double the voltage from one amplifier output. To do otherwise contradicts basic laws of electricity as it is a series circuit, the current cannot go anywhere else.
While at 0.3 Hz the response may be 3dB down unfortunately when measured the 'resonance' shows a 2dB boost at 10 Hz. As i wrote earlier, reducing the cap predictably changes the resonant frequency, giving a 6dB peak at 15 Hz and very sharp cutoff below 15 Hz. If I said 'the basis of a Pultec equaliser' others may take notice.
Heavily loading the output with 600 Ohms (the way the transformer was designed) the resonance disappears and you get the predictable roll off at HF and LF.

[Tim Farrant]

Originally Posted by Matt Syson

Hi
The situation of having a grounded centre tap is entirely different compared to a non grounded centre tap. When it is not grounded you have a series circuit where the voltage the primary 'sees' is simply double the voltage from one amplifier output. To do otherwise contradicts basic laws of electricity as it is a series circuit, the current cannot go anywhere else.

This is true if both legs were in phase, but because they are out of phase, then the centre of the transformer primary winding has zero signal, because the signals cancel. Common mode rejection relies on this situation. If you did have a floating centre tap on the primary, and you looked at it with a scope, there would be no signal (assuming perfect balance).

[Matt Syson]

Hi
When fed with the centre tap floating yes indeed there would be no signal COMPARED TO GROUND but the transformer only cares about the potential from the end of the primary winding to the other end, which is in this case is the algebraic sum of the 2 amplifier outputs. The transformer does not care if it is a balanced signal or unbalanced, only the Voltage difference presented to the ends of the windings.
Since these transformers have half primaries (1+1:2) ratio then indeed the voltage compared to ground at the centre tap will be theoretically zero with only manufacturing tolerances and imbalance allowing some signal to appear.
Common mode 'rejection' works because the transformer only 'transmits' a potential (current) DIFFERENCE as presented to the ends of the primary. If both ends have the SAME signal (voltage and phase) then there is no DIFFERENCE so it will not transmit any signal to the secondary.
This situation is only spoiled by imbalance of the windings, usually capacitive.
Matt S
PS, not wishing to cloud this issue but if you have a centre tapped transformer at each end of a balanced line you can insert DC or indeed AC onto the centre taps with respect to ground, then extract the DC or AC from the second transformers centre tap, again with respect to ground. The DC case is commonly seen as phantom for mics of course but the same argument applies for an AC signal, but it does lead to crosstalk due to imbalance. Mic amps usually do not have centre tapped input transformers at least not with phantom applied to the centre, but rather a pair of resistors. This has an advantage of economy and improved 'balance' by using matched resistors (should be 0.1 percent or better).

[Tim Farrant]

Originally Posted by Matt Syson

Hi
When fed with the centre tap floating yes indeed there would be no signal COMPARED TO GROUND but the transformer only cares about the potential from the end of the primary winding to the other end, which is in this case is the algebraic sum of the 2 amplifier outputs. The transformer does not care if it is a balanced signal or unbalanced, only the Voltage difference presented to the ends of the windings.
Since these transformers have half primaries (1+1:2) ratio then indeed the voltage compared to ground at the centre tap will be theoretically zero with only manufacturing tolerances and imbalance allowing some signal to appear.
Common mode 'rejection' works because the transformer only 'transmits' a potential (current) DIFFERENCE as presented to the ends of the primary. If both ends have the SAME signal (voltage and phase) then there is no DIFFERENCE so it will not transmit any signal to the secondary.
This situation is only spoiled by imbalance of the windings, usually capacitive.
Matt S

Yes, this is my understanding as well, so for this reason, the 100uF are not in series as such, but they do see 1/2 of the output load each.

And, omitting the transformer from the summing circuit may cause problems if an unbalanced input is used. The transformer ensures that all inputs are "re-balanced" despite the input format.

Also, the network on the secondary of the transformer should tame any high frequency rise due to the transformer being loaded higher than 600 ohms.

[Matt Syson]

Hi
I have used an Audio Precision to test these results and for one the output compensation is not sufficient to tame the HF peak when fed a decent length of cable (different values needed) and
the 100uF caps result in a 2dB rise at 10Hz. Take out one cap and the rise is about 1dB and take out both and it is 'flat' (actually about 0.2dB fall which is correct for this transformer).
The caps ARE in series because your schematic shows the physical 'centre tap' not being used so the current HAS to flow through both and not into a 'mystical' null point half way along the windings.
Matt S

[ulysses]

Originally Posted by Tim Farrant

And, omitting the transformer from the summing circuit may cause problems if an unbalanced input is used. The transformer ensures that all inputs are "re-balanced" despite the input format.

The output is balanced with or without the transformer, regardless of whether or not the input sources are balanced.

[Matt Syson]

Hi Ulysses and Tim
I think what Tim was getting at is that the transformer ensures the output is balanced. Without it if you had only unbalanced sources, while the output is strictly 'balanced' there would be no great contribution from the 'opposite' phase being thus the same as an unbalanced output.
It would still of course 'contribute' to hum and interference cancellation which would be lost if the transformer were omitted and it is fed into an unbalanced stage.
Matt S

[Tim Farrant]

I'm just about ready to give up, but I'll try and say this another way...

Assuming the inputs are perfectly balanced, then the output of each opamp will be 180 degrees opposed. If you took those two outputs and summed them together, there would be no signal, because they cancel each other, yes?

Now, if the transformer was replaced with 2 5k resistors in series, and you fed the output of the opamps (180 degress opposed) into each end of this network, what would you see at the centre tap of the two resistors? It's a summing point, yes?

Transformers are impedance convertors, so in reality, with a 10k load and 1:1 ratio, the primary looks like 2 5k resistors in series with a grounded "no signal" (voltage wise) centre tap with respect to the 180 degrees opposed signals. If the signals were in phase, it's a different story.

It's exactly the same as driving a loudspeaker coil with a bridged power amplifier.

There is nothing mystical about it.

Tim.

[Matt Syson]

Hi
[Assuming the inputs are perfectly balanced, then the output of each opamp will be 180 degrees opposed. If you took those two outputs and summed them together, there would be no signal, because they cancel each other, yes?]
No signal relative to 'ground' BUT a there is still signal current flowing through the summing resistors.

[Now, if the transformer was replaced with 2 5k resistors in series, and you fed the output of the opamps (180 degress opposed) into each end of this network, what would you see at the centre tap of the two resistors? It's a summing point, yes?}
It is a 'summing point' only to the extent that the potential difference relative to 'ground' would be zero. Mr Kirchoff's law is upheld.

[Transformers are impedance convertors, so in reality, with a 10k load and 1:1 ratio, the primary looks like 2 5k resistors in series with a grounded "no signal" (voltage wise) centre tap with respect to the 180 degrees opposed signals. If the signals were in phase, it's a different story.]

Assuming the outputs of the amplifiers are 180 degrees in opposite phase then a 'centre tap' would 'appear' to be grounded. If however the signal is NOT balanced going into the transformer then the centre tap will still have half the overall voltage which is presented to the windings on it but this would NOT be zero volts relative to 'ground'. Take for example the case if the inputs (and by implication the output of one of the amplifiers) is UNbalanced then the voltage you would see on the centre tap is half the 'active' voltage, relative to ground.
In this situation, joining the centre tap to ground would be a bad idea as by transformer action the active half would try to force current back into the inactive amplifier.
There is nothing mystical as you say but you have to consider where 'real' electrons are flowing and they certainly can't from a mystical node part way down a transformer winding, it HAS to come out of the 'ends'. A transformer is a current operated device and only cares about the current flowing through the wire, not where it is referenced to in the wider world.
Matt S

[ulysses]

Originally Posted by Matt Syson

I think what Tim was getting at is that the transformer ensures the output is balanced. Without it if you had only unbalanced sources, while the output is strictly 'balanced' there would be no great contribution from the 'opposite' phase being thus the same as an unbalanced output.
It would still of course 'contribute' to hum and interference cancellation which would be lost if the transformer were omitted and it is fed into an unbalanced stage.

It doesn't matter if there's any signal on the cold output or not. It's still a balanced output, with or without the transformer. Balanced interfacing has nothing to do with signal symmetry. It only requires impedance symmetry.

In fact, this circuit will still be balanced without the transformer OR the op amps.

[Matt Syson]

Hi
While I agree it is balanced in all the cases mentioned, the way it would react to unbalanced sources would be that it would 'clip' at about +21dBu output whereas if fed from symmetrical sources it would not clip until about +27 dBu.
Matt S

[JAY X]

Hi All,

Today i Made an audio test of the summing mixer. OP275 opamp.
No transformer outputs. No preamp for gain recovery. Outputs are balanced with 0 volts comming from the ssl9k psu. I did not remove the output capacitors.

The setup is: Pioneer DVD464 to Summing Mixer inputs 1 and 2 , Summing mixer Stereo outputs to Waveterminal 192M inputs, to Cubase sx3.

Samples recorded at 24 bits 44,1khz then exported as mp3 320 kb/s

There are 3 tracks:

1. Original DAW mix
2. Summing mixer no compressor
3. Summing mixer Stereo track compressor.
(compressed in Cubase, sonnox BUS 1)

I wait for your comments regarding sound quality, Highs and lows, etc..

Have Fun.

JAY X

[JAY X]

Hi friends,

As you may noticed after the sound samples, there is a slight level imbalance between the left and right channels.

¿can anyone guess why? ¿maybe the feedback resistors, or the output Capacitors?.
¿maybe this is normal because of the veroboard layout?

Note: I have left out the output transformers, but not the output capacitors.

Thank you,

JAY X

[Matt Syson]

Hi
Using the DAW you should be able to tell us what the level difference is.
Put the same signal level tone into both channels and measure the outputs. Use the SAME channel input on the DAW to measure left hen right as it may be the DAW is incorrect.
Assuming the board layout is the same for both channels the significant difference would be due to resistor tolerances but using 1 percent types should only give a small fraction of a dB difference (you could work it out with one resistor 1 percent low and the feedback 1 percent high to give the 'worst case').
Matt S

[ulysses]

The worst-case scenario for a single pair of mismatched 1% resistors is about 0.17dB. This could compound across a circuit but that's statistically unlikely.

The most likely cause of your level mismatch (assuming it's in the range of a dB or two) is uncalibrated DAC and ADC channels.

But let's find out how big the mismatch is. If it's 6dB, we'll have a different assessment for you.

[JAY X]

Hi Friends,

Today i tested an LCR switch (by new york dave), recording a bass drum from a drum machine (Roland CR8000). Notice the difference in sound quality when routed to the left, center and right speaker. 9k1 resistors used.

I still have not made major changes to the circuit.(pending)

Please comment over the sound and Cubase statistics.




Sound clip mp3:

[JAY X]

Hi All,

Today the LCR circuit works perfectly. No audible differences between channels.

It seems that this mixer takes its time to get used to its new extensions..

The more i get into this project i see its many possibilities. I think, about the end of October i will finish it.

JAY X

[JAY X]

Hi All,

Today i completed all audio/power connections.

Finaly i decided to make the inputs and outputs of the preamp with normal polarity.
The outputs of the summing amplifier go with inverted polarity.

Now, the problem is the fuse, (100ma). With a fast fuse i got current, but those blow down at the second or third time i power up. With a slow blow fuse i get no current at all, no voltage, zero. Maybe i have to use a semikick fuse.
The slow fuse is a SHURTER, type FST. T 100ma.

The toroidal transformer is 30w 2x110v primary, 2x 18v secondary

¿Any suggestions?

Some specs of the project:

SUMMING SECTION
16 unbalanced jack inputs.
2 balanced outputs
16 LCR rotary switches

PREAMPLIFIER
2 XLR front/Rear inputs
2 balanced outputs
2 JLM go between

VUMETER (output volume)

2 JLM Audio Vumeter+ Vu buffer kit

Pictures coming soon!

Thank you for yor help,

JAY X

[Matt Syson]

Hi
If it is working OK before the 'fast' fuse blows then your unit is fine.
I would expect this fuse to blow at 'switch on', but if it has held in, then the unit should work.
If the T rated 100mA is not working, you either have some incorrectly labeled fuses or if it is a different fuseholder you are using for this type, then you have wired it incorrectly.
Matt S

[JAY X]

Hi Matt,

I went to a hardware store this afternoon, and they told me that probably the fuses (5x20 T100ma 250v), may have a defect. It is not normal that the fast fuse works, and the slow doesn't.

I will try all the 12 fuses i ordered. At least one should work.

JAY X

[JAY X]

Hi Matt,

Today i tested 3 125ma T fuses that worked well at first time, then blew at second power up.

I think the inrush current of the toroidal is too big and need to be limited. I think i need a Varistor circuit, between the toroidal primary and the mains switch.

I will see if i find something out there...

JAY X

[Matt Syson]

Hi
You should be looking for a 'surgegard' or it's equivalent.
they are NOT varistors which limit over voltage but they are Negative Temperature Coeficient (NTC) resistors. Basically they have a relatively high resistance when COLD but when curent flows through them they get HOT and the resistance falls drastically. THEY RUN HOT so you must allow for this when placing them.
I have used larger current types very successfully for bigger transformers but you need one with say 120 Ohms or more when cold but are suitable for use at 100mA or even a bit less.
I am a bit suprised that the fuses you have don't hold in, have you got very big capacitors on the supply before the regulators?
I suppose the transformer could be taking 130mA if running 'flat out' so you could simply use a 200mA T fuse and see how you get on.
You are messing with mains now, take care!
Matt S

[Harpo]

Originally Posted by JAY X

The toroidal transformer is 30w 2x110v primary, 2x 18v secondary

Hi Jay,
for whatever reason you need a 30VA toroid here, when powering up the empty caps after the bridge rectifier nearly represent a short for some AC cycles with inrush current limited by transformer VA rating. For 110V mains I'd probably try (30VA/110V=272mA) a 250mA/(T)ime-lag or for 230V mains a 125mA/T fuse. Voltage rating of these fuses is 250V to prevent arcing when they blow. As Matt already suggested, a NTC might help, but these are most often only needed above 50VA rated toroids. As always YMMV.
-Harpo

[JAY X]

Thanks Matt,

I use gloves...

The PSU is a SSL9k psu.
Capacitors:
2 x 1000uf/35 v
2 x 100uf/100v
1 x 220uf/100v
6 x 100uf

the 1000 uf followed by the 100nf, before the regulators.(LM7818/7918) see the attached pdf.

¿How should i place these NTC Resistors? ¿Could you please draw me a schematic?

At home my wall sockets give 15 amps at 230v, and sometimes i read transients at 1350volts... ¿maybe i need a surge protected socket?




JAY X

[spdrman288]

Here is my humble Passive summer.

[Matt Syson]

Hi Jay
Look at the transformer makers website for suggested fuse ratings.
You should not need a big fuse as the caps are small. Have you measured the current taken from the transformer secondaries, it may be you have messed up connecting the 2 power supply boards and have some strange short going on.
Check the mains current when running, does it come to about the right amount?
Say DC current times DC voltage times 1.6 is about output power times 2 for being inefficient plus say 2 watts for transformer losses.
Rather than a NTC you could simply wire a 39 Ohm 25 watt 'gold block' resistior in series with the mains. It will lower the AC into the transformer a bit but for a preamp it probably won't matter.
Matt S

[svart]

Wow I haven't checked this thread in a while!

I'm kinda late to the game but the "balanced" discussion intrigued me.

It seems that you guys were talking about both "balanced" and "differential" without coming to the conclusion that balancing means matching the source, series and termination impedances of a pair of conductors while differential signals are identical signals of opposite polarity that are generally transmitted ON balanced conductors. The signal itself cannot truly be "balanced" without the conductors being balanced but the conductors CAN be balanced without the signal being differential.

The transformer does not care if it is a balanced signal or unbalanced

On the primary side it does care because differences between the output impedance of one of the differential driver outputs and the other output driver impedance can cause a mismatch in balancing. This is likely to be very small and inaudible but academically it's present and can allow noise ingress.

The other side of this coin could be that what if 1/2 of the primary is wound slightly different than the other half? In an ungrounded CT transformer, ground is not referenced and therefor the outputs will essentially be balanced by design whereas a grounded CT will force the difference in windings to be apparent. Again this is academic since the difference will be so small that it won't likely matter, especially at AF.

[Matt Syson]

Hi
The amount that a transformer would care would be largely down to the properties of the transformer, how the windings stack, capacitance between differing windings and so on.
It all depends how well 'balanced' you want it to be and over which passband and out of band.
It is easier to predict or design for accurate balance on parts other than the transformer.
For a 'line level' signal, a balance which allowed say 60 or 70 dB common mode noise rejection is probably adequate and quite easy to achieve. Mic inputs should be better if possible (it is possible of course but at rather greater cost). Just sticking a pair of unselected 1 percent tolerance phantom power resistors can completely mess up the best intentions.
Matt S

[JAY X]

Hi All,

Today i talked with the maker of the transformer, and suggested me to use a 200 or 250 ma Slow blow fuse.

30w x 1.4 = 42W/220 = 0,190 ma.

That is why i Blown all my fuses..

JAY X