Best ways to add balanced output to line-level circuit?

DirtyRegis

Member
2019-12-12 6:24 am
Hey everyone,


I'm working on a circuit and the output stage is currently a 1Vpp wave coming out of a TL074 op-amp. I would like to include a balanced output (1/4" TRS), specifically one that can also be OK with an unbalanced situation (1/4" TS).


I have a few Cinemag transformers (CMOB-4: 600:600ohm & CM-10100: 10k:10k) on hand. I tried the CMOB-4 and it did work, but I don't know if the situation is "ideal". Is a TL074 OK with driving a 600 ohm primary? and is a 600 ohm secondary appropriate for going out as Line-Level audio - where presumably it would be going into things with 100k input impedance.


I've also seen that there are some dedicated chips for "balanced line output" but I've never tried one before.


Maybe there are some other circuits I should be looking at?



I'm also wondering if I can I just invert the signal with a 1x gain inverting op-amp stage as the negative part of the balanced signal? If it has a 1K resistor in series with the output, presumably it wouldn't freak out if it gets grounded by the TS cable in the unbalanced application?


Thanks in advance!




body,div,table,thead,tbody,tfoot,tr,th,td,p { font-family:"Liberation Sans"; font-size:x-small } a.comment-indicator:hover + comment { background:#ffd; position:absolute; display:block; border:1px solid black; padding:0.5em; } a.comment-indicator { background:red; display:inline-block; border:1px solid black; width:0.5em; height:0.5em; } comment { display:none; }
 
The TL07* has a 100R output impedance so 600R is ideal as is a 600R line impedance; that is the standard.


This is an optimal method of driving a balanced line.
I would choose TL072.
 

Attachments

  • Screenshot 2019-12-12 at 10.02.33  00.png
    Screenshot 2019-12-12 at 10.02.33 00.png
    272.9 KB · Views: 530
Some version of this question comes up every so often. If you don't require differential drive, the most simple way to convert an unbalanced output to balanced output is shown in Jensen Transformers App. Note AN003, figure 2.4.

Basically, you can retain your existing single-ended circuit and add only one (an resistor) or two (possibly an capacitor) passive components. The result is very interesting, it being an single-ended output drive with a balanced interconnection. This solution enables every decibel of common-mode noise rejection as does a circuit featuring differential drive. Differential drive is NOT necessary to produce a balanced interconnection.
 

DirtyRegis

Member
2019-12-12 6:24 am
Some version of this question comes up every so often. If you don't require differential drive, the most simple way to convert an unbalanced output to balanced output is shown in Jensen Transformers App. Note AN003, figure 2.4.


Hi Ken, I've never seen this document before - thank you. It makes sense and is very simple.


But if I have an extra op-amp cell, what about implementing the "TYPICAL 'PRO' OUTPUT" in fig 3.1? The only issue I see is that since I'm using only a TRS jack, if someone plugs in a TS jack then the output of the second OPAMP would be grounded through the 100 ohm resistor (as shown).
 

DirtyRegis

Member
2019-12-12 6:24 am
Here's my circuit, adding the passive "Balanced output" from the Jensen AN003. Curious to get some feedback on this...


As you can see, the stage before I am mixing a dry and reverberated signal and then going into a volume control. Then I'm going into the NE5532 as a Voltage Follower to present a very high impedance to my volume potentiometer (so it's response/taper isn't modified by a low impedance).



The Jensen doc shows this circuit coming off of a non-inverting amplifier stage, but I need no further amplification - so is it OK to send the output direct from the Voltage Follower configuration as shown? (one of my concerns here is that the with the voltage follower configuration, I don't know if I can place a capacitor anywhere to roll-off high frequency stuff and avoid oscillating - if even necessary?).
 

Attachments

  • bal-out.jpg
    bal-out.jpg
    76.8 KB · Views: 416
Here's my circuit, adding the passive "Balanced output" from the Jensen AN003. Curious to get some feedback on this...

The Jensen doc shows this circuit coming off of a non-inverting amplifier stage, but I need no further amplification - so is it OK to send the output direct from the Voltage Follower configuration as shown? (one of my concerns here is that the with the voltage follower configuration, I don't know if I can place a capacitor anywhere to roll-off high frequency stuff and avoid oscillating - if even necessary?).

1) Your output circuit looks fine.

2) Jensen's passive balancing technique makes it possible to ground the passive 'output' phase without the concern that grounding an actively driven output would cause.

3) It does not matter whether the output op-amp is inverting or non-inverting. All that matters is that the balanced output phases feature the same A.C. impedance to ground, which your circuit does.

4) I'm uncertain of what capacitor you're referring to. If you refer to having a capacitor in the op-amp's feedback loop, then that is not relevant to an op-amp voltage follower configuration. The feedback loop is a short, so a capacitor in parallel with that short has no effect. Just be certain to use an op-amp that is unity-gain stable, which the NE5532 is.
 
Last edited:
Hi Ken, I've never seen this document before - thank you. It makes sense and is very simple.


But if I have an extra op-amp cell, what about implementing the "TYPICAL 'PRO' OUTPUT" in fig 3.1? The only issue I see is that since I'm using only a TRS jack, if someone plugs in a TS jack then the output of the second OPAMP would be grounded through the 100 ohm resistor (as shown).

Implementing differential drive with a second op-amp essentially provides no benefit unless you require the additional voltage swing that differential drive enables. In addition, as you rightly realize, it makes grounding of that output phase potentially problematic.
 
One other thought on your output circuit:

You may possibly find that the two 0.47uF D.C. blocking caps. are undersized, producing some audible deep bass roll-off in conjunction with the 100k resistors to ground. Which is made even lower because they are electrically in parallel with the input impedance of whatever is the following connected device. If you don't percieve the deep bass as lacking, I wouldn't worry about it though.
 
One other thought on your output circuit:

You may possibly find that the two 0.47uF D.C. blocking caps. are undersized, producing some audible deep bass roll-off in conjunction with the 100k resistors to ground. Which is made even lower because they are electrically in parallel with the input impedance of whatever is the following connected device. If you don't percieve the deep bass as lacking, I wouldn't worry about it though.




Hi Ken, thanks for the reply, and the bass roll off is concerning me too (at least on paper). Previously I used bipolar 10uF electrolytics as DC blocking caps, but their lifespan always concerns me (especially seeing lots of failed electrolytics moonlighting as a guitar amp repair tech). I could probably up the film caps to 1uF though without the cost or size going up too drastically.


I also have a question about the 100k resistors to ground in this design. If this was plugged into something with a 100k input impedance, these resistors would essentially drop that to 50k impedance right? Why not have them large like 1M to avoid this?
 
Here's my circuit, adding the passive "Balanced output" from the Jensen AN003. Curious to get some feedback on this...


As you can see, the stage before I am mixing a dry and reverberated signal and then going into a volume control. Then I'm going into the NE5532 as a Voltage Follower to present a very high impedance to my volume potentiometer (so it's response/taper isn't modified by a low impedance).


Do you know that the relative phasing will be right going into the dry/wet mixer pot?


Why is the 47pF cap not on the input to the NE5532 where is can reduce pot-crackle (normally you'd isolate a bipolar opamp input from any pot wiper with a blocking cap and provide alternative DC bias to avoid loud scratching as the wiper moves and interrupts the DC bias). I think the 47pF is designed to reduce this effect, but is in the wrong place?


The NE5532 has good output drive so its a good choice to push out the signal, but its not ideal DC-connected to a pot like that - all pots can be a little bit scratchy and if the bias current flows through the wiper that translates to full swing output noise in the worst case. JFET opamps have only pA of input current, easily smoothed by their input capacitance at all audio frequencies in the scratchy-wiper scenario!
 
Do you know that the relative phasing will be right going into the dry/wet mixer pot?


Hi Mark, the reverb phase is correct. Interestingly, in testing it's actually irrelevant because the reverb is so random that there's no phase correlation with the original anyways.



Why is the 47pF cap not on the input to the NE5532 where is can reduce pot-crackle (normally you'd isolate a bipolar opamp input from any pot wiper with a blocking cap and provide alternative DC bias to avoid loud scratching as the wiper moves and interrupts the DC bias). I think the 47pF is designed to reduce this effect, but is in the wrong place?


I added the 47pF to roll off any high frequencies just as a precaution, but I could be wrong in it's placement for sure.. or even its necessity. It was more on there as a "test to see if this is even needed".




The NE5532 has good output drive so its a good choice to push out the signal, but its not ideal DC-connected to a pot like that - all pots can be a little bit scratchy and if the bias current flows through the wiper that translates to full swing output noise in the worst case. JFET opamps have only pA of input current, easily smoothed by their input capacitance at all audio frequencies in the scratchy-wiper scenario!


Thank you. I'm only familiar with JFET opamps and have tested this configuration without any scratchy-wiper. This will be my first time using the NE5532. So I put a capacitor between the wiper and the NE5532 to block DC from hitting the wiper?



How do I calculate the roll-off frequency of this blocking capacitor in that position? Does the wiper position affect this calculation? Do any other components need to be added to avoid scratchy noise (a resistor to ground after the cap & before the NE5532 non-inverting input?)
 
The TL07* has a 100R output impedance so 600R is ideal as is a 600R line impedance; that is the standard.


This is an optimal method of driving a balanced line.
I would choose TL072.


The TL07* can drive loads down to 2k or so, not 100. I think you confuse output drive capability with open loop output impedance (which isn't relevant, close loop output impedance will be sub-ohm levels typically anyway).


Overloading the output of an opamp will cause distortion to start rising rapidly. With less output swing it won't be as bad though, so perhaps TL07* can handle 1k at low levels without too much problem.



NE5532 can drive down to 500 ohm loads or so, one of its strengths.
 
I'm only familiar with JFET opamps and have tested this configuration without any scratchy-wiper. This will be my first time using the NE5532. So I put a capacitor between the wiper and the NE5532 to block DC from hitting the wiper?
So long as you add a DC bias path too - but that could load down the volume control, or be such a large resistor the offset voltage becomes an issue (premature clipping). Input bias current x input resistor = extra input offset voltage.
The implication is that for bipolar opamps the volume pots should be quite low in value 10k or less, so biasing resistors aren't too huge. You get the added benefit of less thermal noise from the pot too.
 
So long as you add a DC bias path too - but that could load down the volume control, or be such a large resistor the offset voltage becomes an issue (premature clipping). Input bias current x input resistor = extra input offset voltage.


The implication is that for bipolar opamps the volume pots should be quite low in value 10k or less, so biasing resistors aren't too huge. You get the added benefit of less thermal noise from the pot too.

Thanks Mark. Okay - this is the first time I've ever had to think about input bias current. I see on the datasheet that it is .2mA typ (.8mA max). So let me think out loud:

So I'm choosing an input leak resistor that is small enough to give me a an acceptable DC offset: 100k would give me 20mV typical or 80mV max. 1M would give me 200mV to 800mV, etc. And if I went with the 100k resistor, it now loads down my 100k volume pot, and I minimise that by changing the pot to 10k.

That 20-80mV DC offset that is building up at the input would also be on the output - but since it's AC coupled with the caps I don't need to worry about it. The only worry really is if the DC offset got so high as to push the opamp into clipping? (which would happen if there was no leak resistor at all).
 
Thanks Mark. Okay - this is the first time I've ever had to think about input bias current. I see on the datasheet that it is .2mA typ (.8mA max). So let me think out loud:
No, you're out by a factor of 1000. 200nA = 0.2µA - but your calcs seem correct. With bipolar opamps you always need to think about bias and offset currents affecting the offset voltage (if both inputs see the same DC resistance then usually only the current offset spec matters).

So I'm choosing an input leak resistor that is small enough to give me a an acceptable DC offset: 100k would give me 20mV typical or 80mV max. 1M would give me 200mV to 800mV, etc. And if I went with the 100k resistor, it now loads down my 100k volume pot, and I minimise that by changing the pot to 10k.

That 20-80mV DC offset that is building up at the input would also be on the output - but since it's AC coupled with the caps I don't need to worry about it. The only worry really is if the DC offset got so high as to push the opamp into clipping? (which would happen if there was no leak resistor at all).
Yup.