Attenuating Unbalanced to Balanced Signal. Instrument level to Mic Level

[Colortone]

Hi,
Ive been using the Dual Opamp schematic from sound-au.com below to convert unbalanced to balanced however the circuit assumes you want a gain of 1 or over.
Im using it to convert a instrument level signal to a balanced mic level and adding a 40db Upad on the output however I can see that it is adding quite a lot of noise to the waveform, not necessarily audible. The Circuit is possibly also loosing a very slight bit of 'presence'.

As an alternative would it work to attenuate within the circuit block starting with the - inverting input for U1A and then to the noninverting + input and if so would it achieve better results than adding a large Upad to the output?

Thanks,
AD

 

[k brown]

If your your source is OK with a rather low load impedance, a simple 2-5k pot on the input would allow you to attenuate the input signal as much as you want.

Raising all the 10k resistors to 50-100k would allow use of a larger value pot for a higher input impedance, at a small cost in self noise.

 

[Colortone]

Thanks K Brown,
Interesting idea. The input will be a guitar signal looking for a @1M load. My current version of the circuit has R2 at 1M and removes R1.

 

[k brown]

Thanks K Brown,
Interesting idea. The input will be a guitar signal looking for a @1M load. My current version of the circuit has R2 at 1M and removes R1.

For the circuit to present that high a load, you'd have to replace all the 10k resistors with 1M (and NOT remove any), for the circuit to remain balanced.

For a pot at the input - say a 1M pot, and 3-5M resistors in the circuit (must all be the same value as you see the 10ks in the original); a lot of self noise, but with guitar as input, may not be a problem.

 

[ccaudle]

I thought maybe you meant line level. Guitar isn't too high level even with humbuckers, so you could use an impedance balanced output, single op-amp to buffer the guitar and drive the hot output through an attenuator, and the cold output through just the same attenuator arrangement to "ground." I almost hate using that word (ground), because the circuit as drawn doesn't indicate that the pin 1 of the balanced output is the shield pin, and shouldn't tie directly to the circuit reference node, but that discussion can wait until you nail down the circuit topology.

 

[Colortone]

For the circuit to present that high a load, you'd have to replace all the 10k resistors with 1M (and NOT remove any), for the circuit to remain balanced.

For a pot at the input - say a 1M pot, and 3-5M resistors in the circuit (must all be the same value as you see the 10ks in the original); a lot of self noise, but with guitar as input, may not be a problem.

Sorry drawing in Gimp :/ I added some DC filtering on the output as well in case its connected to a Desk with Phantom power which I dont want to see as this has previously caused issues.
So instrument level in - Balanced Mic Level out. Attenuation at the start instead of Upad at the end.
Would the idea of using a inverted opamp input first work? ( Assuming id have to swap the hot and cold outputs to keep the phase correct?)

 

[Colortone]

Also because U1A has a native gain of 2 and im trying to attenuate is it ok 'balance wize' to make the Feedback resistors R3 and R6 1/10th of the others to take my output gain down?

 

[k brown]

No, all resitors in the original schem changed to 3-5M; 1M pot placed on the input, before the junction of R1 and R5 - wired as a pot, not a variable resistor as you drew.

 

[Colortone]

No, all resitors in the original schem changed to 3-5M; 1M pot placed on the input, before the junction of R1 and R5 - wired as a pot, not a variable resistor as you drew.

Im Soooo sorry, please bear with me but I pasted the wrong starting point circuit however I think what youve been kindly telling me will remain the same.

The circuit I have been using is actually this one. where the "-" block is driven by the output of U1A. This one from the musique.com site which is similar topology to the correct one I meant to reference on the sound-au site.
Do the same rules still apply in terms of just adding a POT at the very beginning of the circuit and changing up all the 10k resistors to 3-5Meg?
Here my initial question will make more sense, can I swap U1A and U1B for this to work with gain reduction or am I still better off attenuating the input first?

 

[k brown]

Personally, I'd just replace R5, C4 and R6 with with a 10k resistor, raise R3 to 3-5M, and put a 1M pot on the input (before C2). Then the circuit will have very low gain and the input impedance will be a little less than 1M. Then the pot can set any amount of attenuation you require.

 

[Colortone]

Personally, I'd just replace R5, C4 and R6 with with a 10k resistor, raise R3 to 3-5M, and put a 1M pot on the input (before C2). Then the circuit will have very low gain and the input impedance will be a little less than 1M. Then the pot can set any amount of attenuation you require.

Thanks for your time K Brown I really appreciate it.
Ill try this in the next day and compare the results on the scope. Are there any other circuit topologies or THAT style chips that may give me a better cleaner output instead of this slightly 'hacked' method I started down?

 

[JohnRoberts]

If you are trying to get it down to mic level that still looks too hot.

Maybe just buffer it then add a passive pad, -30/-40dB will get you down to mic level.

JR

 

[Colortone]

Thanks John, I currently have a passive -40db Upad at the tail but on the scopes at least this seems to introduce a lot of noise and potentially loosing a little life/specularity on output. I read a H-Pad may help with the loss of life compared to a UPad?
Given Ill try K Browns approach which would end up a fixed LPad am I better off using a T-Pad for this?

 

[k brown]

If you are trying to get it down to mic level that still looks too hot.

Maybe just buffer it then add a passive pad, -30/-40dB will get you down to mic level.

JR

How could a pot on the input, as I suggested, be 'too hot'? - It can be turned down to zero.

Plus, instrument level is between mic and line level.

What I suggested requires no pad anywhere.

 

[Newmarket]

Personally, I'd just replace R5, C4 and R6 with with a 10k resistor, raise R3 to 3-5M, and put a 1M pot on the input (before C2). Then the circuit will have very low gain and the input impedance will be a little less than 1M. Then the pot can set any amount of attenuation you require.

Why have any gain at U1A ? Just run it as a unity gain buffer.

 

[Newmarket]

I read a H-Pad may help with the loss of life compared to a UPad?

Pads are resistive only and don't have any effect on the frequency response (ignoring stray reactive impedances etc).
There may be interaction with the following input but that (obviously) depends on the impedance of that (unknown) input - presumably some flavour of mic pre ?
Listening at different levels has a significant effect on what sounds "better".

 

[abbey road d enfer]

How could a pot on the input, as I suggested, be 'too hot'? - It can be turned down to zero.

And result in very poor signal-to-noise ratio.
There is no reason for attenuating the signal from the guitar before hitting the active circuitry, which is perfectly capable of handling it.
All it needs is an adequate output pad.

 

[abbey road d enfer]

, I currently have a passive -40db Upad at the tail but on the scopes at least this seems to introduce a lot of noise

I don't know how a passive attenuator could introduce noise. Measuring noise with a scope is utterly inadequate, particularly, but not only, because it sees noise across its whole bandwidth, which is many times more than the audio range; noise increases with bandwidth, so you see a lot of RF noise that is never heard.

and potentially loosing a little life/specularity on output.

Probably because its impedance is inadequate.
You need an attenuator with an input impedance of about 5-10k and an output impedance of about 200 ohms.
A U-pad with 3.3k series and 220 shunt will get you there, with about 300 dB attenuation.

I read a H-Pad may help with the loss of life compared to a UPad?
Given Ill try K Browns approach which would end up a fixed LPad am I better off using a T-Pad for this?

Textbook attenuators are constant impedance; it means that the input impedance is equal to the output impedance.
It's not what you want here.

 

[Colortone]

I don't know how a passive attenuator could introduce noise. Measuring noise with a scope is utterly inadequate, particularly, but not only, because it sees noise across its whole bandwidth, which is many times more than the audio range; noise increases with bandwidth, so you see a lot of RF noise that is never heard.

Probably because its impedance is inadequate.
You need an attenuator with an input impedance of about 5-10k and an output impedance of about 200 ohms.
A U-pad with 3.3k series and 220 shunt will get you there, with about 300 dB attenuation.


Textbook attenuators are constant impedance; it means that the input impedance is equal to the output impedance.
It's not what you want here.

Thanks Abbey road d enfer. You helped me on this some time back and I have your output UPad recommendation currently on the circuit. I started going down this road of attenuation options because the vocalist mentioned the circuit was loosing a little presence, and I mean a little bit that I found difficult to discern the difference but her voice is her job so Im looking for flaws. Good to know the scope is not ideal for looking for this. Im re-testing all my input and output impedances tomorrow. The current output impedance is 47 ohms, using the pot on output method to measure 1/2 input level. Are you aware of a THAT style unbalanced to Balanced chip that would natively attenuate that I could look at for future iterations?

 

[Newmarket]

Are you aware of a THAT style unbalanced to Balanced chip that would natively attenuate that I could look at for future iterations?

Why not simply buffer the guitar signal (Op amp stage / 1M input Z) then a resistive divider to give you the chosen attenuation, or use a pot to make it variable if desired, - then another unity gain buffer op amp stage that is the active element in an impedance balanced output. Sorted !
Or you could use that buffer output feed a THAT Outsmarts balanced driver device.
You'll want a FET type Opamp for the input stage but a low noise bipolar type would be preferable for the output. Although by no means absolutely necessary given the unity gain operation.
Remember that the THAT device will give you +6dB so you'd need to attenuate further to allow for this.