Signal tapping line level audio

[The Magic Rabbit]

I have a mixing desk that has inserts. I made a passive attenuator before but it wasn't very successful as it affected the signal on the desk too much.

So, my new thinking, is an active version.

There are 3 lines available, Ground, Send (signal out) and Return (signal in). The outputs are from the pre-amp, but before anything else, and are stated as being "at line level".

When the insert is unused the desk simply shorts the send and return internally.

Am I right in thinking that if I short the send and return externally (in my box) and put an Op-Amp in unity gain (1) tapping from that line, I should then get a copy of the signal from the desk on the output of the Op Amp that I can mess around with without affecting the desk (at least not noticeably)?

Do I need to find an Op-Amp that can take high input voltages (2V peak?) or does that become irrelevant if the gain is at 1?

Do I need a dual rail power supply, or can I use single rail?

Sent from my XT1039 using Tapatalk

[Hideki]

if I short the send and return externally (in my box) and put an Op-Amp in unity gain (1) tapping from that line, I should then get a copy of the signal from the desk on the output of the Op Amp that I can mess around with without affecting the desk (at least not noticeably)?

Yes.

Do I need to find an Op-Amp that can take high input voltages (2V peak?) or does that become irrelevant if the gain is at 1?

2V is not a high input voltage for most op-amps. A good old NE5532 is fine for such an application.

Do I need a dual rail power supply, or can I use single rail?

You can use a single rail if you AC-couple the input and output with capacitors and bias the op-amp input to half the supply voltage.

[The Magic Rabbit]

Thank you.

I'm not sure what you mean by AC couple. Is that the same as DC blocking?

I think I understand why I need it at output, since I need to remove the DC bias I added... but what is the one on the input for?

If I do single rail, do I still connect the Gnd from the desk to the power supply Gnd?

Sent from my XT1039 using Tapatalk

[The Magic Rabbit]

Additional thought: Since I plan to duplicate this for potentially 8 channels on the desk, would it make more sense to use something like a BUF634 to create a virtual ground (24VDC input to +12V, VGnd, -12V) - eliminating the need to bias and then DC block each channel?

Sent from my XT1039 using Tapatalk

[Hideki]

AC coupling is connecting a capacitor in series with the signal, so you can have different DC levels on each side but still pass (couple) AC through the capacitor. You can call it DC blocking if that makes it easier to understand.

If the send signal already has a capacitor on its output, you might not need to add one more cap. But there is most likely a resistor going to ground after the cap, to keep the output at ground level when nothing is connected and thus avoid a loud click when you do connect something. Study the mixer schematic. It could be a high value like 100k, but it could be 10k, and mess up the biasing to your chosen virtual ground voltage (depends on your chosen input impedance). Study the mixer schematic.

You can definitely use a BUF634 or similar device to make the virtual ground a bit more stable, but I take it you want to connect the output of the BUF directly to the mixer ground? That might work if the 24V supply is truly floating, but I would consider using a proper dual rail supply. DC/DC converter modules with +/-12V outputs are also available.

[The Magic Rabbit]

I was more thinking of cost saving, since dual rail power supplies are expensive, and I have quite a few spare DC power supplies from printers, laptops, etc.

I don't really understand what is meant by truly floating. All the supplies I have are switched mode type. How would I know?

Sent from my XT1039 using Tapatalk

[RJFreeman]

you can use the inserts as a line out (I have done it in the past) simply use a TRS plug into the Insert socket and short the Tip and the ring, then tap that off to a mono plug to feed whatever you are feeding.
The Op-amps (inside the desk)  will usually happily drive a load down to 1K ohm while input impedances of equipment (including the return on the insert socket) are usually over 10K ohm if you set it up correctly (if you need a passive attenuator, aim at an input impedance of 5K or more) you should not have excessive loading on the Insert.

Maybe though more specific information would help;
what desk are you using?
what equipment are you feeding from this?
how was your "passive attenuator" made up?
and why did you feel the need to pad the signal down?

I have quite a few spare DC power supplies from printers, laptops, etc.

Some of these switchmode power supplies can be noisy and even inject noise back into audio gear they might be connected to.

[The Magic Rabbit]

The problem I had with the passive setup was that the desk outputs too hot a signal for the computer interface, which results in a lot of clipping. The passive attenuator worked, but resulted in a noticeable drop in signal volume at the desk, and some loss of high frequencies. The computer interface also introduced noise, although that was solved by simply grounding the chassis to the desk chassis.

There is more to the project than just wanting to get a desk output. Once this works, I plan to add some kind of VU display for each channel, and maybe even its own microphone inputs. At this stage, I'm just wanting to keep things simple though. Get a good starting point and then build upon it.

Some of the cheap power supplies may be noisy, but some of them are expensive supplies, and I would be adding some caps on input anyway just to add smoothing. I have a box of various reclaimed electrolytic to use.

Sent from my XT1039 using Tapatalk

[fivefish]

BUF634s are high current line drivers (expensive, but if you'll be driving long lines, or an output transformer... sure go for it). But for use as a virtual ground? I think its overkill and waste of a part.

If you use a virtual ground to split a single 24VDC supply, your "virtual ground" will be different from the ground of the equipment. Your virtual ground is really 1/2 Vcc and you're just biasing your opamp to operate on that level... that's why you need a capacitor/AC coupling to isolate your circuit from the mixer.  You don't want to use DC coupling and sending 1/2Vcc back to the mixer's line level circuitry... that will be a huge dc offset,  and will cause distortion/clipping on the rest of the circuitry after the mixer's RETURNs.

Another option is use transformer coupling, that way there is complete galvanic isolation between your circuitry and the mixer's. Downside is cost and bulkiness.

[The Magic Rabbit]

I was thinking the BUF634 because it has a virtual ground application in the data sheet, and thinking of possibly 8 x Op Amps at maybe 20mA a piece meant I needed stable virtual ground at 160mA. Maybe I've got it completely wrong. I'm only a beginner.

Hence also looking at premium Op Amps. Intending to use the TL071 while experimenting, then maybe swapping out for AD843 or LME49990 since they're supposedly a much better sounding amp, for when I try to add microphone capability later.

The AD823 is also apparently a reasonable sounding chip, although not as good as the AD843, but can run off a single supply rail.

The LME49990 I just like the look of because you can get insane gain from it with extremely low noise and distortion... but the input voltage seems to be +/-0.3V Max which I would assume means microphone only.

At the moment a lot of things are a bit of a blur because I don't understand how the stuff in the datasheet correlates to sound quality. Determined to learn though.



Sent from my XT1039 using Tapatalk

[RJFreeman]

The problem I had with the passive set-up was that the desk outputs too hot a signal for the computer interface, which results in a lot of clipping. The passive attenuator worked, but resulted in a noticeable drop in signal volume at the desk, and some loss of high frequencies.

So it sounds like you loaded down the output too much, but to comment we would need to know specific details, mind you to understand what are trying to achieve would need specific details anyway, so more information would be useful.

 

The computer interface also introduced noise, although that was solved by simply grounding the chassis to the desk chassis.

So you already have a computer power supply with a noise problem (sounds like a double insulated supply) which does illustrate why using a random power supply could cause problems, or on the other hand might not cause problems.

There is more to the project than just wanting to get a desk output. Once this works, I plan to add some kind of VU display for each channel, and maybe even its own microphone inputs. At this stage, I'm just wanting to keep things simple though. Get a good starting point and then build upon it.

Fair enough, sounds a good idea to start with in which case you probably want to start taking this into account as you begin the build so you don't just end up making something several times over.

Some of the cheap power supplies may be noisy, but some of them are expensive supplies, and I would be adding some caps on input anyway just to add smoothing. I have a box of various reclaimed electrolytic to use.

Its not always cost, but more design requirements, for example a power supply designed for a printer might allow for higher surge currents, while noise may be less of a consideration.

On the other hand one of your supplies might already provide a dual rails, or you might have a 24V supply that is grounded which could cause issues even with decoupling and a virtual split rail, again without more information we wouldn't know?

the circuitry is fairly simple, just set-up your op-amps as Voltage followers (i.e. with the inverting input connected to the output, and use the non-inverting input) - Technically in this set-up there is no 0V rail involved so you don't even need to worry about the virtual ground.
at unity gain most Op-amps will provide pretty good results

[fivefish]

I think you read the 4990 datasheet wrong.  It's (V-) - 0.3V to (V+) + 0.3V 
Where V- and V+ is your supply voltage. (+/-19V max)

Maximum input is 12V (Common-Mode Input Voltage Range)

8 x Op Amps at maybe 20mA a piece meant I needed stable virtual ground at 160mA.

No, you need a power supply that can supply at least 160mA. 
Also.... 20mA per opamp is large. They're typically around 10-12mA or even less.

[The Magic Rabbit]

The computer power supply is fine. The issue is that computers are not designed for audio necessarily, so noise from the graphics card and hard drives can creep in. I'm addressing that too by putting some bypass caps on the lines to the hard drive, and changing the graphics card for one that is not part of the main board, since they are much better on terms of electrical noise.

Reading up on power supplies, ALL the power supplies I have are floating. Apparently it's very easy to tell - they have no Earth connected. Mine all have the figure 8 mains- no Earth pin. :-)

Sent from my XT1039 using Tapatalk

[The Magic Rabbit]

'I think you read the 4990 datasheet wrong.  It's (V-) - 0.3V to (V+) + 0.3V 
Where V- and V+ is your supply voltage. (+/-19V max) "

Well aren't I a Wally... yeah, I read the data sheet wrong, lol.



"No, you need a power supply that can supply at least 160mA.'

Ooooooh... what about when I start adding on other things, like LEDs driven from an LM3915? Do I just ignore the virtual ground and AC couple the signal? ... Almost treating it like two separate projects?

Would something like aTLE2426 be sufficient to provide virtual ground, even with 8 Op Amps / channels?

Sent from my XT1039 using Tapatalk

[RJFreeman]

The computer power supply is fine. The issue is that computers are not designed for audio necessarily,

Yep, that is exactly the point I am trying to make.
simply adding more decoupling capacitors may not be as successful as you might think, since the noise will be brief spikes which could have quite a bit of current driving them, this can result in you simply injecting more noise into your ground.
you may think that this noise is not significant but even fractions of a millivolt can be audible on a line level signal. 

The channel strip in a mixing console will usually have resistors in series with the power supply (10 ohms or so) and then capacitors from there to ground. This reduces the noise current significantly (since the noise voltage is fairly low, it doesn't take much resistance to reduce noise current to insignificant levels) - and this is with Linear regulators, which usually have noise voltages of less than 1mV, while even a perfectly working Switch-mode supply can have noise levels of 10mV or more.

To decouple a switch mode supply I would also consider inductors in series with the supply rail, and capacitors from there to Ground, since most of the noise is high frequencies this will help reduce the noise current.

[The Magic Rabbit]

OK - what kind of values should I be looking at for caps and coils?

Sent from my XT1039 using Tapatalk

[RJFreeman]

I think you read the 4990 datasheet wrong.  It's (V-) - 0.3V to (V+) + 0.3V 
Where V- and V+ is your supply voltage. (+/-19V max)

Maximum input is 12V (Common-Mode Input Voltage Range)

8 x Op Amps at maybe 20mA a piece meant I needed stable virtual ground at 160mA.

No, you need a power supply that can supply at least 160mA. 
Also.... 20mA per opamp is large. They're typically around 10-12mA or even less.

actually, the virtual ground for an op-amp circuit doesn't need to handle anywhere near that amount of current, since it is only really used by the feedback circuit.
For example, if you had a differential amplifier using 4 x 10K resistors as per the attachment, then the maximum current going to ground will be the input voltage at V2 divided by 20,000. so presuming 12V peak on the input, the peak current to ground will be 0.6mA

[RJFreeman]

OK - what kind of values should I be looking at for caps and coils?

Again, the variables are unknown,
what frequency is your supply running at?
what is the noise characteristics of the supply?
what are the current requirements of your circuit?
 generally the rule of thumb would be more inductance in series and as much capacitance as you can get, but then that can cause other issues.
what is the internal series inductance of your capacitors? (they might be useless at the switching frequency of the supply) does that matter? (high frequency noise might not be audible anyway)
If you use too much capacitance will your supply go into over current shut-down and or do something strange?
If you use too much inductance is that going to interfere with the high frequency response of your op-amps? or will your capacitors happily deal with all that?

[RJFreeman]

I think you read the 4990 datasheet wrong.  It's (V-) - 0.3V to (V+) + 0.3V 
Where V- and V+ is your supply voltage. (+/-19V max)

Maximum input is 12V (Common-Mode Input Voltage Range)

8 x Op Amps at maybe 20mA a piece meant I needed stable virtual ground at 160mA.

No, you need a power supply that can supply at least 160mA. 
Also.... 20mA per opamp is large. They're typically around 10-12mA or even less.

actually, the virtual ground for an op-amp circuit doesn't need to handle anywhere near that amount of current, since it is only really used by the feedback circuit.
For example, if you had a differential amplifier using 4 x 10K resistors as per the attachment, then the maximum current going to ground will be the input voltage at V2 divided by 20,000. so presuming 12V peak on the input, the peak current to ground will be 0.6mA

sorry forgot to allow for the input impedance of whatever equipment the op-amp is feeding, this will usually be greater than 10K (probably more like 50K or so) which would add maybe, worst case, another 1.2mA

[The Magic Rabbit]

Those sound like complete unknowns for virtually any off-the-shelf supply to me. Surely all I need to worry about withdrew be anything in the audio range, and SMPS will be over that, so why not just low pass at, say, 20Hz? ;-)

Sent from my XT1039 using Tapatalk

[fivefish]

actually, the virtual ground for an op-amp circuit doesn't need to handle anywhere near that amount of current, since it is only really used by the feedback circuit.

Is your comment for me or the OP?

Because what I said is No, he doesn't even need that BUF634 opamp for that purpose.  It's a waste of a part for that purpose.
If his load is XX mA, then he needs a power supply that can supply XX mA. (not an opamp that can supply XX mA).

[RJFreeman]

Is your comment for me or the OP?

that would have been for the OP since I was pointing out what the ground was used for.

[Hideki]

If you want to use the same supply for lots of LED drivers, they will draw a lot of current, and that current should absolutely not go into the virtual ground.

[The Magic Rabbit]

It's not a huge amount, but there's 8 channels x 10 LED x 10mA so not far off 1Amp once you allow for the support chips and margin of error.

Sent from my XT1039 using Tapatalk

[Hideki]

Not "huge", but that's why you can't connect them between the virtual ground and 24V, because then the 1A would have to pass through the BUF634 chip.