Direct Injection Circuit Diagram

[exabrial]

Hey guys, I'm trying to design my own buffered direct injection box that has a buffered unbalanced send and of course a buffered balanced send. I've pieced together various circuits from around the web and came up with this. Most if it was stolen from here: http://electronics.stackexchange.com/questions/167025/op-amp-based-guitar-direct-box


Anyway, can you guys see anything terrible wrong with this? Thanks!



https://123d.circuits.io/circuits/1705298-opamp-buffered-balanced-direct-injection-for-audio

[Transmogrifox]

Looks like it will work well.

Put something between 100R and 1k in series with the output of each op amp to ensure the op amps stay stable when connected to long lines or capacitive inputs.   Probably better to stick toward the 100R side of things to minimize loss, particularly at high frequencies with long cables.  It's a 2-edged sword.

[exabrial]

Thanks. Is this correct?

I copied this form the last circuit, but what's the purpose of c9 and r5? Are they needed?

Thanks!

[dschwartz]

A couple of things i learned about DI..

Use a ne5532 opamp or one that can cope with 600 ohms loads..
The two outputs shouldn't be connected directly, if the xlr is connected to a 600ohm load, it will load down the other output..ideally, use a buffer to separate them or put a series resistor (10k or more) to minimize the outputs loading each other..

Typical DI boxes have a 6dB boost to get the signal into "line" level..put a 10 k  resistor to vref on the - input of ic 1 a

[exabrial]

Ok thanks, I'll work on those things. I will buffer the unbalanced output separately. Let me work on that and I'll repost.


Any feedback on c9 and r5? Are they necessary? Seems like they're trying to couple two parts of the circuit but I'm not sure why the isolation would be needed.

[dschwartz]

They are not isolatong, on the contrary they feed the non inverted signal to the inverting opamp..im not sure if c9 is necessary though..the signal is at vref potential anyway..

[Transmogrifox]

Any feedback on c9 and r5? Are they necessary? Seems like they're trying to couple two parts of the circuit but I'm not sure why the isolation would be needed.

R5 is definitely necessary.  It sets the -1.0 gain for the differential output.  C9 hypothetically prevents DC offset but there probably isn't enough to be worried about.  You could probably do away with C9 without any ill effects.

I wouldn't worry about driving 600 ohms.  Almost any mixer has a switch between "mic" or "line", (low-Z or hi-Z).  Line inputs are nominally 10k.  I really like looking at RANE discontinued product schematics for this kind of thing.  One good example of what you are doing can be found on schematic page for for the MLM42 mixer:
http://www.rane.com/pdf/old/mlm42sch.pdf 

It looks an awful lot like the circuit you've presented so I think you have something good.  I agree with dschwartz that something like an NE5532 (or as Rane used MC33078) is a good choice for this application.

Generally it would be considered best practice to drive the unbalanced output from a separate buffer. You will see in the RANE schematic they did buffer the unbalanced outputs separate from the balanced driver. 

However, I don't see any real issue with driving the unbalanced output off the same op amp as you have it drawn.  The advantage of the way you have drawn your circuit is that it requires only one dual op amp.  If you add another buffer then you add another IC, which seems like a waste.

The NE5532 states 0.3 ohm output impedance tested into 600 ohm load.  The signal loss will be minuscule, so that's why I don't think it will actually be an issue.  I think the NE5532 could easily drive a 300 ohm load at 4V peak-peak (2V amplitude), which is equal to about 8 dBu if I did my math right.  Either way you won't have enough headroom on a 9V circuit to get much more than that, so I say don't worry about adding another buffer.

I say it's fine just the way it's drawn.  You can probably eliminate C9 without issue, and you can get rid of R3.  A resistor placed in a unity buffer like R3 is going to be selected purposely for a specific op amp.  Its presence is often related to noise performance and/or stability, and sometimes it's only for managing DC offset due to input bias currents.  You're better off just using a wire than randomly selecting a value -- more likely to cause problems than fix any unless you know why it's there and have selected the value for a specific op amp.

[exabrial]

Thanks! wow, appreciate everything. I'm definitely learning a ton as I hang out here.

If I get rid of r3 should I also get rid of r6 then?

[dschwartz]

Nopes, leave r6, it sets the gain of the inverted signal together with r5..

I recommended the buffered output because i made one exactly as the schematic shown..and when the xlr was connected, there was a huge volume drop on the other output..i corrected it with a 10k resistor which isolated both outputs (since tje other output was intended for hi z amp inputs..

[Transmogrifox]

n..and when the xlr was connected, there was a huge volume drop on the other output..i corrected it with a 10k resistor which isolated both outputs (since tje other output was intended for hi z amp inputs..

Good to know that.  I can talk theory all day, but I can't argue with the real thing.  Listen to the guy who did it, not the guy who tells you how he thinks it will work

It leaves me puzzled though.  Maybe I should order some NE5532's and figure out what is happening.

[dschwartz]

n..and when the xlr was connected, there was a huge volume drop on the other output..i corrected it with a 10k resistor which isolated both outputs (since tje other output was intended for hi z amp inputs..

Good to know that.  I can talk theory all day, but I can't argue with the real thing.  Listen to the guy who did it, not the guy who tells you how he thinks it will work

It leaves me puzzled though.  Maybe I should order some NE5532's and figure out what is happening.

It was with a tl074 so maybevthat was the culprit..i changed to ne5532 and a buffered extra output after that..

[PRR]

> if the xlr is connected to a 600ohm load

Where do you find true 600 Ohm loads today?

A '5532 will drive 600 fine. A TL072 is wimpy in 600. If the "balanced" output is really shorted to ground one side, the opamp sees 100 ohms, and the TL072 will just about quit; the 5532 will chug some signal into 100r.

> it will load down the other output..ideally, use a buffer to separate them or put a series resistor (10k or more)

If you only expect a single shorted/loaded output, 2K isolation is fine for TL072; 5532 will easily drive 470r isolation resistor and get full signal to the un-shorted output.

I used to provide for sixteen destinations, many not under my control. And live performance, no do-over. The usual rule is to assume that up-to 1/3rd of your outputs may get shorted. (To be very conservative, all-but-one output.) I used a rather beefy line-amp, with 470r to "my" loads and 2.2K to "strange" loads. 2.2K driving 100 feet of cable will be down a hair at the top of the audio band (-3dB at 25KHz) which was acceptable here. 1/3rd of 10 stranger feeds shorted is say 2.2K/4= 550 Ohms, with the "good" 10K loads is 500 Ohms, and my line-amp would drive well under 600 r with ease.

Other possibilities are independent buffers for each output (costly in bulk) or one very powerful amp with resistors to many outputs. Here the better loudspeaker chips apply. If you can drive 8 Ohms, you can drive TWENTY 150 ohm resistors to shorts, and still have full signal to any un-shorted loads. The car-sound "16 Watt" chips give a push-pull (balanced) output at 4 Ohm min, and you usually run out of space for connectors before you can connect enough 150r-isolated shorts to hurt.

In less-antique broadcast practice, "nominal 600" lines were driven with 60 Ohm isolation resistors. But broadcast lines sometimes run FAR longer than anything musicians face. (We have to work close enough to be heard and get paid.)

[exabrial]

OpAmps are cheap. I'll drive each output with a separate buffer.

New circuit coming later... standby!

[exabrial]

Alright guys,

Changes:

Eliminated first buffer
Eliminated feedback resistors on u+ b+
Eliminated Coupling capacitor for b- buffer
Added buffer to u+


Questions:
Do I need to buffer my buffers or is this ok?
Did I terminate the unused opamp correctly?
Anything else stupid?

https://123d.circuits.io/circuits/1705298-opamp-buffered-balanced-direct-injection-for-audio

[PRR]

> Anything else

This plan is 10K (11K) input impedance.

The Ji source has to drive R5. The other end of R5 is a Virtual Ground, typically less than 10 Ohm impedance due to IC1_b NFB action.

Since you have IC2_b standing out there scratching its nose, move that over to buffer Ji source and drive all the other buffers.

Alternatively, move R5 to say IC1_a output (same signal, now buffed). And go ahead and wire IC2_b as an extra unbalanced output.... it may come in handy some day.