Best way to split a signal?

[DSV]

I'd like to build a multi-cannel project, which requires the splitting of the signal at the beginning of the chain, while providing some gain at the same time. At the end, one of the channels will be selected through an electronic switch.

There are 2 possible solutions I have in mind for the switching:

1) The input signal goes to 1 non-inverting opamp per channel:

2) The input signal is splitted after a non-inverting opamp:


Is one of these alternatives the best, or should I consider a third implementation?

Thanks,

DSV

[brett]

As far as I'm aware, use Option 1. It isolates each channel from loading on the others.

[mth5044]

Slightly OT, but to go with splitting, how would one join two signals? Would it be in a similar fashon, or could you just put a resistor after each signal then put the ends of the resistors together? How do you determine the resistor value?

Sorry for the slight highjack 

[aziltz]

Slightly OT, but to go with splitting, how would one join two signals? Would it be in a similar fashon, or could you just put a resistor after each signal then put the ends of the resistors together? How do you determine the resistor value?

Sorry for the slight highjack 

look up summing amplifier (op-amp)

[mth5044]

Thanks 

[brett]

Hi
I forgot to mention that you can also use inverting op amps.  Saves a few resistors.  Something like 220k or 330k input and feedback resistors are good.  A few people say to use higher value resistors for higher input impedance, but I believe that its no use trying to reproduce frequencies only dogs can hear.  Also, noise (hissss) goes up as resistance goes up.  (My choice is always 220k).
have a great day

[R.G.]

...Something like 220k or 330k input and feedback resistors are good.  A few people say to use higher value resistors for higher input impedance, but I believe that its no use trying to reproduce frequencies only dogs can hear. 

It's not really a few people; 1M ohm is the most widely accepted value. Part of this is from history from tube amps, but a lot of it is from the math of the situation.

The impedance of a guitar pickup varies from about 1H to maybe 4H depending on single coil/humbucker. There's some distributed capacitance there as well, but let's ignore it for the moment. What's the cutoff frequency for a 2H (just taking a median kind of value) inductive signal source into 220K?

Xl = 2* pi*F*L, and the cutoff is where Xl = R, or where F = R/(2*pi*L) = 220K/6.28*2 = 17516Hz, well within the 20kHz of human hearing.
One could argue that is high enough, as I guess you are; however, there is a group who insists that guitars get sparklier and livelier if you use input impedances over 5M. Then there is the "curly cord" brigade who likes the "brown sound" of a 20 foot curly cord eating their treble.

It is also feasible to do a high input impedance inverting output without high value feedback resistors by using T-attenuators as feedback and sidestepping the problems with large resistance feedback networks too.

[brett]

Hi

as always RG, you have thought this through.

But as for -

well within the 20kHz of human hearing

it is only young people who can hear to 20kHz.  Generally, you lose about 2kHz per decade after age 10.  Next year when I'm fifty that rule-of-thumb suggests that I will only be able to hear to 12 kHz.  Even though I've listened to some loud rock and roll, my hearing is, indeed good to about 12 kHz.  It was 20 or more years ago that I stopped hearing the 16 kHz "whistle" from the TV set (The HDF of NTSC sets is 15.75 kHz).  If you are saying "What 16 kHz whistle?" then you know that it's been some time since your hearing  extended to 16 kHz.

But one strange thing I've never been able to work out is why do I like filters that roll off treble with Fc around 7 kHz if my hearing is only good to 12 kHz?  In theory, such filters would only take out a small amount of treble (ie the 7 to 12 kHz range, or less than one octave).  Can anyone explain that?

cheers

[DSV]

Thanks for the answers. I'll go for solution 1 then (I don't want an inversion of the signal). It seems an interesting discussion has also started  I'll report my experience then.

On my current project, the input buffer has no resistor for referencing voltage to ground. Noise is increased, but I had the impression that a resistor (1Meg) cut some harmonics in the note attack (nothing noticeable in a mix, but a different feeling when recording single tracks).

About the 7kHz, I always noticed that by boosting those frequencies the sound became really "fizzy". On the other hand, boosting frequencies up to 6kHz was fine, as well as those > 9kHz. If I remember correctly a thread at the other forum, there was also some kind of effect supposed to cut off frequencies at 7-8kHz (a deep notch, -20 / 40dB). Some forumites used it with digital modelers, and they reported that the sound was significantly improved.

[MohiZ]

Slightly OT, but to go with splitting, how would one join two signals? Would it be in a similar fashon, or could you just put a resistor after each signal then put the ends of the resistors together? How do you determine the resistor value?

I've seen some schematics that just connect the signals together with resistors. It's probably better practice to use the summing amplifier, as suggested, though. I guess the resistor values depend on how large your two signals are, and how much of each you want to be heard in the mix.

[Andi]

I'd opt for version 2 - much higher input impedance. Though I'd change to using an op-amp follower instead of a unity gain non-inverting configuration - simpler and fewer components.