active tri line splitter using BJT transistor

[hozone]

hello,

i'm building an active line splitter based on BJT transistor
i need to split the line output signal of my tascam us122.
( specs of the card:
Nominal output level : –10 dB
Maximum output levels : +6 dBV )
i will use this splitter to send signal to
* speaker amp 1
* speaker amp 2
* headphone amp
for my audio recording studio


- question 1 -
how can i transform the output db/dbV value to voltage?

i've already build a line splitter, based upon
* AMZ - Guitar Effects & JFET Splitter
* Basic Buffers
schematics attached (VCC is +9v)

i've checked this with a scope and a sine signal sweep (0.3V - from 10Hz to 25Kz), it seams to me that the FFT is 1:1 to the original sine signal.

- question 2 -
is this FFT check right?, i mean, is it right to check the frequency response by looking at the generate FFT during a sine sweep?

- question 3 -
i've also check that the output sine is the same amplitude that the input one, for the sweep, and it seams so. is it a good test?

- question 4 -
there are errors in my schematics?

- question 5 -
there is any opamp good schematics for this?


thank you...

[Mooly]

Your circuit is fine although unusual.
R1 could be usefully lower in value.
C1 could be larger depending on the LF response you want.
C2,3 and 4 could likewise be a little larger if you want extended LF response.

Thats all a matter of personal choice.

Using 3 followers is unusual. Remember the bias network has to supply 3 lots of Ib so you want high gain transistors to minimise loading.

It's all very workable but unusual.

Do you need 3 followers ? One opamp output would easily drive 3 outputs, or you could use a quad opamp and use one as a buffer and have the other three as followers.
Even a soundcard output should drive 3 inputs OK.

If you only have a single rail supply avalable the transistor option is absolutely fine.

Even 1 follower would drive 3 inputs.
You could also replace the emitter load resistors with constant current sources.

A follower like this will be good up to many hundreds of Khz when loaded lightly.

Sine wave test is fine, it will also be essentially perfect on squarewave testing too.

For db's and db/v you will have to do some reading and maths...
this covers what you are asking but you need to know impedances and levels,
Decibel - Wikipedia, the free encyclopedia

[hozone]

Originally Posted by Mooly

Your circuit is fine although unusual.
R1 could be usefully lower in value.
C1 could be larger depending on the LF response you want.
C2,3 and 4 could likewise be a little larger if you want extended LF response.

Thats all a matter of personal choice.

Using 3 followers is unusual. Remember the bias network has to supply 3 lots of Ib so you want high gain transistors to minimise loading.

It's all very workable but unusual.

Do you need 3 followers ? One opamp output would easily drive 3 outputs, or you could use a quad opamp and use one as a buffer and have the other three as followers.
Even a soundcard output should drive 3 inputs OK.

If you only have a single rail supply avalable the transistor option is absolutely fine.

Even 1 follower would drive 3 inputs.
You could also replace the emitter load resistors with constant current sources.

A follower like this will be good up to many hundreds of Khz when loaded lightly.

Sine wave test is fine, it will also be essentially perfect on squarewave testing too.

For db's and db/v you will have to do some reading and maths...
this covers what you are asking but you need to know impedances and levels,
Decibel - Wikipedia, the free encyclopedia

thanks,

R1 => 1M, ok?
C1,2,3,4 => i would like an almost perfect freq response, even on LF, from 50Hz to 25Khz, better C1 => 0.3uF - C2,3,4 => 10uF?

dbu, tks i've found the solution with some math.

i can use dual supply, or make a virtual ground circuit, so i can also use an opamp for this project. do you know any good schematics?

[Mooly]

If you want an opamp it's text book stuff. You must use "unity gain stable" opamps such as OPA2134 (which are duals so you need two IC's). FET opamps also produce zero DC offset.

If you're stuck I can draw a simple schematic later but it's all text book stuff. All stages are connected as followers with the three (if that's what you really want) separate driver stages all fed directly from the first stage. Good practice to isolate each final output via a small resistor of say 68 ohms to ensure total stability with long or capacitive leads.

[sreten]

Hi,

Why use circuitry at all ? -10dB is 300mV - typical home audio line level.

Most line outputs can drive multiple line inputs with no difficulty.

rgds, sreten.

[hozone]

this is interesting:
How Do I Make a Distribution Amplifier?

expecially this link:
HeadWize - Project: Designing an Opamp Headphone Amplifier (A HeadWize Design Series Paper)

maybe i can use this as a buffer opamp.. ?

[hozone]

Originally Posted by sreten

Hi,

Why use circuitry at all ? -10dB is 300mV - typical home audio line level.

Most line outputs can drive multiple line inputs with no difficulty.

rgds, sreten.

1) to learn something new - for me, cause i'm a software eng. -
2) to split the source output signal without losing power 1:1~ signal

[Mooly]

Originally Posted by hozone

1) to learn something new - for me, cause i'm a software eng. -

I can understand that.

Try something like this. Also with the addition of two resistors in the feedback network the gain of any buffer can be increased.

Haven't shown supply decoupling but it's good practice to add a small electroylitic across the opamp supply pins.

The TL072 is a good choice for a first project as it's stable and draws little current. Very good performance too.

[benb]

Originally Posted by hozone

1) to learn something new - for me, cause i'm a software eng. -
2) to split the source output signal without losing power 1:1~ signal

Both good points, but regards to 2) what is the output impedance of the thing you're going to plug into this? What are the input impedances of the things you're going to drive?

An offhand guess at the answers are 1k, 100k, 100k, and 100k. With those numbers, plugging the output directly into one input causes a 1 percent (much less than 1dB) signal loss. Having it drive all three inputs would cause a 3 percent drop, still MUCH less than 1dB. Furthermore there's no active electronics added for the signal to go through.

There still could be other reasons for using a buffer like this. If you have very long cables (maybe 20 feet or more) where the capacitive loading of the cable could reduce the high frequencies, such a thing would help (put this thing at the source and put the long cables at the output) . Do that opanp circuit Mooly just posted, and be sure to include the buildout resistors.

[hozone]

Originally Posted by Mooly

I can understand that.

Try something like this. Also with the addition of two resistors in the feedback network the gain of any buffer can be increased.

Haven't shown supply decoupling but it's good practice to add a small electroylitic across the opamp supply pins.

The TL072 is a good choice for a first project as it's stable and draws little current. Very good performance too.

thank you,

if i've understaind
* the 0.47uF cap is to remove dc offset
* the 470K res set the input impedance (the same way the R1 res do in my schematics)
* opamps are in voltage follower configuration
* 68ohm output resistor quite some noise
right?