transistor summer

There must be 50 ways to do this. (Make a little plan, Stan, no need to be coy, Roy....)

I'm assuming you're talking about a low-level or line-level audio application, right? In that case you care a lot about linearity and not as much about isolation between channels, i.e. it isn't a big problem if a little bit of the other channels's signals come back out the input. (The idea of 'input' and 'output' terminals is a convenient approximation--strictly speaking you can't completely prevent some stuff coming back out of the inputs.)

Generally if you have an active device per channel, it will improve the between-channel isolation. If it's a simple BJT or MOSFET stage, it'll probably hurt the linearity.

Cheers

Phil Hobbs

A summing junction with an op amp can be relied on to be at or near pseudo-ground; with a lower voltage gain item, like a transistor or MOSFET with simple biasing, there's just a bit more voltage swing at the summing junction.

It's all easily modeled, no need for experimentation if you know your gain element's parameters.

A transistor can't swing output far below the base voltage, of course, THAT can matter.

For summing AC signals, common-base isn't bad. The summing node impedance can be a few ohms, wideband, and the input resistors can be many K.

The voltage at the summing point of an opamp isn't zero, either.

John

True, but it's a bit more symmetrical than the common base approach, so there isn't so much second harmonic distortion.

I have a design in layout right now that uses NPN and PNP capacitance multipliers wired in parallel, to get rid of the ripple on a bipolar thermoelectric cooler supply. That's one way of reducing the second harmonic!

Cheers

Phil Hobbs

Don't you have crossover disto. & switch noise problems that way? Or is that then out of band?

-- Les Cargill

How BIG is it John ?:-) ...Jim Thompson

It's controlling a thermoelectric cooler, so it doesn't switch between positive and negative very often!

Cheers

Phil Hobbs

Why the hell would you want to?

What, have you un-killfiled me?

You can't even do that right.

John

It's imaginary.

John

If you're talking about a single stage summer, the main differences are, of the bjt versus the opamp:

• less isolation between inputs

• less open loop gain

• more distortion

• less PSRR

Moving to a fet gets you high input impedance, but all the above issues still apply. In practice one can make linearity good enough with low level audio signals by using a source degenerating R to give bias and local nfb.

NT