LM334 questions, and it's role in this schematic.

[user 41968]

It doesn't make a difference. The Shockley equation is valid for all sorts of PN junctions.

Of course it is not the highest of the two. I used the term "analog-OR" as a simplification.
And I stand corrected regarding the power summing.

My point was that the timing capacitor node is not an internal emitter follower output providing a stiff voltage source.

It is better analyzed as a current source through a diode which just happens to have an emitter as cathode.

The current domain is the thing.

When two channel's detectors are coupled at the Ct pin the currents sum into a single capacitor.

It appears to be a diode OR - and structurally is in the voltage domain - but what is happening is current summation through diodes.

JR is correct in that the output of the "RMS" detector is really not true RMS. I think this is where we began to disagree some time ago.

The final square root and antilog mathematical operations to make it be RMS-responding occur in the VCA control scale and law when the RMS detector is paired with a VCA.

The square root is due to the scale factor of the VCA being twice that of the "RMS" detector. (√ is /2 in log domain)

The antilog operation occurs due to the VCA's exponential control law.

If one wanted a linear DC representation of true RMS for measurement the way to get it would be to use a feed-forward topology and (1) put a reference DC voltage into the VCA's input pin, (2) control its' Ec with a "RMS" detector and (3) read the final VCA-IV converter output.

As to the merits of true RMS versus averaging I recently measured some UPS outputs. On the APC the averaging responding meter read about 93V but the true RMS read 120V. The Belkin read about 85V avg and 110V true RMS. Not sure how this relates to music other than both are complex waveforms.

 

[user 41968]

I also want to pint out something about the LM334 current source used to provide reference voltages.

IIRC this circuit was first suggested by THAT in a battery-powered application. This is Gary Hebert's invention.

Part of its brilliance is hidden. When a Zener or voltage reference like a TL431 is used additional current is required to bias the shunt reference.

The LM334 reference does not have this additional current demand: The reference current is the bias current. Zero current overhead.

I used an LM334 to drive a resistor string to provide reference voltages for a simple LED bar graph. The regulated current is far superior than a Zener with varying supply voltage and it didn't have the current overhead of the TL431.