Paraphase inverter values

[EFK]

Below is a little drawing of a Supro 6497 paraphase. The values are pretty typical of what Valco was using as 'standard' by the mid 60s, although w/ different amps usually you see variations in the coupling cap values and the bypass cap around the inverting cathode R. However, the resistor values were pretty much settled. What I'm wondering is, WHY is the second stage grid leak resistor always larger at 470K? That's a pretty hefty difference from combined 282K off the first stage (270K + 12K). I thought about a balance issue, but these resistor values were always the same despite different coupling cap values and variations in the cathode bypass (some were 25 or 35 uf, some like on the 1624T were .05), which ought to affect the balance in such a way that a 'one size fits all' 470K would not do much to maintain a balance. And they used this in 6L6 amps, 6V6 amps and 6793 amps generically.



Anyone have any thoughts?

[jmaf]

The first stage has way more gain than the second.

My guess is that the second stage in a paraphase inverter needs less attenuation(a voltage divider formed by the output impedance of the plate and the resistor in question). That value must have been found through experimentation, a signal is injected and the value was increased until the outputs were symmetrical.

[Chuck H]

And it's probably true that it isn't always balanced. Generic values for arbitrary components are used all the time in design and manufacture. I suppose you could put it on a scope and make some adjustment to balance the outputs. That doesn't mean it would sound better. There are differeing opinions, but some imbalance in the PI has both pros and cons depending on what kind of tone you want to achieve.

It could be I'm missing something, and there is some formula that dictates the ratio of that voltage divider, but it seems to me that if you vary other things like Vp and bias conditions the ratio would need to change also. But it doesn't have to for the amp to work and sound fine, so...

[Enzo]

Looking at the schematic I see the left triode putting out a signal on pin 1 for one power tube, then the second triode needs to put out a similar level of signal at pin 6. Since we are using the pin 1 signal as a source for that, how much does the pin 1 signal have to be reduced for the grid at pin 7? The 270k and 12k form a 20/1 voltage divider to accomplish that.

The ratio of signal level is what makes for "balance." Cap size and bypass affect tonal response or bandwidth, but that is different from balancing a PI.