Supply voltage dropping networks

[Ears]

I'm designed a supply using the usual series ladder of dropping resistors. In addition I thought I'd try O'Connor's lead of using a filter cap for each triode in preamp rather than one for each dual unit.

It works fine, but in order to keep the keep the supply voltages up each RC filter in the chain had to only drop a few volts and the resistor was only about a K in size. I assume that larger resistors and voltage drops would provide more filtering so I've a query about parallel configured supplies. See picture. Most designs use the first scheme (1) but I've seen the mixture of series parallel as in (3) occasionally. I can't recall any (?) using parallel (2) exclusively but maybe they are in use too. In a parallel configuration I figure I could use larger value resistors for more ripple/coupling attenuation.

Duncan Munro's simulator only allows the series option and I'm left wondering if there is an obvious reason why.

Can anyone kindly expain why the parallel option isn't used as much as the series configuration?

[mhuss]

The "series" connection has the advantage that each succeeding stage benefits from the filtering of the previous stage. In the series/parallel connection, the parallel R/C sections are each starting with the more ripple-laden screen supply, potentially injecting more hum into the input stage.

--mark

[skyboltone]

I'm wondering why he's made no provision for bleeder resisors in the first stage.

[Ears]

I'm wondering why he's made no provision for bleeder resisors in the first stage.

Well, it's only a general schematic, sufficient to the illustrate the question

And thanks also to Mark, I will think on that. I thought that the reason you gave is true in the case in terms of residual power supply ripple, but I suspect parallel supply would be better to eliminate supply coupling or signal modulation effects between the stages, But I'm unsure of my ground.

Edit:

I'm thinking in terms of signal modulation between adjacent stages. Say for sake of illustration, in diagram (1) point (D) requires 3 x 1k resistors beween it and point (A) and all caps are 20uF. Then the RC filter between adjacent take-off points (D) and (C) consist of 20uF and 1K ohm.
If we are in diagram (2) then the RC filter between adjacent take-off points (A) and (D) would consist of 20uF and approx 3Kohm, the higher resistance meaning greater isolation between stages.

If the "parallel" config is of little or no merit over "series" then why is it sometimes used as in Mesa's dual rectifier?

[tele_player]

It's my understanding that the multi-stage filter setup we're used to seeing is more about stage decoupling than about ripple reduction. In other words, you'd still do it if you were running off batteries.

I just checked an old textbook from the 60's, it offered a few paragraphs on decoupling. A recent textbook (1999) covered cascaded stages, but made no mention of decoupling.

Tubes are almost a dead art, but analog is dying, too.

An interesting bit from the old book was the observation that some coupling between adjacent stages is OK. Since they're typically out of phase, it amounts to a small amount of negative feedback, which stabilizes the circuit at the expense of some gain. The troublesome coupling would be, for instance, between the 1st and 3rd stage, which would be positive feedback (they're in phase), and could cause oscillation.

I'd step back and think "What am I trying to solve?" before investing too much on this line of thinking. The normal setup we're used to seeing provides the lowest ripple in the supply to the highest gain stages, which is probably more important than further reducing a harmless small amount of coupling.

[Ears]

Tubes are almost a dead art, ...

No! we, the brave 600, will will carry the flame!

An interesting bit from the old book was the observation that some coupling between adjacent stages is OK. Since they're typically out of phase, it amounts to a small amount of negative feedback, which stabilizes the circuit at the expense of some gain. The troublesome coupling would be, for instance, between the 1st and 3rd stage, which would be positive feedback (they're in phase), and could cause oscillation.

That rings a bell in my memory.

I'd step back and think "What am I trying to solve?" before investing too much on this line of thinking. The normal setup we're used to seeing provides the lowest ripple in the supply to the highest gain stages, which is probably more important than further reducing a harmless small amount of coupling.

When I fired up this build I forgot one connection (to the supply of the driver's plate). The SE amp still worked albeit at I guess less than half a watt (into JBL E120) - well it was almost loud enough to hear over conversation when cranked. Great overdriven sound, colorfully distorted, with no grid blocking... it rocked. The only obvious way I figure the signal could have got through to the output tube was through power supply coupling, (maybe I'm missing some other capacitative means) but it got me thinking about this issue and if such signals are finding all their way through the amp in any case.

I'm still wondering why Mesa and others use the "parallel" option, maybe the current through the resistors is lower and they don't have to be so highly rated (wow, that would save them a fortune).

[LOUDthud]

One of O'Connor's motovations is to carefully star ground each stage with it's filter cap. He locates the filter caps throughout the preamp.

On the plus side, if you want to change the B+ on one stage, there isn't so much interaction with other stages.