Additional filter caps....

[Chuck H]

I'm hardly offended. I just remeber times when I've done a few 12oz. curls and spoke up. A couple of times causing enough trouble that in retrospect I should have just let it go. I'm trying (key word) to keep my etiquette at the same level I would use in public. As in "Gee, if I actually say that it could cause a fight, maybe I'll let it go." It's real easy for me (and others) to forget manners with such anonymity as an internet forum.

Of course I still fire back unreserved when shot at. That's one reason Gary's post was a dissapointment. He didn't use a weapon.

The whole R.G., Gary, Merlin thing is still entertaining enough that I always read on when it happens. But I kinda worry sometimes when I consider how angry Gary has become and that this may be the last amp forum that still allows him to post. If we pick on him it might be partly our fault when he gets drunk and sticks a gun in his mouth.

[trobbins]

I strongly doubt that you would notice HT induced preamp hum, or any form of common HT interaction between V1 and V3 if you are using a new electrolytic for the 20uF filter cap. I would rather suggest that you focus on getting your distributed star grounding scheme in good order, and set up an elevated positive DC bias with a humdinger pot for your heaters (circa 40-50V should get you well above the output stage bias - not that hum would be a contributor from that stage). Most people I think would then end up with a 'humless' amp.

If noise is a concern, then probably worthwhile targetting metal film for all resistors around the input stages at least.

Ciao, Tim

[trem]

First......I am sorry to see this happen again in one of my posts. May be do to the beginner level of questions I ask.....?? I do not have a dog in this fight, and I know tension can run high on these forums. I respect all of you, and appreciate your help, more than you know.
trobbins....I will have to post some pictures and a long list of "things" I have tried to get this amp more quiet than it is now.
Where octals pres are concerned......I understand how a "hum-bucker" pick up works, and I understand the theory of running heater wires in a close and consistent proximity. What I Do Not understand is what it means to wire a tube for "hum-bucking".
Thank You

[trobbins]

Humdinger - not humbucker ;-) (although maybe it's called a humbucker in other places !?)

I'm sure there are many excellent references for getting an understanding of humdingers. Merlins is a great place to start (The Valve Wizard), and he also has a great page on grounding to check how well you've done your layout.

Ciao, Tim

[Chuck H]

If this is specifically related to filament circuit hum I think your idea is sound. The more typical 12ax7 has a tapped filament so each side can be wired reverse phase. This is how the typical twisted pair filament circuit works. The 6sl7 has a single, untapped filament heating both cathodes so they cannot be wired the same way. But, and I think this is where your going with this, it may be possible to wire two complete dual triodes out of phase to achieve some cancellation. I should think this would be somewhat dependant on both dual triodes being in the same AC path though. Still, I think some improvement could be made. You can also try elevating the 0V reference for the filament supply with a DC bias. I've never used an elevated DC reference with a straight wired filament supply but I've read that some Champ owners have used it to reduce hum.

[J M Fahey]

Going back to the original doubt(s), expressed as:

But what I am wondering is, if all I am trying to do is isolate the plates from each other, so they have their own filter cap....What resistor do you go with.? Is there a minimum value I need to use.?
If I do not need a voltage drop of any particular value (I do not need the voltage to change at all) do I just pick a resistor and make sure it has a wattage rating that will handle the ohms law formula.?

Nice wish, but there's no free lunch. For any effective filtering you will lose *some* voltage.
It probably will not mind, anyway.
The wattage problem is secondary. What you need to calculate first is the filtering action, the original reason for adding a filter !!!
I like to consider these as low pass filters, with a given (expected) attenuation at 100 or 120Hz, the annoying ripple frequencies.
My "almost_no_math" rule of thumb is to always remember that a 100K resistor grounded with a .1uF capacitor has a crossover frequency of 16Hz, and will attenuate higher frequencies at 6dB/octave (easy)
From that on, I calculate (mentally) with very simple math what I need.
100K+ .015uF= 100Hz so 1K+1.5uF=100Hz also .
If I increase those 1.5uF 10X to 15uF we are attenuating those 100Hz by an extra 10X, which means -20dB.
So now I know that 1K in series with 15uF will attenuate whatever 100Hz hum I have there by 20dB, which is very good.
You have 20uF? even better !!, you are still in the ballpark, you'll get an extra couple dB.
How much voltage will you lose?
If you have a single triode with a typical 100K load resistor it will pass around 1mA.
How much will you lose?: 1mAx1K=1V .
Can you spare 1V?
I thought so.
And if I have, say, 4 triodes?
You'll lose 4x as much, or 4V.
You can get somewhat more exact values with pencil and paper plus a calculator, of course.
I just love to exercise my mind when waiting for the bus or at any queue.

[diagrammatiks]

If you want to decouple the 2 nodes just parallel them from the previous node.

if it was resistor/cap/v1a/b before just make it

resistor/cap/v1b and resistor/cap/v1a off the output from the previous cap.

if it's the first 2 nodes and everything else is in series you hardly need any filtering there anyway since the effect is cumulative from the reservoir.
double the resistor value.

you'll have roughly the same amount of filtering.

It's a lot more complicated if you wanted to decouple every node. If you just want to do 2 then you can use the parallel resistors instead of theory.

[TheTinMan]

Tradeoffs

To simplify the points JM made, if you cut the size of the dropping resistor on the power rail in half, then you need to double the size of the next capacitor to get the same filtering. So theoretically if you have a 10K resistor between two 20uf caps now, you could make it a 1K resistor and use a 200uf cap with the same filtering. (& yes, I know that 200uf caps would be too big and too expensive)

However, I think that the resistors on the power rail serve a second function as well (the first function being part of a low pass filter). They help isolate each node on the power rail from its neighbor(s).

So to re-state the original question in this thread, what is a reasonable minimum resistance between preamp nodes on the power rail to avoid crosstalk/interference/whatever? Is there a rule of thumb like "5K" or do you need to take the plate resistors into consideration as well?

Also, would the branching of the power rail that Enzo & RG suggest improve the isolation between the preamp nodes because there would be, in effect, two resistors separating (isolating) the last two nodes on the power rail?

I often see 1K or 1.5K between the OT node and the screen grid node, but larger resistors between those high current nodes and the PI and/or preamp. Very often there's one relatively big resistor between the screen grid node and the PI node which handles all of the preamp current and, therefore, is sized to get specific preamp voltages. Fender seemed to stick with 4.7K or higher (to make a gross generalization), and Leo tended to use the smallest capacitors possible. Classic Marshalls tend to employ 10K resistors on the preamp portion of the power rail with much larger capacitors.

[R.G.]

To simplify the points JM made, if you cut the size of the dropping resistor on the power rail in half, then you need to double the size of the next capacitor to get the same filtering. So theoretically if you have a 10K resistor between two 20uf caps now, you could make it a 1K resistor and use a 200uf cap with the same filtering. (& yes, I know that 200uf caps would be too big and too expensive)

This is true if filtering is the issue being resolved - as you realize from your next comments. Filtering becomes very much a secondary issue after the third section of R-C filtering from the first filter cap. The first filter cap typically reduces the ripple on the DC to under 5% of the DC level. After that, each R-C section reduces 120Hz ripple by the RC filter factor. A 1K/20uF cap reduces 120Hz by a factor of 16 just on the magnitude of the capacitive impedance to resistive. The impedance of the cap at 120Hz is about 66 ohms, so the filter filter cap ripple is reduced by this by 24db. If the resistor had been 10k, the reduction in ripple would be by 44db. That's for one additional section. After about two sections, the ripple contribution is really, really small.

However, I think that the resistors on the power rail serve a second function as well (the first function being part of a low pass filter). They help isolate each node on the power rail from its neighbor(s).

And you're right - this is the big one.

So to re-state the original question in this thread, what is a reasonable minimum resistance between preamp nodes on the power rail to avoid crosstalk/interference/whatever? Is there a rule of thumb like "5K" or do you need to take the plate resistors into consideration as well?

You can't solve it that way. It's the R-C filtering that introduces the isolation, reducing anything that one stage puts on the power line. The reduction is not exactly, but simplistically, the ratio of the capacitor to ground impedance and the series resistor acting as a voltage divider at the lowest frequency of interest. It gets much, much better as frequency goes up because the capacitive impedance heads for zero.

So 1K/20uF, 10K/2uF, 100K, 0.2uF all have the same isolating effect, they just have steadily increasing DC impedances for the DC that goes down the power line. They all have the SAME reduction in AC interaction, but steadily increasing absolute impedance level.

Also, would the branching of the power rail that Enzo & RG suggest improve the isolation between the preamp nodes because there would be, in effect, two resistors separating (isolating) the last two nodes on the power rail?

Two sections of filtering of interaction can be made to have a higher degree of AC rejection without as much DC loss, done properly.

To net it out, you use the value of the resistor to drop the DC to the level you have to have, and then the value of the capacitor to ground to get the AC rejection you need and power ripple filtering you need.

[Enzo]

They are all "filter caps" in one sense, but I tend to reserve the term filter for smoothing out ripple. Since as RG mentions, the ripple is pretty much gone after the first couple filter stages, the remaining nodes - usually for the preamp - are more commonly referred to as decoupling caps. They are there to prevent the signal from riding the B+ between preamp stages.

The OP question was really about increasing the decoupling between hus preamp stages after all.