Minimum Power-node dropping resistor Ω-value

[deci belle]

I want to have an extra pwr-node after the cathode follower tube on a 5f8 circuit, but I don't necessarily want the voltage feed to the preamp channels' input plates to be lessened much at all (as far as I know).

Is there a minimum generally recommended Ω-value for power resistors that are separating power node filter caps? (I'm using 3w IRC resistors). How low can I go and serve the needs of the filter cap it feeds?

I ask because I don't know what's going on in the rc relationship of the dropping resistor and the filter cap other than creating a current reserve and lowering of the voltage to the next stage…

[Bruce / Mission Amps]

It's the dropping resistor of the node in front of that new one that will have additional current flowing and drop more voltage... so, you'll be installing two new resistors, with the original resistor having less resistance then before.
Basic Ohm's law.
Except for the very first node carrying all the preamp tubes following and the screen node.... I think you are OK with any 1 watt resistor you choose.... most vintage amps use 1/2 watt.

[daz]

I think he's asking for a resistance value more than a wattage. And as that goes, I've used 2.2k which is what they use on the 18 watt marshall clones and derivatives. So i know u can go at least that low, tho I've never seen any amps using lower dropping resistors than that.

[deci belle]

What daz said! I'm already using the IRC 3w. My '66 Showman uses a 1k right after the choke for that matter.

I just thought it mattered …hehe! If someone thinks it does, let me know.

I'm going to put in a 10Ω to start. I have a 47Ω too.

Thanks you guys!!

[eschertron]

So I'm tempted to ask what's the intended result? My basic understanding is that the stiffness of the node results from the R*C of the node. You haven't mentioned what size cap you'll add...

And then I'll ask (and Bruce might be able to answer this one, since I love to ask questions that I don't know the answer to) what difference is there between a small value R and another cap, and simply tacking another cap onto an existing node?

[potatofarmer]

This is useful: The Valve Wizard

So, you want the first triode pair to have a separate power supply node from the summing gain stage and cathode follower, right? Not adding a new tube or some such?

Have you considered adding another branch from the phase inverter node, 10K + 8uF or 20 uF? Then just adjust the 10k going to V1 and the 10k going to V2 to get the voltages back to where they were on the stock schematic.

[Chuck H]

+1

I assume the goal is node isolation for noise control as it relates to grounding and such. You can probably get away with fairly small resistances in a "string" because of the relatively low impedance of the power supply. I'm not a good enough tech to spout formulas on how to calculate the lowest practical value but I would think that 100x the impedance of any given node should isolate adequately. Why 100x??? Because guitar amps can be VERY high gain compared to other amps and I think that level of isolation might be practical to avoid the amplification of even very small signals ghosting on the supply. So if a filter cap has maybe a few ohms of impedance at most any considered frequency a 470 ohm resistor is practical. You could probably get away with 220 ohms actually since most filters will perform better than a few ohms. Since preamp tubes don't draw much current at all the voltage drop will actually be very small. Such that the voltage difference between a 47 ohm resistor and a 470 ohm resistor is negligible. That said...

Branching is easy and offers the opportunity for much greater node isolation and voltage control via the resistance value. I've done it with good results. Some Matchless amp schematics show a branched B+ string If you want to look at one.

[deci belle]

I want to have an extra pwr-node after the cathode follower tube on a 5f8 circuit, but I don't necessarily want the voltage feed to the preamp channels' input plates to be lessened much at all (as far as I know).

So, that would be a typical 5F8a 8µF cap, escherton— (haven't actually made the amp yet, just planning and getting parts)!

Paul said:

So, you want the first triode pair to have a separate power supply node from the summing gain stage and cathode follower, right? Not adding a new tube or some such?

Ya, I thought I could just add another node to the cathode follower with another 8µF cap. No new tube.

Have you considered adding another branch from the phase inverter node, 10K + 8uF or 20 uF? Then just adjust the 10k going to V1 and the 10k going to V2 to get the voltages back to where they were on the stock schematic.

Haha!! I'm making this amp have two channels (channel one is a 6G8, so both the 5F8+6G8 preamp nodes are already feeding off the PI per the 6G8's topography). The PI will have a 20~30µF cap.

I've done this before (last time I gave the last triode pair in a 5F6 an 8µF node each so I could lower the voltage of the extra gain stage triode feeding the normal Bassman input stage to 95v). I used a 10k after the cathode follower that time, and a 390k between the triodes.

I want to see what it does without lowering the voltages this time (the voltages of the 5F8 are about 30v lower than the 5F6). The PT I have is 290-0-290. So that's why I wanted to know how low I could go in dropping resistor value.

I'll have all the caps in a cap-can with leads out to the tubes on the board. This way I can easily change whether or not the input tube gets a separate cap or feeds off the PI. The last dropping resistor will be on the board too. Since the value will be so low, just pulling the lead (at the board) from the last node's cap (in the can) will be enough to change where the first tube gets its juice.

Thank you for the logical suggestion (and reminder to check the Valve Wizard site), Paul!

Wow Chuck! That's great!! A figure (like 100x)!! A tiny bit of math with some back-round application! Yay!! Thanks so much, you guys!!

I'll start with a 510Ω 3w I have.

[Chuck H]

Thanks for the like. I want to qualify my above figure of 100x! It's purely speculation. Suppose you have as much as 1/10th of a volt of crud on the forward power supply node. A 100x resistance to the filter impedance would reduce that to one microvolt on the following node via voltage division with that nodes impedance as the "dropper" in the divider (this is actually further reduced by that stages plate circuit impedance that I'm not figuring for, but is likely a significant consideration). That's probably alright for anything but the first gain stage. Which will likely have another RC filter stage of it's own. But if this one microvolt were to be impressed on the first gain stage, or even the second in an uber gain type preamp, consider that a gain factor of something like 10,000 could easily be in play! So that one microvolt becomes ten volts of hash and noise at the input to the phase inverter!!! This is why HV node and DC filament supply considerations are so important for modern high gain amps. Now, you'll probably never get 1/10th of a volt of noise on the screen node. More like a few hundredths of a volt from what I've seen with a well considered PS. But I just wanted to illustrate that the first node, and even the second for an amp that cascades three stages or more, should definitely be on it's own filtered node with as much resistance as practical between that first stages node and the next up the chain.

I hope I didn't make this too hard to follow... I'm not exactly a teacher or author.

EDIT: I did run a CAD simulation of an "average" guitar amp PS and the preamp voltage difference between a 47R resistor and a 470R resistor following the screen node was only four volts. The difference between the 470R resistor and a 10k (typical value) was 50V.

[pdf64]

My take on this topic is that the B+ supply should be decoupled to a lower frequency than the signal path is coupled; otherwise, with cascading gain stages, motorboating oscillation may occur, due to positive feedback via the B+.

So, if the signal between cascading gain stages is coupled with a 22nF feeding a 1M gain pot, the coupling -3dB frequency F = 1 / (2 x pi x R x C) = 7.2Hz.
Say we've got 22uF decoupling caps, then the decoupling R for the B+ node needs to be >>1k, in order to push the decoupling frequency below the coupling frequency.

From that, it may be appreciated that the more (cascading) signal gain is fed from a B+ node then the lower the decoupling frequency of the node has to be, in order to avoid the likelihood of positive feedback.
Pete

[Chuck H]

Excellent addition! But I want to ask, isn't it possible to reduce the decoupling frequency with a larger filter value rather than change the impedance to alter the knee!?! In fact the filter value brought up for the branched circuit is 8uf. I don't hear enough of a difference between 8uf and 22uf that I don't automatically default to the higher value for what seems to me like practical design reasons. I think the reason for the lower value in vintage amps was economic and not a tonal consideration and I haven't observed that higher values have tonal consequences. Since the price difference between an 8uf and a 22uf cap is truly tiny I just use 22uf.

So it would seem that I neglected to add that the filter value will need to be bumped to effectively use a low value inter-nodal resistance. Pete covered the math. I would just bump to 47uf since that's the next common value.

[deci belle]

Wow!! Even better answers piled on top!! I get to study these— yahoo!!

Thank you Chuck and Pete!!

[Chuck H]

We both get to study!!! The higher minds here are very good to reminding me about what I failed to consider in my typically rote approach to this craft. I hadn't even considered cap value as it relates to circuit impedance in this particular scenario until Pete brought it up. Paramount info. I've been able to neglect it for PS design because I've never used really low value resistors in the string. The kudos go to Pete here in my opinion!