question RE: PSU filtering

[12AX7]

If one were to want less voltage at the PI of a marshall type amp, but did not want to lower voltage further upstream, a parallel filtering circuit will work. (with bigger resistor to lower the PI's voltage) I've done it, but I've wondered about one thing. More filtering at a given node=a tighter feel/sound. Will the preamp stages upstream from the PI lose some of that if the PI node is in parallel to them? Each node has it's own filtering, but I'm not sure how much that filtering affects the overall filtering at stages upstream. Not thinking about ripple here, only tone/feel. I want to lower the PI voltage by giving it it's very own branch from the B+ rail then taking a second parallel rail to V1 and V2, leaving them in series with the PSU rail as normal. Would i need to add more  MF to have the gain stages give the same sound and feel?Of course i realize i would need to add another resistor to the preamp nodes to make up for the missing resistance of the PI not being included.

[HotBluePlates]

... More filtering at a given node=a tighter feel/sound. ...

It might help to think in terms of "Electrical Cause, Electrical Effect, Sonic Result."

Electrical Cause: More filtering (capacitance)
Electrical Effect: Lower power supply impedance
  2nd-order Electrical Effect: Less voltage drop/sag
Sonic Result: Tighter feel/sound

Now what happens if you go in the opposite direction?

Electrical Cause: Less filtering capacitance or Lossier rectifier or Resistor instead of choke or More power supply resistance
Electrical Effect: Higher power supply impedance
  2nd-order Electrical Effect: More voltage drop/sag
Sonic Result: Looser feel/sound (with significant power output)

Now connect the dots between the 1st- and 2nd-order Electrical Effects: "More impedance" is related to "More voltage drop" by Ohm's Law; the missing element is Current.

In a guitar amp, output tube plates draw significant (varying) current with varying power output; this could be 10's to 100's of mA's. Output tubes screens draw some varying current with varying power output; this could be 5's to low-10's of mA's (generally on the very low end of that range). Preamp tubes draw the least current with varying output power; maybe as much as 1 mA, or a hair more.

I would therefore suggest that the preamp current draw and filtering has little effect on the overall amp feel. The rectifier/filter (and PT capability) setup around the power tubes dominates the situation.

[Joe_L]

HotBluePlates is dead on. 

Additionally, preamp stages are typically class A which means the current draw does not vary.  The main job of the filter nodes is to decouple previous and following stages from each other and voltage is a secondary consideration.  For proper decoupling, the resistor and capcitor should be chosen to have a cutoff frequency at around 1Hz or below.

The formula is:

f=1/(2*PI*R*C)

So, for a typical Marshall filter node:

f=1/2 * 3.14 * 10000 * .000047 = 1/2.9516 = .39Hz

As you can see, 47uF with a 10K resistor is more than plenty.  That is why you see 22uF used in some of Doug's circuits.  It is plenty enough for proper decoupling and will not sound any different.

Google "rc time constant" for more info.

..Joe L

[HotBluePlates]

Joe, that's a very interesting point!


Something related to what you said that I didn't think of until now:
You want the filter cap to look like a dead-short to your intended signal. Hence a -3dB point well below any intended audio signal. That's because if the -3dB point was up in the desired pass-band (in the audio range), a feedback signal can develop across power supply impedance. If you're unlucky, that feedback will be positive between a pair of stages, and you get motorboating.


New thought: An alternate and related issue is where are your bass roll-offs in the signal chain of the amp? You may get away with less filtering if you roll off the bass higher (make the amp brighter) so the filter cap still looks like a very small impedance.


The "Gotcha" with that is the feedback loop around the output stage from speaker to (usually) phase inverter... Bass roll-off via a cap also causes phase shift. The more roll-offs (more stages of caps in the circuit), the more the phase shift at some low frequency. If you also use a lot of negative feedback, the phase shifts turn it into positive feedback.


Generally, designer's side-step this issue by including as few roll-offs (phase shifts) inside a feedback loop as possible. But before everyone worries about this issue, guitar amps generally have less feedback than would cause a problem, and you're probably safe until you get "creative".