Choke and sag

[lowell]

Does a choke reduce power supply sag?

[loudthud]

Compared to a resistor that would provide the equivalent amount of filtering, yes.

[imaradiostar]

If you have a supply designed for choke input it can also improve regulation between rest and full load for a class AB1 or 2 amp. It won't make nearly the difference for a class A amp because the current draw shouldn't change much with signal. Many old high-power designs called for choke input- dunno, I guess chokes were cheaper than capacitors.

jamie

[lowell]

Say you have a 4x6V6 amp (5E3x2) with a 5h choke vs a 100ohm resistor. B+/screen filtering at 100uf/100uf. Would the difference between the resistor/choke be audible?

Also, is the theory behind this that the amp draws more current while it's played, and even more when played at higher volumes, thus the voltage drop across the resistor vs the choke increases as current increases?

This leads to another question. The more current that is drawn through a resistor increases the voltage drop across that resistor... right? I'm also thinking that, in the case of the pi filter, the more filtering there is after the resistor the less the effect...?

Finally, is a resistor pi filter indifferent from a choke pi filter (concerning sag only - not ripple) if the filtering is very large?

[ThermionicScott]

Say you have a 4x6V6 amp (5E3x2) with a 5h choke vs a 100ohm resistor. B+/screen filtering at 100uf/100uf. Would the difference between the resistor/choke be audible?

Not if the choke is also 100ohms. What loudthud was getting at was that a choke can offer the same amount of filtering with much less DC resistance.

Also, is the theory behind this that the amp draws more current while it's played, and even more when played at higher volumes, thus the voltage drop across the resistor vs the choke increases as current increases?

That's right. The preamp tubes are pretty constant in their current draw, but power tube screen current goes up with signal, so there will be a voltage drop across the choke or resistor that feeds them.

This leads to another question. The more current that is drawn through a resistor increases the voltage drop across that resistor... right? I'm also thinking that, in the case of the pi filter, the more filtering there is after the resistor the less the effect...?

You mean larger capacitors? There is an RC time constant that affects how fast it responds, but as far as we're concerned about voltage drop, current draw is current draw.

Finally, is a resistor pi filter indifferent from a choke pi filter (concerning sag only - not ripple) if the filtering is very large?

No.

[lowell]

You mean larger capacitors? There is an RC time constant that affects how fast it responds, but as far as we're concerned about voltage drop, current draw is current draw.

But if the screen capacitors are super large then wouldn't they be able to supply more current, thus not pulling as much current through the choke/resistor? - thus reducing sag?

[Wilder Amplification]

Say you have a 4x6V6 amp (5E3x2) with a 5h choke vs a 100ohm resistor. B+/screen filtering at 100uf/100uf. Would the difference between the resistor/choke be audible?

Also, is the theory behind this that the amp draws more current while it's played, and even more when played at higher volumes, thus the voltage drop across the resistor vs the choke increases as current increases?

Correct.

This leads to another question. The more current that is drawn through a resistor increases the voltage drop across that resistor... right? I'm also thinking that, in the case of the pi filter, the more filtering there is after the resistor the less the effect...?

You've got two things that happen. First off, voltage drop across a resistance increases when current flow increases through said resistance.This happens no matter how much filtering you have as the filter caps have to draw that same current that was pulled from the filter cap to replenish the charge that was depleted from the filter cap.

Second off...the filter cap value is what determines how much "on demand" current you have. If the filtering is of a lower value, the power amp taxes the current supply beyond the amount required to recharge the cap in a timely manner. This causes the filter cap voltage to drop (i.e. "sag") to a value that averages out to what's being taken from it vs what it's being re-supplied with from the supply. The resistor adds a higher time constant to the mix...which reduces how quickly the filter cap can replenish the charge taken from it. This effect becomes more apparent the higher the value of the resistor used.

Finally, is a resistor pi filter indifferent from a choke pi filter (concerning sag only - not ripple) if the filtering is very large?

Again...the filter cap has to replenish the charge that was pulled from it. The resistor increases the recharge time constant.

[Wilder Amplification]

But if the screen capacitors are super large then wouldn't they be able to supply more current, thus not pulling as much current through the choke/resistor? - thus reducing sag?

No because whatever current they supply they have to draw through the resistor to replenish it.

Yes...increasing filtering at the screen node reduces the voltage sag, but when you introduce a resistance between said filtering stage and the supply, it offsets the effect of increasing filtering. No matter the filtering value, the cap has to be able to replenish the charge that was taken from it. The added resistor slows that recharge time down, which creates the same effect of reduced filtering. In other words, it has to be able to recharge just as fast as the current was pulled from it to get the full effect of increased filtering.

[tubeswell]

Hi lowell

you could always build a switchable CRC/CLC filter for the price of an extra resistor and a DPDT switch, and then you could hear the difference.

FWIW here again is the scheme from a 5E3 I did a couple of years ago with this in it. I found that running a shared 4k7 screen resistor with the choke switched in sounded better. 2CW (I think - now having read Merlin's book - that a further mod I would make to this would maybe be to put a 2M2 resistor across the 'open' contact for the CLC filter when the CRC filter was switch in (to prevent the screens from instantaneously floating when switching)

[lowell]

I dig... good idea.

[lowell]

tubeswell,
curious, how does that amp sound w/ the whopping 4k7 screen resistor there as far as sag is concerned? Also, what's the screen to plate voltage?

Another choke related question:
An inductor opposes change in current. So when the power tubes in a push-pull circuit are greatly fluctuating in current, does the choke attempt to counteract this? And if it does why does a choke not cause MORE sag than a non-wirewound resistor? I'm thinking maybe the frequency at which current is changing is relevent here. If someone could clear this up for me and possibly use a freq= equation relating to the B+ current that'd be great.

[Steve Conner]

:
An inductor opposes change in current. So when the power tubes in a push-pull circuit are greatly fluctuating in current, does the choke attempt to counteract this? And if it does why does a choke not cause MORE sag than a non-wirewound resistor? I'm thinking maybe the frequency at which current is changing is relevent here.

If you get the design wrong, a choke can indeed be worse than a resistor. The voltage can not only sag but oscillate up and down.

Luckily with the large electrolytics used nowadays this isn't a problem, the resonance is at a very low frequency and highly damped. But back in the days of paper-in-oil smoothing caps it could cause major hassles. Choke input filters were supposedly worse than capacitor input.

The filter is just an RLC circuit, so the normal equations for resonant frequency and Q apply, except that the load damps the Q too. You can read about it on ham radio mailing lists, where they still use these kinds of filters for their big tube linear amps, and have to stop the B+ from pogoing around as they pound on their Morse keys.

[loudthud]

I have done two simulations. One with an choke and one with a resisitor. The stepped load feature was used with I1 going from 80mA to 220mA at the 1.1 second mark. Note how the green line (screen voltage) shows some ripple in the RC sim, drops farther but slower. In the LC sim the screen voltage shows some ringing as Steve suggested but the 120Hz ripple is very low.

[Bruce / Mission Amps]

High Z, 100Hz and 120Hz tank circuit

Here's an old Ham Radio, Hi-V, Hi-Z, power supply tank circuit filter trick I've used a number of times for those 250-500 watt, transmitter RF tube amps.
For amps with lots of high current screens, use a higher current rated choke.
It's good for lower screen voltage tubes like 6146Bs and the like.... I can't see any reason it wouldn't work with other pentode tubes using 350v-400v screen nodes either.
Anybody see a flaw there?

[imaradiostar]

Hey Bruce, that's interesting. What kind of input and output voltages?

jamie