Resistor in series with choke

[Gaz]

So I figured out a new "trick" to extract a little more power from a quad of EL84s, and keep the screens in check at the same time. The trick is to add a resistor in series with the existing choke before the screen node, while simultaneously increasing the plate voltage.

Before I was running 310vdc on plate and screens (wrt cathode) with individual 1K screen resistors. I was getting about 25W rms and the screens were still dissipating a disconcerting 4.5W each with max signal.

With the new setup I have 360vdc on the plates and 320 on the screens, achieved by a 2.5K resistor in series with the choke. I now get 30W rms, and have been able to reduce individual screen resistors to 100R. Screens are only dissipating 2W each now with max signal.

In result I have more power, better tone, and longer tube life! Bias had to be cooled off a little, of course, which result in a little more crossover distortion, but I don't noticed, and prefer the slightly less 'furry' tone that super hot bias gives.

Anyway, I do have a question here: With the resistor in series with the choke, has the choke become redundant? It has a resistance of 710R, and I know I'd have to compensate for that, but am I gaining anything by keeping the choke there, or can I take it out for another project?!

Thanks!

[David Root]

The choke will filter more ripple making a smoother screen supply that will strongly resist voltage changes, and therefore also current changes. This is good, especially with tubes that have less-than-heavily-built screens.

I'm not knowlegeable about the EL84, but the EL34 has a relatively lightly built screen, and screen failure is a very common way for tubes to go to Valvehalla!

2K5 is a good value for a screen gridstpper too, but I would keep the choke in this instance, especially as you have upped the voltages a bit. Sounds like it is helping to keep the screens under control, so... if it ain't broke, don't fix it!

In my latest build I put a small 20H choke immediately behind the screen gridstopper, which is a little different than what you are doing but has essentially the same effect, but focused solely on the screen. I also have another choke in the main PS line too.

[Gaz]

Thanks for the reply, David. You're right to assume, the EL84s have very fragile screens, and my setup seems to take care to stay under their maximum ratings (plus they are not visually glowing). And to just to be clear, the resistor is in series with the choke before the screen filter cap, not a common screen resistor.

Since posting, I a/b'd the choke in and out while keeping the series resistor and there was no audible change in hum at the output, and I think this is simply because the the screen filter at 50uf in combination with the 2.5K resistor must be plenty of filtering to keep an PS hum abated.

I would just leave it, but I'm just wondering if the choke is redundant now, kinda like $25 just sitting there atop the chassis. Sure the filtering is technically better, but if I can't hear it, is there any worth in keeping it?

I'm curious to see if I can measure the regulation being any better with choke, but I feel like DC resistance is DC resistance, and as long a that's equal the regulation should be equal as well.

[Merlinb]

Sure the filtering is technically better, but if I can't hear it, is there any worth in keeping it?

You are correct- the choke still gives superior ripple filtering; your added resistor simply makes it 'even better'. But as you say, if you can't hear it, then you can do away with the choke. You have a decent amount of RC filtering already (just compare it with other amps that don't use a choke!). The low frequency regulation will be no different if you replace the choke with an ordinary 710R resistor.

[martin manning]

Gaz, I'm curious as to how you measured screen dissipation under full signal conditions. Can you elaborate?

[JazzGuitarGimp]

Gaz, I'm curious as to how you measured screen dissipation under full signal conditions. Can you elaborate?

I have wondered about the best way to do this. It seems to me that you could inject a 400Hz sine wave into the amp, run the amp up to full clean power output, then measure the AC voltage across the screen grid resistor, then divide that voltage by the screen grid resistor's resistance to get the current into the screen. What isn't clear to me is whether you'd use the measured RMS voltage across the resistor, or convert it to peak voltage before doing the current calculation. Peak might be better, because if you run the amp full out (square waves on the output tube grids), this might yield a more accurate result.

[David Root]

I should have asked what the screen node uF was. 50uF is a lot more than I was thinking. What is the plate node uF? 100uF?

[pdf64]

Before deciding to get rid of the choke, it may be a good idea to check the amp with a 4x12 (assuming that's not what you are using) or something with good bass output.
If you are used to it with an open back 1x12 or something, the cab may be filtering out nasty low frequency by-products.
Another option would be to record it with a good mic close on your speaker - a recording can also reveal such artifacts.
I think that pentodes tend to have a higher gain on the screen grid than beam tetrodes, so more sensitive to ripple on their supply node.
Pete

[Gaz]

Thanks, Merlin, I was just reading about the subject in your PS book

The way I measure the cranked screen dissipation is measure the DC voltage across (JazzGuiatrGimp, maybe you made a mistake when you said AC?) an individual screen resistor, divide by its resistance to find current, then multiply that times the sagged screen voltage (wrt the cathode in this case).

I use an rms meter, and it seems to give me pretty good results, because sure enough when my calculations give me results under 2W, the visual glow I can see on the screens goes away

@ David, yes plates at 100uf, screen at 47uf (I think I stole it from the Rocket).

[JazzGuitarGimp]

When I wrote it, I meant to say AC. Now, after reading your response (Gaz), I would like to change AC to DC. I think I hoodwinked myself, thinking there is an AC component present on the screen under full output conditions - and there probably is, but I imagine its magnitude us dwarfed by the DC component of the screen's signal. Thanks for setting me straight.

[Gaz]

Haha, that's the fist time I've set anyone straight on this forum

Right now I have a bunch of stuff on rigged up on switches:

- bypassable choke in series with resistor (actually a decade box so I can try different values).

- 100R or 1K individual screen resistors

- Variable cathode resistor for output tubes

- power scaling on Plates, screens, and PI so I can hear different voltages

Now, I'll be honest, I don't know if I could hear the difference between any of this stuff unless was A/B'ing like I am. In general, lowering screen voltage through any method softens the sound slightly, but there's not a big apparent change in headroom. I would say making the screen resistors larger gives the most 'softness' on the attack and compression when compared to lowering the screen voltage via the large choke resistor, which I guess is due to the screen node's filter cap providing some regulation.

As far as bias, while a really hot typical 14W at idle gives a few more watts on the scope and no crossover distortion, a cooler bias actually sounds and feels cleaner, louder even though it's technically not. I also can't really pick out any cross-over distortion by ear even though it's visible on the scope. I have to go really cold to hear it, and even then it still just sounds like rock and roll to me, just a different sound I guess.

Again, the cool thing about this mod (high plates volts, low screens through the larger-ish 'choke resistor', and 100R individual resistors) is that it's easier on the screens than the way I had it before, which was 310vdc on the plates/screens with a choke and and individual 1K resistors, and it technically puts out more power, albeit it's probably negligible. The reduced individual resistors and high B+ do add some noticeable punch, which I can say for sure since I can truly A/B it in a few seconds with my setup at the moment.

Haha, just a few thoughts!

[JazzGuitarGimp]

Research: it's a beautiful thing... good stuff there, Gaz.

[Gaz]

Thanks, it's amazing what a little bench time can accomplish. I think a lot of amp designing is finding out what 'really makes a difference'.

[pdf64]

I would say making the screen resistors larger gives the most 'softness' on the attack and compression when compared to lowering the screen voltage via the large choke resistor

Individual screen grid resistors are a source of negative feedback, in a similar manner to an unbypassed individual cathode resistor - the signal Vac on the screen grids is inverted wrt the control grid signal.
With large signal swings there's a lot of signal Vac.
However, screen grid current is not linear wrt plate current, rather it seems to increase heavily with plate current, under regular guitar amp operating conditions.
Hence the compression effect with high value individual screen grid resistors.
Pete