voltage divider for screen grids

[Liquids]

The thing I was wondering about using a voltage divider to set screen voltages in an output stage is:

Allowing for the required amount screen current, wouldn't you need to set up the resistances in the divider so that they pass a relatively high amount of current through the divider - such that any changes in screen current do not hugely affect the voltage at the knee of the divider so much? In which case wouldn't the resistors in the divider need to be lowish resistance (and high-power) resistors, and in which case would they not use up a whole heap of B+ current?

That's an element of what I am curious about and was asking myself...
I know of one guy (Steve/Hangman @ ampegv4.com) who used 20 Watt resistor pair of 2.2k/10k voltage divider, for a time, and said newcomb (sic?) amps used a similar configuration. Of course, the lower the overall resistance of the divider, the more heat yielded apparently, and this amp is already "a pretty good space heater" as is.

I assume the highest wattage/lowest resistance pair was used in order to minimize current. That said, the screens shouldn't be drawing (much) current anyhow, and while most people are aware that they do, it's evaluated in amps that run the standard configuration where the screens and plate supply are rather close in voltage. Likewise, the preamp valves shouldn't draw much current, but it's a design factor if you are 'tapping off' the B+ for the voltage divider and only using that for the screens and feeding the preamp's tubes separately through it's own tap, or if the preamp tubes are following the voltage divider/screen tap in series.

I know when I was screwing with a tube preamp, I wanted to mess with dropping voltage to hear tonal differences. Indeed, lower voltages sounded different than using higher voltages (think Soldano, Mesa Dual Rectifier) for preamp tubes. However, the 'sag' effect yielded by using very large resistors before a filter capacitor to ground, in order to effectively drop the preamp voltage incrementally became increasingly noticeable after a certain point of resistance, too. Hard to isolate the variables without a vast variety of parts and components, at your fingertips (including time and motivation), and know what difference one think's is being heard may best be attributed to...

Again though..my question is...does bypassing each resistor of the voltage divider with a capacitor effectively reduce/eliminate the 'seen' resistance in the voltage divider? Indeed, does significant stiffening and reduction of the screen supply cause the screens to draw significantly less current than the common (for guitar amps) configuration where each power tube screen is effectively isolated from the plate supply but at/just above/just below the voltage of the plates?

Has anyone analyzed an amp, such as an SVT with modded with proper screen resistors, 6550s run properly (with the screen voltage on it's own rectified transformer tap, at a voltage way below the plate voltage) for how much screen current they're putting out, and if it's noticeably lower than the more common, aforementioned configuration in guitar amps?

[Andy Le Blanc]

I believe the answer is yes...

" bypassing each resistor of the voltage divider with a capacitor effectively reduce/eliminate the 'seen' resistance in the voltage divider", at audio frequencies. Its the same rational as bypassing the screen of a sharp cutoff pentode such as a 6sj7, when its used as a gain stage for audio.

getting back to the dropping resistor value and wattage...

the advantage is that you can use a low wattage bleeder resistor and use the value of a normal wattage screen grid resistor to set the voltage present at the screen grids of the power tubes.

the larger the voltage you can drop the better the regulation you can achieve in this very simple approach, but... the practicality is limited by doing so, just as tubeswell stated.

Now I haven't gone so far as Liquids has with a screens voltage supply, but that smacks of "preferred circuit conditions" I've seen in data sheets for some power pentode types, thank you for the continued interest and exploration, I think your creating circuit stability with the increased capacitance, it becomes voltage supply circuit the same as a "bias" supply. Increasing the capacitance values in the power supply is a commonly stated method to improve the low frequency stability,don't know if you've run into a "motor boating condition"... but I think you can keep it conceptually as a voltage supply without the same bypass concerns that the screen grids of a pentode gain stage , like a ef86 or 6au6 etc...

[Colossal]

Good to see some activity on this thread. This is a very interesting topic. Andy, sorry I haven't posted 'official' voltages for my previously stated amp. I believe I have 1k/221k on 4xEL84. I am running around 360V on the plates but have not measured the screens. Sounds fantastic.

[LeftyStrat]

This is a very interesting topic, especially since I have been trying to determine the magic behind the AC10, an amp with 100R resistors on the plates and the screens tied directly to the same B+ node as the plates.

I'm starting a build of a minimal version (just the EF86 without the tremolo), and dug around in my parts and found a big 25w rheostat to experiment with.

[Structo]

Glad you guys are still working on this.

What I still have a question about is, take my 100w ODS clone.
4x 6L6GC power amp.

I have 500R screen resistors on each of the sockets.

Is it better to have a 220K to ground from each screen resistor?

Or just put one 220K from the screen node to ground?

Which way develops the least amount of heat?

Sorry for being so dense, I just can't wrap my brain around this.

[Colossal]

Is it better to have a 220K to ground from each screen resistor?

Yes, this is the way you want it. Each screen resistor gets its own divider resistor to ground. Least amount of heat this way as well if divider resistor is a high(ish) value.

[Andy Le Blanc]

Oh ya... try and grab a before, and after installing a bleeder with the plate/screen measurement.
see where things are sitting at the bias point your setting. I've been landing around a 5v difference between the plate and screen. Funny...
It alludes to the possibility of using the bias to set a preferred difference as a primary objective, one more way to look at bias.

when you find the spot you like you can repeat it critically...

[Structo]

Sounds good.

Does this affect the voltages downstream to the preamp?

[Liquids]

Sounds good.

Does this affect the voltages downstream to the preamp?

It is an option - it can if you want it to; it can have no effect if you'd like things to otherwise remain as-is. It just depends how you wire it, intentionally or inadvertently...

The voltage divider does create a 'load' on the B+ due to the added resistance between the power supply and ground. The amount of resistance added, 'and' how many times you add it (if you do it individually at each screen, for instance) will have some effect the apparent B+ voltage, and/or the DC ripple to some degree. That can have an effect on the amp's power supply/voltages globally.

However, to say it has an effect is not to suggest it's definitely going to be measurable, let alone notable. It may be negligable, or it may indeed be significant. It depends a lot upon the details of the values used, the transformers, and the approach taken, etc....

I believe the approach(es) Andy suggested for those specific circuits have little to no direct effect on downstream voltages, beyond the aforementioned, since he is doing mods on a 'side tap' off the screen supply, or the "B+2" if you will, while the preamp voltage taps effectively see the same circuitry as before.

[Firestorm]

Sounds good.

Does this affect the voltages downstream to the preamp?

Interesting question. But for all the talk of reduced preamp voltages sounding "browner," the truth is that preamp plate voltage has pretty much zero effect on stage gain and only a little more on the needed center bias voltage to maintain headroom. If you drop a few volts, you will never hear it. It probably takes 50-100V to make a difference and you can compensate for that completely by changing the bias point (Rk).

[gingertube]

We seem to be talking about 2 things here.

1) Is the dropping of the screen "idle" voltage below the anode voltage by the addition of that divider. This will help limit screen current and hence dissipation, it should be good for tube life too, not just tonal change.

2) Is the presence or absence of the filter capacitor after the top dropping resistor, which changes the AC impedance seen by the screen. That is it changes the amount the screen voltage changes with signal current swing. Inclusion of the filter cap will make the screen supply "stiffer" and give you a more HiFi like response. Leaving out the filter cap allows the screen to move with signal and this is where you could expect to see some (extra) compression effect.


Is this the way you folk see it too?

Also along these lines I would like to mention to you the old Mullard HiFi Amp trick which was to use a common screen resistor for both sides of the push pull. As one side (of the push pull) drew more screen current, the other drew less and then swapped for the next half cycle, the result being a more constant current draw through that common screen resistor and hence a more constant screen voltage. This gave the "tight" sound which is good for HiFi but not necessarily what you want for guitar or at least for some guitar sounds.

This old Mullard trick gives you one more design option. A common divider with individual screen resistors off the divider. To fit the discussion above that would mean a 500 Ohm / 220K divider with individual 1K screen resistors from the divider. That would give similar sound to the individual dividers with bypass caps option but without the need for any bypass cap (always nice to ditch electrolytic caps when ever possible).

Cheers,
Ian

[Andy Le Blanc]

What I noticed before the extra bleeders was that if the screen was slightly positive it would first have to swing to 0
before going negative relative to the plate under dynamic conditions, so you have an extra swing in the voltage across the device
as it handled the current imposed with signal, an extra source of distortion.

The condition is alleviated by simply keeping the screen negative to the plate.

The difference measure is pretty precise, you can dial in "0".
The simplest solution with the bleeder lands usually in the negative.
I've set up the extra caps and bleeders in various voltage divider configurations, you can even throw in a diode
to decouple a bleeder and cap from the rest of the PS, you can go very large with the cap or you can go small and treat it as bypass, so far I like the simple bleeder.

the stiffness of the supply usually reflects upon the first couple PS filter stages, I haven't noticed that and extra cap in the screen supplies stiffened it up

[chocopower]

Just for share proposits, years ago i repaired an old spanish tube amp with a voltage divider for G2:

http://s101.photobucket.com/albums/m74/ ... IGINAL.gif

[tubeswell]

Just for share proposits, years ago i repaired an old spanish tube amp with a voltage divider for G2:

http://s101.photobucket.com/albums/m74/ ... IGINAL.gif

That's an example of what I mean as a poor regulator.

I make that the input voltage is 570 (so from top-to-bottom that divider passes 25.33mA of B+ to ground). The upper leg of the voltage divider drops 20V across the 500R. That is 40mA. The remaining 550V is dropped across the 22000R, which is 25mA. So 15mA is available to go to the screens (7.5mA per screen). But 15mA is a relatively significant proportion of the total 25.33mA going to ground. So as the screen current varies, the output voltage of the voltage divider must also vary quite a bit. This may be okay if you like dynamic screen compression, but it won't provide very constant regulation when the screen current shoots up.

Whereas if the voltage divider was configured to pass say 10x the amount of screen current to ground, then the variations in screen current wouldn't have such a noticeable effect on the output voltage at the knee. But that would mean configuring the voltage divider to pass 150mA of B+ to ground! (which is a huge waste of B+) which would mean having 134R/3666R voltage divider, where the 3666R resistor would be dissipating 82.5W!

I rest my case.

[surfsup]

Couple questions as I follow along:

The upper leg of the voltage divider drops 20V across the 500R.

Where do you get the 20V? half the 40 volts total drop on the schematic? I dont think that is correct because the first 500R has all the preamp current through it plus the screen current, the screen 500R only has the screen current so it wouldnt be 50/50. I guess it would only be a few volts difference but just curious on the clarification on the 20V.

That is 40mA. The remaining 550V is dropped across the 22000R, which is 25mA. So 15mA is available to go to the screens (7.5mA per screen).

Aren't you also assuming here the max current is 40mA through the divider network when really the max current is if the screen went to zero, which would be 550v across the 500+200 = 785mA (assuming the screen went to zero (unlikely) and the resistors didn't fry - which they would). My point is I don't think 40mA is a "limiting" current value for the screens. It is only valid across the 220k.

What I noticed before the extra bleeders was that if the screen was slightly positive it would first have to swing to 0 before going negative relative to the plate under dynamic conditions, so you have an extra swing in the voltage across the device as it handled the current imposed with signal, an extra source of distortion.

Bear with me here but why would this cause distortion? The screen grid simply attracts electrons in its attempt to draw them to the plate, as I understand it. So any "signal" landing on the screen doesnt make it to the speaker because it is filtered out by the screen filter cap. Thinking about this more, wouldnt this screen voltage simply be a way to control power?

ok now I'm ready to get b*$&ch slapped back to reality...!