Drop 140V Screen Voltage / Reservoir Calculation

[tschwarz]

Hi builders,

a while back I asked another amp forum's experts how I could drop lots of screen voltage from HT. The best advice I got was to use a PT with a center tap and feed the screen off the center tap at 1/2 HT. Excellent! But that requires a low voltage CT power transformer, in the 120 - 0 - 120 range, with 120mA max current, which is very hard to find.

The "ideal" load line for a russian 6P18P P/P stage requires roughly 170V on the screens with 320V at the plates. Let's just assume this to be the ideal setup for my question - I'm well aware that by changing HT, screen voltage and OT impedance one could arrive at a completely different setup. Lets also assume that the "ideal" PT is not available. 250V / 120mA no CT with a bridge rectifier it is for this build.

If I use a resistor to drop the ~140V from HT to feed the screens at ca. 6 mA quiescent screen current per screen, I'd need a 22k resistor. At full power the screen current would rise to 14 mA. Which it can't, though, because that would theoretically drop 300V over the 22k! As I understand it, this setup creates a huge amount of sag and limits power output to some value far below the calculated 18W.

Now for the question: since the 8mA jump in screen current is highly dynamic, could a (large) reservoir cap be used to provide at least parts of the current spike to the screen and keep its voltage nearer to the quiescent value? How would I calculate or at least estimate a value for this capacitor?

Thanks for sticking with me to the end of this long winded question!

Cheers, Tom

PS: How can I embed the graphics I've attached directly into the text?

[pdf64]

A classic solution would be to use an OD3 as a dropper for the screen grid HT node
https://frank.pocnet.net/sheets/137/0/0B3.pdf
As per the Leslie 122
http://www.captain-foldback.com/Leslie_sub/Leslie_schematics/122.GIF

A modern option would be to use a MOSFET amplified zener. Or just a beefy zener.

[thetragichero]

how about: use two caps in a totem pole. run the screens from the point where the two electro caps meet. that should give you half of the b+. i would suggest a bit more filtering before the screens, maybe 1k resistor and another electro cap?

[PRR]

120V AC transformer and a voltage doubler

120+120V AC isolation transformer

A proper audio tube such as 6V6, 6L6, EL34

[thetragichero]

A proper audio tube such as 6V6, 6L6, EL34

well that's no fun!

sincerely,

a guy sketching out a build to put into an old reel-to-reel using a quad of 25l6 output tubes (and isolation transformer!)

[sluckey]

"PS: How can I embed the graphics I've attached directly into the text?" Do you mean like this?

First you must convert the pdf to some image format. PDF is not an image format. Then you must put the image file somewhere on the net. Copy the image URL to this message, select it, and click the "Insert Image" button to place the image code around the image URL. Then the image will appear inline within the text. (Quote my message to see the what the code looks like).

It's usually easier to just attach the image file to the bottom of your message in the same manner that you attached the two PDF files, but you still must convert the PDF to some image format. I converted to PNG format for this example.

[HotBluePlates]

... 170V on the screens with 320V at the plates. ... 250V / 120mA no CT with a bridge rectifier it is for this build.

... use a resistor to drop the ~140V from HT to feed the screens at ca. 6 mA quiescent ... At full power the screen current would rise to 14 mA. ... How would I calculate or at least estimate a value for this capacitor? ...

You wouldn't rely on a capacitor at all.   You would:

     -  do as pdf64 says & use a voltage regulator tube to hold the screen at a roughly-fixed voltage, after a suitably-sized dropping resistor, OR

     -  Use a series dropping resistor to the screens, tack a filter cap to ground (just for these tube screens) AND place a resistor-to-ground across this screen filter cap.  To keep the voltage at the screen relatively-stable, the resistor-to-ground should pass around 5-10 times more current than the screen draws at full-tilt!!

What does that 2nd bullet look like?  You say the 6P18P will draw 14mA of screen current at full-power.

     -  Current through the shunt resistor is 14mA x 5 = 70mA
     -  Current through series resistor = 14mA + 70mA = 84mA
     -  Resistor-to-Ground ("shunt" across capacitor) = 170v / 0.07A = ~2.4kΩ (2.2k or 2.7k is close enough)
     -  Resistor-to-Ground Rating:  (170v x 0.07A x 0.07A) x 2 = 1.7w (but make that 3w or 5w in a real build)
     -  Want to drop from 320v to 170v = 150v ---->  150v / 0.084mA = ~1786Ω (1.8kΩ is available in 10% parts)
     -  Series Resistor Rating:  (150v x 0.084A x 0.084A) x 2 = 2.12w (but make it a 5w resistor)

The point of this very-old (but also wasteful) approach is to make the current-change in the series resistor small compared to total current.  That also made the voltage-drop-change across that resistor small, at the cost of using more current to create heat.

Whatever current your output tubes will pull to make audio might only be 80-100mA in a 12-15w amplifier.  So you'll literally waste as much in heat knock the voltage down by half for the screens, while also stabilizing the voltage.

Or you use the regulator tube.

Or get a power transformer with a center-tap.

The "ideal" load line for a russian 6P18P P/P stage requires roughly 170V on the screens with 320V at the plates. ...

Why is that "ideal"?

The 6P18P data sheet says max plate & screen voltage is 250v.  There's no special magic to having the screen voltage much less than the plate, unless you're using a tube like a 6550 that can accommodate a very-high plate voltage for higher power output.

I would be looking for a PT with a 250v/1.414 = ~175vac secondary, or less.  The EL82 data sheet (which the 6P18P supposedly copies) also shows a 250v max voltage for plate & screen, and shows example audio conditions with only 170-200v plate & screen.

A proper audio tube such as 6V6, 6L6, EL34

I suspect low price is the appeal.

[PRR]

> I suspect low price is the appeal.

Then he's at the end of a very long line of geeks re-re-re-reading datasheets for possible audio uses. That this one is still $2 does not bode well. Though it may do OK. And I confess I have used odder tubes.

An audio amplifier is not primarily tube cost, unless you have very long hours, or abuse them. Load and Power Supply costs are significant. If a $2 tube wants custom $85 PT or $95 OT, you may do well with a $25 tube and generic PT and OT. Or maybe not? Or maybe in pandemic when your ship never comes in?

The 6V6 et al were designed to take a single power source lightly dropped for G2. In fact thetragichero's 25L6 can be used that way as a 120V 2W plan (instead of 200Vp+110Vg2 for 4W).

[tschwarz]

...

The "ideal" load line for a russian 6P18P P/P stage requires roughly 170V on the screens with 320V at the plates. ...

Why is that "ideal"?

The 6P18P data sheet says max plate & screen voltage is 250v.  There's no special magic to having the screen voltage much less than the plate, unless you're using a tube like a 6550 that can accommodate a very-high plate voltage for higher power output.
...

Several concurrent limits apply here:

- Pmax for the screen is 2.5W. At 170V and 14mA we're already almost there. Screen current rises exponentially the further left you go.
- The cathode can handle 75mA continuous current max. Need to estimate this one for a dynamic guitar signal under real-world playing conditions. I'm assuming 50% between max and quiescent plate currents and RMSsing that, so (135-30)/1.4 or roughly 75mA. Already there. I might be completely wrong with this homegrown formula, though. If I am, please set me straight!
- Current handling and efficiency in the output transformer - higher current requires a larger copper cross-section, requires a larger core.
- Same goes for the PT that needs to provide these higher currents. More copper, larger core, more chassis real-estate. Off-the-shelf low voltage & high current PTs with heater supply windings are difficult to find. Unless you have them custom-made. The Hammond 369AX (I need a 230V primary) that HotBluePlates suggested  can only provide 2A of heater current. That's just 2 power tubes and 1 pre-amp tube...

To stay away from the high screen current area the load line needs to cross the 0V grid line in the knee or to the right. Two ways to achieve that - as far as I understand it:

1.) aim higher with the load line: smaller OT primary impedance (Raa) and less HT and a (much) higher plate current. See my points above why I'm trying to stay away from there
2.) lower the knee - i.e. drop the screen voltage. Now you can up the HT and use a higher OT primary impedance. I do realize I'm over the max. plate voltage by 70V. From what I've read so far, this seems to be common practice in many power amp designs. The data sheet does list 2500V as the max. cold plate voltage...

[tschwarz]

Sorry, I forgot to mention earlier that I have a TW Rocket derivative with a 6P18P P/P power section running on the breadboard right now. I've been tweaking it for the last two weekends and really like it. It has a tremolo built from the extra triode V1b that feeds into the power pentode's grid bias, the tone stack can be disabled (lifted off ground) and a post PI master volume that can be removed (Matchless MV).
I also added a second switchable cathode resistor set to run the pentodes hotter (25mA -> 32mA). This is really noticeable in the sound and responsiveness. I'd be happy to provide the schema if anybody's interested...

[tschwarz]

> I suspect low price is the appeal.

Then he's at the end of a very long line of geeks re-re-re-reading datasheets for possible audio uses. That this one is still $2 does not bode well. Though it may do OK. And I confess I have used odder tubes.

An audio amplifier is not primarily tube cost, unless you have very long hours, or abuse them. Load and Power Supply costs are significant. If a $2 tube wants custom $85 PT or $95 OT, you may do well with a $25 tube and generic PT and OT. Or maybe not? Or maybe in pandemic when your ship never comes in?

The 6V6 et al were designed to take a single power source lightly dropped for G2. In fact thetragichero's 25L6 can be used that way as a 120V 2W plan (instead of 200Vp+110Vg2 for 4W).

Well, I don't know about everybody here, but for me this is a hobby. An intense one. I work a day job in software development and spend weekends in the lab. Others fit 8-cylinder big blocks into tiny cars or run model train tracks through their living rooms, or go golfing...

And, for me it's going to get a lot "worse" than 6P18Ps. Here's an excerpt from my power tube stash in no particular order: QQE03/2, ECL80, ECL81, ECL82 (6F3P), ECL84 (6F4P), ECL85 (6F5P), 6P1P, ECL113, ECL805, ECLL800, EFL200, 6S19P (a power triode biased at around -90V - this one's going to be real fun!), E80L, E81L, EL802, EL803, EL821, EL83, EL84 (of course), 6P14P, 6P15P (another one with a very feeble screen grid), EL86, EL183, EL504, EL82 (the "real" ones), 6P43P (~EL86), 6P36S, 6P41S, 6P44S, 6V6, EL34. Then there's the PLxyz TV tubes with all sorts of heater voltages - thinking I'll run those in series @300mA off some separate heater transformer.

I'll try to get a usable guitar amp out of each and every one of those - that's the mission. Truckloads of fun waiting...

[tschwarz]

...
A modern option would be to use a MOSFET amplified zener. Or just a beefy zener.

Thanks alot, pdf64. This solution looks like it would also provide an elegant option for power scaling. A switchable array of zeners. I'll definitely try this!
Found an excellent youtube vid that explains the concept :

Cheers, Tom