Feeding 807 G2 - Disquisitions ........... (and questions)

[kagliostro]

Some tubes like to have G2 feed with a lower voltage than B+

I mean tubes like EL500 - EL504 - GU50 - 807 .....

to achieve a lower voltage to supply G2 on some tubes like this requires a voltage that has a large drop if B+ is considered

so a resistor in series with G2 isn't sufficient because the voltage will go up and down of a large amount depending on G2 current

and there is the risk to damage the tube when the current is low and so the voltage that feed G2 is high

the better solution will be to have a dedicated tap on the PT as to have the exact required voltage, but often we don't have

a specific PT and we would like to use a "standard" PT

Using a "standard" PT the lower voltage for G2 may be obtained using different methods, someone uses Zener, other a VVR

in some old circuit I've seen Gas voltage regulators Tubes

On the 807 datasheet the spec says you must use a voltage divider (a pot)



Assuming I've B+ @ 380v and I want to supply G2 @ 250V (to be within the indication of the datasheet) I must use a voltage divider like this (I think)



but ... which is the consumption in current from the PT ?

Which is the correct way to establish the values of the two resistors respect the current flowing on the Voltage Divider ?

on the datasheet there is a current of 2-5 mA flowing on G2 but I'm not able to use this data to go further

---

One other thing regarding this arrangement

As far as I can know in an SE amp if I want to introduce SAG (consider SS rectifier PS but Vacuum rectified PS will act the same as there isn't a real difference in current from idle to full power) I can add a resistor in series with G2, when current increases, voltage drop and SAG effect appears

But in the eventuality G2 is supplied using a Zener or VVR or Gas Stabilizer, is this trick usable ?

A resistor placed between the Zener (or VVR or Gas Tube) and G2 will do his job or his action will be counteracted by the preceding circuit ??

I think that the in series resistor will anyway result in SAG, but not sure about this

Thanks

Franco

[sluckey]

When you use a simple resistive voltage divider like that to develop a lower voltage, the old rule of thumb was to have 10 times as much current flowing through the resistor divider as the load will actually need. IOW, if you expect the total screen current to be 20mA then you need to have 200mA flowing through the divider. So, a high resistance divider is not very good. There are other much better methods of achieving lower screen voltage.

[kagliostro]

Ciao Steve

There are other much better methods of achieving lower screen voltage.

Which is your preferred method ?

Franco

[sluckey]

I like the method that uses a full wave bridge rectifier and connects the PT CT to the junction of two series connected caps. The voltage at the PT CT will be exactly half of the voltage at the output of the bridge. If you want to add some extra filtering, just connect that junction through a 500Ω resistor to another filter cap and use that cap to feed the screens.

[TIMBO]

Hi K, The Local made Goldentone's used the PT CT and dropping resistor to achieve a lower screen voltage.
OR as in the MAXIM a 6V6 also to get the desired voltage.

[HotBluePlates]

Some tubes like to have G2 feed with a lower voltage than B+

I mean tubes like EL500 - EL504 - GU50 - 807 .....

There are layers of misunderstanding driving this thread. Not because you don't understand tubes, but because you're looking at the 807 outside of the context of its times.

The 807 is nothing more than a 6L6G with a top cap, which means you can apply a plate voltage higher than the 6L6G's ratings and not risk arcing to another pin on the bottom of the tube. Don't give it any magic, just treat it like another 6L6 (within the limits of its ratings).

Screen (G2) voltage determines the peak plate current possible from a tube. Plate current * Load impedance determines the peak plate voltage swing possible from a tube, as long as the B+ is higher than this by some amount (the saturation plate voltage). Peak plate current * Peak plate voltage / 2 = RMS Power output.

If plate current (swing) stays the same, you can increase the load impedance and B+ to raise the peak plate voltage swing. If you notice all the 807 conditions you posted kept roughly the same screen voltage, so the plate current capability of the tube stayed the same. But to get more power, the supply voltage and load impedance were raised. Same current swing * bigger load impedance = bigger plate voltage swing (which was allowed by higher B+) --> more output power.

But notice the tube bias is roughly the same in all the conditions... The similar G2 voltages means bias voltages remained similar. And so did "Peak A.F. grid voltage", which is the signal input to the 807. Raising load & supply voltage while keeping G2 and bias similar yielded more output power with similar driving signal, or "better power sensitivity".

Now let's recall this is an 807, with a plate brought out to a top cap, well away from the other tube pins. As a result, the plate voltage rating is much higher, because it's less likely to arc to other tube elements at the base pins. Now you could run this tube in class AB just as you would with a 6L6, with 350-400v on both plate and screen. But you paid for a tube with a top cap, so the manufacturer assumes you'll want to run the 807 plate much higher than the rating for the 6L6.

When plate voltage is very high, it might not take a lot of screen voltage to get enough plate current swing for the power output you want. It would be especially handy if the screen voltage were 1/2 the plate voltage (then you could use Sluckey's method of getting 1/2 B+ for the screen, and sending full-B+ to the plate).

Hopefully you now see why it seems like when you see an 807, you see a higher plate voltage & lower screen voltage. If you weren't running the plate voltage well above screen voltage (or the 6L6 plate voltage rating of the day), it would've seemed silly in the old days to bother with using an 807 at all.

... On the 807 datasheet the spec says you must use a voltage divider (a pot) ...

What it really says is, "Don't use (just) a series resistor to get your screen voltage."

The series-resistor idea is, "I've got 500v and want 250v on my screen, and the screen current will be as high as 20mA, so let me just stick 250v/0.02A = 12.5kΩ between the plate supply node & screen to get my reduced screen voltage."

Imagine one of the preamp pentode stages you've looked at recently. Each had a cap from screen to cathode or ground, along with a series resistor from B+ to screen. This arrangement, minus the cap to ground, is using "a series resistor to set screen voltage." The screen current varies with applied voltage, so the screen voltage will vary if you do this, because there is a relatively high impedance (resistance) between the screen and the voltage source.

So stick a 20-40µF cap from the screen to ground. The cap supplies much of the varying screen current, and this starts looking a lot like an average guitar power supply (and it is exactly that, except for deriving the lower screen voltage, if needed).

You could add a resistor to ground across the screen node filter cap, but it will create a lot of waste heat and probably won't add much stability to the screen voltage without that waste power. You could use a regulator to get the stable screen voltage, but you just transferred the waste power from a resistor to a solid-state device. The worst case with the screen node cap is that there might be a little screen voltage sag, but no more than any other guitar amp, especially if you derive the lower screen voltage as in Sluckey's method (it's a relatively low-impedance voltage source).

[kagliostro]

Many Thanks HotBluePlates

I think I begin to understand

Yesterday I posted a replay thanking Steve & TIMBO but seems it is lost

I was saying that at the moment I've under hand a 270V @ 80mA toroidal PT (14V for heaters)

and so I can't apply the trick using a PT with CT and feed the screen from the CT

I was planning to use ~378V B+ (270V x 1.4) and to lower it in some way for G2 till 250V using as output a 4500ohm OT (following the spec on datasheet)

In the Standel 25L15 I've seen that G2 is supplied at a much more higher voltage (~310V G2 with ~410V B+), and reading your explanation

I suppose I can go with more of 250V on G2, but this confuses me about the output impedance, rising G2 voltage till ... (till how much with

378V B+ ???) but, above all, which primary load impedance requires this higher G2 ? Can I assume that the same Z (4500ohm) is OK ?

Franco

[PRR]

Load impedance should be very close to V/I.

25W tube, 378V of B+, say 350V across the tube, 100% Pdiss. 25W/350V= 0.071A. 350V/0.071A= 4,900 Ohm happy load.

The datasheet for 6L6-type pentodes will usually suggest 10%-25% lower, because this gives a higher number for maximum power, by forcing current to increase above the maximum safe idle current. So 4.5K.

For listening, I would go a bit high, and 5K is right there.

G2 voltage has to be high enough to pull the peak current, 142mA, without G1 going positive. 200V-250V is probably plenty. You can run G2 higher, if it is allowed, and if you accept the larger bias and drive required to balance the excess G2 voltage.

I'd just set a MOSFET on a voltage divider. In fact that's what I did, once. A passive divider heavy enough to be stable despite large and variable G2 current is a LOT of heat. It was the main fault of a Fisher Hi-Fi I once had.... you could cook dinner on it.

[kagliostro]

Many thanks PRR

So 5K is a good load for that conditions

I'll adopt this value for the OT

and a Mosfet will be used for G2 supply

Franco

[kagliostro]

I've recently discovered the spec of this regulator

LR8, input max 450V output 1.2 - 440V @ 30mA

http://it.rs-online.com/web/p/regolatori-di-tensione-lineari/8293266/




I think it will be an interesting component for G2 VVR

What do you think ?

Franco

[PRR]

Why do you want a "regulator"?

IMO Vg2 should stay proportional to Vp.

What is your Dissipation? I had calculated 380V-250V at 2.5mA to 5mA as 1/3rd to 2/3 Watt. That tiny package may be rated over 300mW (0.3W), but must be derated for ambient heat and general safety.

[kagliostro]

I don't really want a regulator, it seemed me a simply way to tame B+ at lower level to feed G2 at 250V (as in the datasheet)

The LR8 is very cheap and this is one other reason to be interested to use it (also if the price of an IRF820  isn't a problem)

Previously you told to use a mosfet as voltage divider, my poor knowledge bring me to think about a VVR used as fixed VR

so the use of a regulator didn't seemed (to me) odd

Franco

p.s.: Ahem ....... how is the scematic of a voltage divider that uses a mosfet instead of a resistor as dissipation unit ???

[PRR]

STP6NK90Z, canale N, TO-220, 3 pin € 2,15
12V (7V-20V) Zener, any watts

[kagliostro]

Thanks PRR

but in presence of a zener didn't we have a regulator ??

Is the difference in the fact that the zener isn't connected to ground ?



Franco

[HotBluePlates]

but in presence of a zener didn't we have a regulator ??

Is the difference in the fact that the zener isn't connected to ground ?

Ultimately, all you're doing is taking a high voltage and knocking it down to some lower voltage.

You could do that with a simple series resistor (sized based on screen current draw), but then output voltage varies with current.

You could do it with a resistor voltage divider, but output voltage again will vary with changing current through the resistor between B+ and screen.

You could use an active device (like a MOSFET) instead of a resistor voltage divider. In PRR's drawing the 470kΩ and 1MΩ resistors divide the raw B+ and present it as a gate input. Source output for the MOSFET will be near the voltage of that gate voltage set by the resistor divider. Now the MOSFET assume whatever conduction it needs to maintain its source near gate voltage (so its internal resistance is variable).

Your drawing substitutes a zener in place of the 1MΩ lower resistor, so in the end it just adds another semiconductor to the circuit. Functionally, it is the same as what PRR posted.

True regulation would require some kind of sampling of the MOSFET output, and comparison to a reference voltage (maybe from a resistive divider, but more likely from an independent voltage reference like a voltage reference tube or a zener) to automatically control the output voltage.

I know you're used to seeing/thinking of the MOSFET circuit as a "regulator" but perhaps instead you should just think of it as an "active (controllable) resistor".

[PRR]

> in presence of a zener didn't we have a regulator ??

No.

The to-ground Zener conducts all the time and sets the Gate voltage (and thus the Source voltage).

The Zener in the plan I sketched normally never conducts. The plan works without it, except when you turn-off the cap would discharge into the falling Gate and zap the Gate oxide. It is only there for protection in transient conditions.

[kagliostro]

MANY THANKS HOTBLUEPLATE & PRR

finally I've understand

Franco

p.s.: thinking to the mosfet as a controllable resistor is of great help