Elevated heaters and maximum heater-to-cathode resistance

[theRagman]

I'm looking at elevating the heaters in a build that's similar to a Marshall 2203 by using a voltage divider off HT. There is a warning in Merlin's book about exceeding maximum heater-to-cathode resistance by using too large a resistor in the lower leg of the voltage divider. In the data sheet I have for EL34s, max heater-to-cathode resistance is 20k ohms.

How do I apply that limit to calculating an appropriate value for the lower leg of the voltage divider? Counting the preamp, I've got seven tubes sharing that reference point, and I have no idea how that resistor interfaces with each tube's heater-to-cathode resistance.

There are plenty of examples online of voltage dividers for heater elevation in this and amps like it, and I'm sure I could use any one of them and not blow up my tubes, but I'd like to understand the principle here.

[tubeswell]

Page D4 https://frank.pocnet.net/sheets/129/e/EL34.pdf


Vg-k max = 100V

[theRagman]

Page D4 https://frank.pocnet.net/sheets/129/e/EL34.pdf


Vg-k max = 100V

Isn't that a distinct issue from maximum Rh-k (given on the same page as 20k ohms), or have I misunderstood that?

[Merlin]

How do I apply that limit to calculating an appropriate value for the lower leg of the voltage divider? Counting the preamp, I've got seven tubes sharing that reference point, and I have no idea how that resistor interfaces with each tube's heater-to-cathode resistance.

Rhk(max) is the maximum recommended resistance between heater and cathode. In a normal circuit that will amount to the cathode bias resistor (if used), plus the lower resistor in the elevation divider. So yeah, 20k is kind of a difficult limit if you want to get the elevation voltage from the HT. It's not clear exactly what happens if you exceed Rhk, but a lot of amps do, and seem to get away with it.

[stratomaster]

If you want to stay under 100v and 20k then you can just solve this:

20k + R′ - where R' is the upper leg of the voltage divider

[20k/(20k+R')]*Vin=100v.  Vin is the voltage into the divider.

I don't know where you're taking the voltage from, but let's say it's 450v. Solving for R' gives you 70k. 

So keep the top leg over 70k and you're good on both voltage and resistance. 

Personally I don't like relatively low resistance to ground off the HT.  220k is a minimum I use (at 450v it's another 2mA of draw, and depending on where you put it affects voltage drop at the power supply nodes).  So maybe 220k/18k would be a good place to start.

[theRagman]

Rhk(max) is the maximum recommended resistance between heater and cathode. In a normal circuit that will amount to the cathode bias resistor (if used), plus the lower resistor in the elevation divider. So yeah, 20k is kind of a difficult limit if you want to get the elevation voltage from the HT. It's not clear exactly what happens if you exceed Rhk, but a lot of amps do, and seem to get away with it.

Does the fact that there's a whole mess of tubes with heaters all pegged to the same resistor have any effect on the resistance each individual tube "sees?"

[theRagman]

If you want to stay under 100v and 20k then you can just solve this:

20k + R′ - where R' is the upper leg of the voltage divider

[20k/(20k+R')]*Vin=100v.  Vin is the voltage into the divider.

I don't know where you're taking the voltage from, but let's say it's 450v. Solving for R' gives you 70k. 

So keep the top leg over 70k and you're good on both voltage and resistance. 

Personally I don't like relatively low resistance to ground off the HT.  220k is a minimum I use (at 450v it's another 2mA of draw, and depending on where you put it affects voltage drop at the power supply nodes).  So maybe 220k/18k would be a good place to start.

The math is pretty straightforward--I want around 50V off a 500V supply, so that's easy on its own. But, just like you said, that looks like an uncomfortably low resistance between ground and HT. Might be I choose to live with 30V or something rather than press my luck.

[dwinstonwood]

Thanks for the post theRagman, and to those who replied. I'm in the process of building a fixed-bias amp with a cathode follower, so this is all good info.

In fact, I adjusted my voltage divider circuit to increase the upper half, and moved it to the screen node. It will still give me about ~40VDC of elevation.

[Merlin]

That looks like a good compromise.

[pdf64]

That looks like a good compromise.

Any thoughts on why that limiting value is so low with EL34 / 84, what the mechanism is behind that limit being required?
It's just I don't recall it being noted with other valve types.

Edit - the Jan 69 Philips EL34 info doesn't note that limit, whereas the Feb 58 info does
https://frank.pocnet.net/sheets/030/e/EL34.pdf
https://frank.pocnet.net/sheets/010/e/EL34.pdf

[Merlin]

Any thoughts on why that limiting value is so low with EL34 / 84, what the mechanism is behind that limit being required?

I would guess it is because they have such high-power heaters, so thermal stress on the insulation is harsher (most materials become more conductive as they heat up), compared with a little 1W heater in a preamp tube.

[dwinstonwood]

Thanks Merlin.

The original Fender 6G6-B schematic shows 136VDC on the CF. I'm using a 6SL7GT in this position. Its datasheet states a peak heater-cathode voltage of 90 volts. So, with ~40VDC of elevation I should be fine with my CF set at ≤130VDC, which is close enough to the schematic.

[stratomaster]

You can always add a leg off the bias supply, as I've discussed elsewhere. Doesn't look like there's a bias tap, but the principle is the same.  In this case, the added current draw is pre rectification, so there's no added voltage drop/current draw in the HT line to worry about.

Basically build a Princeton Reverb type bias circuit with the diode reversed and put your center tap on that.

[theRagman]

You can always add a leg off the bias supply, as I've discussed elsewhere. Doesn't look like there's a bias tap, but the principle is the same.  In this case, the added current draw is pre rectification, so there's no added voltage drop in the HT line to worry about.

Basically build a Princeton Reverb type bias circuit with the diode reversed and put your center tap on that.

That's a great idea! There is a separate bias supply on the PT I'm working with, and copying the existing bias circuit would work great with its current values, just with the diode flipped. Thanks!

[tubeswell]

just with the diode flipped. Thanks!

And don’t forget filter cap polarity!

[theRagman]

just with the diode flipped. Thanks!

And don’t forget filter cap polarity!

Good point!