Voltage needed to drive power tubes

[harry]

Hello,

I'm building an amp with a 5d8 power section and PI.

At the moment, the amp goes triode gain stage -> gain control -> cold biased triode gain stage -> dc cathode follower (using two sides of a 12AX7; so fed by another triode gain stage) -> tone stack -> volume control -> triode-gain into paraphase -> 2x6L6 cathode biased

I'm only getting 8W output. It should be about 25W.

The PI is driving the 6L6s with about 60V peak to peak. Is this the cause of the low output? Should it be driving the tubes harder?



Cathode voltage on power tubes is 24V. Plates are 363V.
IIRC, PI has 340V through a 100K, resulting in about 160V on the plate.


Any help would be much appreciated!

Thanks,

Harry



(also, preamp sounds great - thanks to Merlin's book, which is absolutely fantastic. I am sure this is going to be the best amp I have played when I'm finished)

[Wakculloch]

According to my data book, with the configuration you have, you should be getting about 12W. 60 volts drive signal sounds about right.
If you need more power either reconfigure with fixed bias or raise the HT voltage. The Brimar 6L6 data sheet at Franks tube pages will have the exact voltages and loading necessary for this.

[Wakculloch]

Humf! Just re-checked and indeed you have it right and it should put out 24W. If the drive signal is correct then there must be a problem with output valves, output transformer, wiring in this area, power measurement method or something like that.

[harry]

Thanks, now I know what route to go down.

I had to try and deduce how the OT was wired - no labelling or documentation, so it's possible that might be wrong. Time to experiment.

[harry]

This transformer's from a DSL401/JTM310 I think. I don't know if those terminals were snapped off on-purpose or what... Is anyone familiar with this terminal arrangement?

clutching at straws here.

[R.G.]

A quick mini-lesson in tube drive.

The linear range of tube operation as far as grid voltage is concerned is from the cathode voltage down to cutoff. That is, the grid starts drawing current if the grid goes positive with respect to the cathode, and the tube is cut off entirely when the grid voltage gets that negative.

It is *possible* to drive the grid positive with respect to the cathode, but this takes a low impedance source drive, is very limited in how far you can drive it, and is usually heavily nonlinear.

It is *possible* to drive the grid more negative than the cutoff point, but this is very much beating a dead horse; the tube quits conducting at cutoff and won't get any more "off" if you make it more negative.

So the entire linear range of a tube's grid is between cathode voltage and cutoff. In an A-B output stage, this is very nearly twice the bias voltage on either tube. For a preamp tube, it's very loosely twice the cathode to grid voltage. This varies a lot more than for power stages, as the bias for a preamp tube is often more non-centered in the usable range than for power tubes.

For the purists: Yes, I know there are situations where this is not strictly correct. It's a rule of thumb, which is what to expect things to be close to, not what is always exactly correct.

For the normal people: sorry - I have coped with a lot of nit-picking purists.

[harry]

It's a Dagnall D2512 TXOP 00012

I emailed them and they sent me a datasheet that says this:

Customer Marshall Amplification Plc
Description Output Transformer
Customer Part No: TXOP-00012
Core: 25.4mm stack of 29 laminations
Bobbin: Single section, glass reinforced nylon thermal class "B"
(130oC) UL94V0 core tube thickness 0.9mm typical.
Windings: All winding wires to BSEN60317-4 (BS6811 Pt3 sect 3.1)
Polyurethane round copper wire; Grade 1 or better
Primary To suit 4.5K Ohms
Secondary To suit 0-8-16 Ohms@ 35W
Insulation 2 layers paper crepe tape between primary and secondary.
Terminations: 2.8x0.5mm push on tags
Finish Class "F" polyester varnish
Assembly Interleaved construction.
Electrical Tests: Off load performance
2000v flash test


which doesn't help

Shifting it to what I imagine is the 16 ohm tap gives maybe 13 watts into an 8-ohm load. Still not right...

I'm a bit afraid of trying that terminal that seems to be snapped off.. what could it be? Why is there a wire going to it?

Same exists on the primary...

[harry]

A very useful post R.G., thank you.

[harry]

Marshall® 30 W Output Transformer JTM30, JTM60, DSL401 TXOP-00012 - Etronic Shop, Amplifier & Guitar Parts

Found it! After hours of searching. Those terminals are cut off at the factory. So I'm fairly confident the OT is wired up correctly.

The PT gets a little hot... but heaters are at 6.3V....

Baffled.

[Old Tele man]

...if you have an idea of how much power (Po) you expect from the tubes (and there's sufficient plate voltage and plate current 'swings' to develop that power), and know the OT Zpp and the tubes' rated average transconductance (gm) value, here's the equation that can be backsolved to find the PI output signal (Vg.rms = half of Vgg) needed to produce that power:

Po = 0.9*(Zpp/4)*(gm*Vg.rms)^2

...so:

Vg.rms ~ (1/gm)*SQRT[(4*Po)/(0.9*Zpp)]

...it'll get you within ±10% of the actual design value.

[MWJB]

Post dc voltages throughout the amp.

"IIRC, PI has 340V through a 100K, resulting in about 160V on the plate." OK so what value power supply dropping resistors have you used in place of the 10K originals...I ask because your PI voltage is higher than 5D8, but your B+ is lower?

Output from both PI plates are comparable?

How have you rectified the B+?

Have you copied the 5D8 tone stack? If so lift the ground reference for the volume pot (post tone stack) and elminate series resistance by turning the pot full up, what happens?

[harry]

That is a very useful-looking formula, Old Tele man, thanks.

Is Zpp just the primary impedance?

I'm still experimenting with the power supply. I've just noticed that at the moment, there's a 330ohm dropping resistor leading to 1k screen resistors that I added. The PI comes from the same point.

The PI is getting 368V. The screens are getting 361V. I will sort this out and add an extra dropping resistor/capacitor stage, but I don't see why it would cause any problems?

It's rectified with 4xUF4007s and 150uF of capacitance.

After the PI, each subsequent stage gets 10k/32uF filters.

I didn't copy the 5d8 tone stack, but I will try your suggestion and see if it makes a difference.

The PI plates were very much unbalanced. One side (the inverted side) clipped very very early. In the mean time I've gone to a non-self-correcting version of the paraphase, which has evened things up a bit.

[Old Tele man]

Originally Posted by harry

Is Zpp just the primary impedance?

...yes, Zpp is the stipulated OT primary plate-to-plate impedance value.

FYI -- the equation is just an extension of the basic: P = R*I^2, with R being replace by 0.9*(Zpp/4) and I^2 being replaced by (gm*Vg.rms)^2

[harry]

Originally Posted by MWJB

Have you copied the 5D8 tone stack? If so lift the ground reference for the volume pot (post tone stack) and elminate series resistance by turning the pot full up, what happens?

I just tried this; it gives about 10W. So a slight increase.

[Old Tele man]

Originally Posted by R.G.

For the purists:

For the normal people: sorry - I have coped with a lot of nit-picking purists.

...not a problem for me, because ALL pepper tends to look like fly-shit to me (ha,ha)!

[harry]

Originally Posted by Old Tele man

...if you have an idea of how much power (Po) you expect from the tubes (and there's sufficient plate voltage and plate current 'swings' to develop that power), and know the OT Zpp and the tubes' rated average transconductance (gm) value, here's the equation that can be backsolved to find the PI output signal (Vg.rms = half of Vgg) needed to produce that power:

Po = 0.9*(Zpp/4)*(gm*Vg.rms)^2

...so:

Vg.rms ~ (1/gm)*SQRT[(4*Po)/(0.9*Zpp)]

...it'll get you within ±10% of the actual design value.

Going off:

Zpp = 4500 ohms
Po = 25W
gm = 9.5mA/V for 6L6s (what happens if I'm using two?)

gives 16.53V rms = 23.37V peak or about 46.75V peak to peak!


So the problem lies in the power stage?

[harry]

The impedance selector switch was wired incorrectly. I worked this out by measuring the RMS voltage on the OT primaries, and looking at the RMS voltage on the secondaries with a second meter.

As it is, 250V across primary gives 10.2V across secondary on 8 ohm tap (although my load is actually 7.5 ohms)

Assuming a primary of 4.5k as per the manufacturer's specs, this is correct.

This means the output is 13.9 watts

Still not there...

[Old Tele man]

...14W for 2 x 6L6GC is indeed low...Zpp = 4.5K should get you around 40-45W, Zpp = 5.6K is needed for 50-55W.

...what Vgg (volts peak-to-peak) do you measure at the PI plates? That's the drive signal going into each respective output tube, so should be around 47Vg.pk...what's the bias voltage at?

[R.G.]

Originally Posted by Old Tele man

...not a problem for me, because ALL pepper tends to look like fly-shit to me (ha,ha)!

Me too. I just hate it when I can dimly see the pepper crawling...

[harry]

Sorry - should have mentioned - it's cathode biased.

[harry]

25.5 Vrms at power tube grid...

so about 36V peak / 72V peak to peak


Actually the PI is unbalanced, so one side is only putting out 28.3V peak


Heater voltage was low - upping that gives me 16.7W at clipping (estimate - my preamp has too much gain for me to gauge properly) or 30W at full tilt...


I'll try balancing the preamp and see what happens...

[Old Tele man]

Originally Posted by harry

Sorry - should have mentioned - it's cathode biased.

...did you remember to subtract the cathode bias voltage from the plate voltage to get the actual "usable" plate voltage?

...that's one reason why a cathode-biased amp ALWAYS produces less output power than an otherwise identical fixed-bias version will--there's less available "usable" plate voltage.

[6267]

The front end has so much gain I'd make sure there's no weird clipping or oscillation coming in and disrupting output.

For kicks, try 16 and 4 ohms and see if that brings up the power.

I can get at least 16-17 volts across 8 ohms peak (not rms!!) with 6v6 in class B, so your 6L6's with that much drive should be able to do at least a 20v burst! Something is fishy. Check all solders and speaker connections. What's your heater voltage (i.e. is power transformer & PSU so overloaded it sags a ton??)

[harry]

What is the B+ on a 5D8?

By calculations, I'm dissipating 19W per tube.. which is about 63% of the maximum 30W.

Some amps bias EL84s at 100% when cathode-biased, right?

I heard it's not a good idea to do this with 6L6s.. but would upping it to 22W per tube dissipation be okay?

[Old Tele man]

...a real 6L6GC has an average plate power dissipation (Ppd) wattage rating of 30W, so 22W is 73% of it's working level...slightly high.

...with Class-AB1 operation it's advisable to keep the idle wattage at or below 70% of the tubes Ppd wattage, regardless of bias method.

...how did you calculate the idle wattage? did you remember to first subtract the cathode bias voltage from the measured plate voltage, ie: (Vp-Vk)?

[harry]

I cheated and used Weber's bias calculator

Weber Bias Calculator

But yes, I have been doing calculations using 363-25 = 338V from plate to cathode.

It sounds good as-is, so I might just keep it slightly cold.