A question for all loadline experts (6L6 amp content)!

@tweedy_woodpecker

"What am I missing here?"

6L6's that match the original spec?

What 6L6's are you working with?

 

With no signal present (at idle) how would the tubes know what the transformer impedance was?

 

I think what @printer2 was trying to say is you need to adjust your load lines to lay on top of your bias point, instead of worrying why your bias point doesn’t fall on a design load line.

 

Just so we can rule out any mismeasurement or misunderstanding, hook your black voltmeter lead to the socket cathode terminal and then measure everything directly from the tube socket terminals. That way we don't have to do any extra math and can go straight to the data.

 

Yes! I still think your current runs a little less than could be expected from the charts at a Vg2 of 365, but it's pretty close!

 

You're welcome... honestly, I was going to back you into the 365V realization by having you measure it directly and then having the a-ha moment. But you figured it out.

Just by looking at the charts, voltages and currents you posted, I would have estimated a bias of about -33V. If you'd like to know how I make that estimation, I'll gladly share. But it sounds like you've got it figured out, so I won't bloviate.

 

Your bias point is a dc parameter. It is set by the voltages and currents running through the tube. It has no knowledge on what your loadlines are. Your impedance is set by your winding ratio and the speaker load. Change your speaker impedance and the load lines shift. Put an 8 ohm speaker on an 8 ohm tap then put the 8 ohm speaker on the four ohm tap and the bias point is the same. It doesn't care or know. Until you put a signal through the amp.

The plate characteristic lines are a guide. Tubes vary, they are all not going to match the curves in the datasheet, from the same manufacturer or another manufacturer. Or one made on Monday and one made on Friday. This is not rocket science, close enough for R&R comes to mind. 10% tolerance and they were doing great back in the day.

 

Here's how I work with plate characteristics when the charted Vg2 (screen voltage) doesn't match my design. This is the "back of the napkin" way. If you search around on the internet, you will find applications that do this and spit out the curves and numbers automatically and more precisely.

We need to generate some new plate characteristics for the Vg2 we are interested in. The thing to know is that when the screen voltage changes, the whole set plate characteristics curves expand and contract (up and down, in the y-axis direction) like an accordion. In your case, you have a screen voltage of 365V, so the curves are all going to contract down because they are originally drawn for 400V.

Think about what this means. It means that as we reduce the screen voltage, the anode current (Ia) for any given control grid voltage (Vg1) will be reduced. But by how much?

Enter the transfer characteristics chart. (Not shown on this chart is the fact that the anode voltage (Va) is a constant for each of these curves. It appears to be something like Va = 600V but I'm only guessing. That's not of any consequence at the moment, but I thought I would point it out since its missing.)

Take a moment to look at the shape and spacing of the transfer characteristics curves.



Don't they look very similar? They look like the same curve, just slid over left and right for different screen voltages (Vg2). What this means to us is this: when it comes to finding anode current, changing the screen voltage can be completely negated by a proportional change in control grid voltage, and it's very nearly a linear relationship.

For example, the conditions at Vg2 = 400 and Vg1 = -30 yield an Ia = 90mA. BUT... the same current (Ia) can be had at Vg2 = 350 at a Vg1 = -24. Changing the screen voltage by -50V while simultaneously changing Vg1 by +6V yields an identical Ia. And that relationship holds pretty close all over the transfer characteristics charts. There are some places where it's more like -50V:-5V, but our estimate holds pretty close. For instance, if we drop from 400V to 300V screens, we can negate that by increasing grid voltage by 12V. It's the same ratio. (-50:6 = -100:12)

All we need to do is just change the Vg1 voltage labels on the plate characteristics and we will have lines for the new Vg2 voltage. In our example we reduced Vg2 by 50V and showed that could be balanced out a 6V increase in control grid voltage. So let's just keep the same 400V Vg2 curves and just change all the Vg1 voltages by +6. That is what I have shown on the chart to the right.

You can see that the Vg1 = 0V curve has dropped quite a bit. This demonstrates the accordion like "contraction" of the plate characteristics as the screen voltage drops.

Now, your actual screen voltage was 365V, not 350. But the proportion -50V:6V still holds. We are only changing the screen volts by -35V so the equivalent control grid voltage change is +4V. (-50V:6V is roughly equal to -35V:4V). So your bias point, which is shown on the chart at Vg1 = -38V when the screen is 400V, is more like -34V when the screen is 365V.

I plugged the numbers into this calculator https://www.vtadiy.com/loadline-calculators/loadline-calculator/ and it came up with -33V for the bias. So either we're both wrong, or we're both pretty right.

I know I probably confused more than enlightened! Feel free to ask any questions, express comments and concerns this raises and I will try to answer to the best of my knowledge!

 

How does it sound?

 

Did a nice job cramming it in.

 

Just realized we never really addressed this question. Judging from the load line in my previous post, with a 4K plate-to-plate load impedance, you are hitting grid current (Vg1 = 0V) above the knee.

If you go with 8K (and hit grid current below the knee), you will probably get more output power and more compression when operating at full power. It might sound more “tubelike.” The catch is you would have to monitor screen dissipation and make sure your screen stopper is appropriately sized both in resistance and wattage. Even though plate dissipation will be lower, 8K might wear out your power tubes faster if the screens are being abused.

You said you have them on a switch? Can you detect an audible difference between the two? The difference would be most apparent when you are at or near full drive/in overdrive... and with 2x 6L6’s... that’s going to be loud!