Understanding Parallel Triode Load Lines - Reverb Driver

[stratomaster]

I have a rudimentary understanding of load lines as they apply to simple triodes.  I would like to understand how to analyze the parallel triodes used on the AB763 style reverb driver. 

My motivation for this is the Princeton Reverb circuit. I find the reverb node inadequately filtered in the stock form--where the reverb driver draws its voltage from the screen node in a no-choke filtering scheme. This leads to excess noise at higher reverb settings.  I'd like to move the transformer connection to a later node, but there's a significant voltage drop.  I want to understand how to find the optimal cathode resistor value to suit the lower voltage.

Has this been tackled formally anywhere that you can point me to?  If not, would someone please help me to understand how to draw parallel triodes load lines to try and match the performance of the stock circuit at a lower plate voltage?

I've read the Valve Wizard page on this subject, but it's not explained in any specific detail to the differences between a single and parallel triodes.  Also the example circuit is different enough to not be directly applicable.

Thanks in advance.

[tubeswell]

For a pair of equivalent parallel triodes, the current is doubled, so double the mA numbers on the y-axis (when drawing the load line)


A transformer driving tube doesn’t have (much of) a DC load* so you need more bias voltage to keep the current in check. The fender reverb driver load line is a Single-ended load line (see example here https://www.valvewizard.co.uk/se.html)


*but it still has an AC load

[tubeswell]

I've read the Valve Wizard page on this subject, but it's not explained in any specific detail to the differences between a single and parallel triodes.  Also the example circuit is different enough to not be directly applicable.

But everything is the same. (Except you use a 12AT7 chart with the y-axis values doubled). You plot the load line gradient using your selected xformer’s reflected load. Slide the line across (or up) to where it touches the Pmax curve and read the current you get for the selected B+ point where the load line intersects. The grid curves tell you what bias voltage to aim for. Ohms law tells you what size resistor you need.

[stratomaster]

Thanks. I was referring to the reverb driver blurb on parallel triodes.

I'll sit down with this and work through it. Appreciate the response.

[tubeswell]

Thanks. I was referring to the reverb driver blurb on parallel triodes.

Yes that’s the wrong page. You want the page for the SE output stage in my first post.

[HotBluePlates]

I have a rudimentary understanding of load lines as they apply to simple triodes.  I would like to understand how to analyze the parallel triodes used on the AB763 style reverb driver. ...

My preference is to recognize the curves available are drawn for a single-triode; so I convert/interpret the real-circuit values for what is seen by a single-triode.

   - If the load impedance is 22kΩ seen by parallel triodes, I convert this to a 44kΩ load to draw on the curves.
   - If the cathodes share a common cathode-bias resistor, I pretend its value is doubled.

   - Why?  The curves are drawn for a single-triode.  The loadline will be accurate if drawn at 2x actual load, because the load seen by a single-triode will then be reflected in the voltages & currents crossed by the loadline.
   - When moving back to "parallel-triodes" I can simply retain same-voltage-change (grid or plate) and double the current indicated in the curves.  This will yield current & voltgage changes for parallel-triode operation.



I prefer to do this stuff on a copy of the curves, taken from a data sheet.  You can use loadline calculators, but you have to payt attention to all variables presented, or they will graph a load & set of curves that don't apply to your case.

[Merlin]

Allow me; it's a four-step process. I'll assume these component values:

B+ = 420V
Bias resistor Rk = 2.2k
Transformer DC primary winding resistance = 2k
Transformer reflected primary impedance = 22k

Step 1: You're using two triodes in parallel, so double the vertical current axis;
Step 2: Draw a cathode load line corresponding to the bias resistor, 2.2k;
Step 3: Draw a DC load line corresponding to the total DC resistance in series with the tube, which is the bias resistor plus the transformer DC resistance, 2k + 2.2k. Where the two lines cross is the bias point.
Step 4: Draw an AC load line through the bias point, corresponding to the total impedance in the anode circuit, which is the reflected impedance plus the DC winding resistance (usually we ignore the DC resistance, but since you want all the detail) 2k + 22k.
Step 5: Profit.

Note: Remember that the horizontal axis only shows you the voltage across the valve itself. In this case the bias point shows about 405V. But there is also about 8.5V across the bias resistor, so the voltage you would measure anode-to-ground is 413V. The original Fender schematic shows 410V so I'm pretty close...

And then you need to remember the load is an inductor, highly reactive, so in reality the load line will trace out an ellipse aligned with the idealized straight line. In this case the bias point is so far to the right that the bottom of that ellips will likely smash into the horizontal axis, which explains why the Fender driver hits cut-off so easily; it has crappy current headroom.

[stratomaster]

In this case the bias point is so far to the right that the bottom of that ellips will likely smash into the horizontal axis, which explains why the Fender driver hits cut-off so easily; it has crappy current headroom.

This is awesome. Thanks for the detailed explanation. 

Your comment implies that reducing the cathode resistor would mitigate cutoff, but could put the tube outside of its max dissipation if you get too aggressive with it.

This explains the failure rate of modern 12AT7s in later Silverfaces that use a 680Ω Rk and no bypass cap.

I'll practice my parallel triode load line drawing to investigate a happy medium between the two.

Thanks again.

[Lectroid]

stratomaster,

I'm working through this stuff with a 1-tube reverb using 12DW7 and I'm interested if you've taken this any further.  Any further thoughts or observations?

[stratomaster]

stratomaster,

I'm working through this stuff with a 1-tube reverb using 12DW7 and I'm interested if you've taken this any further.  Any further thoughts or observations?

A 12DW7 wouldn't need a parallel triode load line. I'm assuming you're doing the driving with the low mu triode and recovery with the high mu triode? If so, you can just draw your lines on the 12AU7 graph.

I'm unsure what you're asking specifically in that context.

[Williamblake]

12AT7 is a weak reverb driver in most cases, two in parallel are a half as weak so you can add more load. I think of it this way: Has anyone ever put more than two 12AT7 triodes in parallel?