Phase Inverter comparisons

[lowell]

So according Merlin at Valve Wizard the Cathodyne PI WITH a preceding gain stage has more gain and better balance that the Long Tail Pair. I've also heard and read from various sources that the LTP is better for driving larger power tubes. Now I'm wondering why one might choose one or the other? For tonal character? All in all I'm just curious what you all know about the two types of PI. Maybe some myths can be debunked/clarified.

[Wakculloch]

The gain thing is simple, though hard to put into words.
The split load has a gain of slightly less than the driver valve, ie about 45 with a 12AX7.
The LTP will have about the same gain, but only when measured across the anodes, so each phase referenced to 0v will have less than half this, ie about 22.

[MWJB]

The cathodyne inverter, if we are talking about the specific triode at which inversion takes place, actually drops gain. Hence cathodyne inverters in guitar amps have a preceeding triode, with little attenuation between them. This means that 250mVAC on the way into the first triode might come out at more like 7-15VAC feeding the power tube coupling cap.

The cathodyne inverter has a little more character.

Sometimes a cathodyne inverter will be followed by another gain stage, but we will overlook this for the moment.

A LTPI (assuming both circuits are using the same tube & also equipped with NFB) might typically turn 4v peak to peak into 30VAC feeding the power tube coupling cap. The LTPI has more headroom & less character.

Balance is adequate with both, if in good working order (a discrepancy in cathode & plate values in a cathodyne PI can be a problem), this is pretty well a moot point.

Because of the higher gain of the cathodyne & lower parts count, cathodyne inverters are usually found in smaller amps, with a lower number of preamp stages. For amps with more preamp stages a LTPI might be more suitable (but not a hard & fast rule). Type of power tube is irrelevant.

I'm not sure that there are really any myths to debunk?

[lowell]

The gain thing is simple, though hard to put into words.
The split load has a gain of slightly less than the driver valve, ie about 45 with a 12AX7.
The LTP will have about the same gain, but only when measured across the anodes, so each phase referenced to 0v will have less than half this, ie about 22.

I'm not sure I follow this. What do you mean the driver valve? Are you referring to a driver stage before the Cathodyne?

MWJB,
So the cathodyne with a preceding driver stage has a gain of 28-60 and the LTP has a gain of 7.5? So one might use the LTP for LESS drive, and the cathodyne for MORE drive? Finally you say the cathodyne imparts more character so one may or may not want that. Also, if you can put it into words how does the LTP have more headroom?

[Wakculloch]

To compare like with like, it's the gain that is created in a dual triode.
With the split load, the first triode is the driver, and creates the stage gain, whilst the second triode does the splitting and has no gain.
If one wanted to be pedantic one could also say that the actual triode doing the splitting has a gain of less than one and so a split load has no gain, but they are hardly ever used like that, so it's inpractical to make such comparisons.

[lowell]

So to be clear are you saying that the Cathodyne w/ gain stage has a gain of 45? Assuming they are DC coupled?

[Old Tele man]

...the "balance" of a triode cathodyne/split-load PI depends soley upon the accuracy of the "split" loads, ie: the plate and cathode resistors, and not upon any of the tube characteristics (gm, mu or rp). Thus, swapping tubes has no affects on the separate output levels.

...this is not so, however, with tetrode or pentode circuits, because of the differing plate and cathode currents, due to screen current!

[MWJB]

...the "balance" of a triode cathodyne/split-load PI depends soley upon the accuracy of the "split" loads, ie: the plate and cathode resistors, and not upon any of the tube characteristics (gm, mu or rp). Thus, swapping tubes has no affects on the separate output levels.

...this is not so, however, with tetrode or pentode circuits, because of the differing plate and cathode currents, due to screen current!

It's also not true in most cases because many (but not all e.g. BF/SF Prinetons with treble & bass controls) have the 2 triodes in the same tube, therefore a drop in gain will be noticed at the first triode.

[Wakculloch]

No, because it depends on the circuit conditions of the driver stage, and the value of the split load resistors, but it should be equal to the gain of the driver stage minus the losses from the split load stage.
I have no figures to hand but i beleive the losses are in single figure percentage terms.
I was using figures to illustrate the different gains of the different topologies.

[MWJB]

MWJB,
So the cathodyne with a preceding driver stage has a gain of 28-60 and the LTP has a gain of 7.5? So one might use the LTP for LESS drive, and the cathodyne for MORE drive? Finally you say the cathodyne imparts more character so one may or may not want that. Also, if you can put it into words how does the LTP have more headroom?[/QUOTE]

LTP has more headroom because it distorts less, it's more transparent, also LTPI can typically be used in higher B+ circuits (whereas the cathodyne might require singificant voltage drop after the screen supply), as voltage developed accross the tail (& therefore cathode) reduces 'actual' plate voltage in the LTP. A cathodyne PI usually has both triodes run form the same node and plate voltage at the first might therefore be limited (<200v?)...I guess you could add another node and run the 2nd triode at higher voltages with respect to ground?

Note that the circuits I compared were both circuits equipped with a 12AX7 and NFB, for the purposes of a reasonable comparison. So gain figures will vary some. Check out the schems at Fender.com.

[MWJB]

"No, because it depends on the circuit conditions of the driver stage, and the value of the split load resistors, but it should be equal to the gain of the driver stage minus the losses from the split load stage." The answer is "Yes" because substituting a 12AU7 will change the plate voltage, mu & bias, when compared to a 12AX7 in the same circuit - it will sound different due to the changes at the first triode, inversion will still be achieved. You typically have to consider the 2 triodes together, because that is how they are often wired. Swapping tubes makes a difference, I've done it enough times.

Oops! Sorry Wakculloch. I think I got the wrong end of the stick with your post, I see now that you were responding to Lowell's earlier post, not on tube subs in cathodyne PI. Apologies, Mark.

[Merlinb]

So the cathodyne with a preceding driver stage has a gain of 28-60 and the LTP has a gain of 7.5? So one might use the LTP for LESS drive, and the cathodyne for MORE drive? Finally you say the cathodyne imparts more character so one may or may not want that. Also, if you can put it into words how does the LTP have more headroom?

Each "half" of an LTP has half the gain of a regular gain stage, so a typical 12AX7 LTP has a gain of about 30 to each output. An LTP also has twice the headroom of an ordinary gain stage using the same type tube (bypassed cathode). Like any push-pull stage, an LTP produces more odd-harmonic distortion than single-ended stages.

A single cathodyne (JUST the PI, not the preceeding gain stage) has a gain of about 1 to each output, and has enormous headroom and very little distortion. However, combining it with a gain stage with a gain of 60 you now have a gain of 60 to each output. Twice what the LTP has! But your gain stage has only half the headroom of the LTP...
Also, most of the distortion now comes from your gain stage.

It is usually said that an LTP can deliver more output swing, so drive the power valves harder. This is true, but can be a bit misleading, because if you simply increase the load resistors on a cathodyne you can get pretty similar levels of drive to a typical LTP.

I think MTBs post might be a little misleading. Although an LTP can be used under higher voltages, a cathodyne can still be used up to the max value given on the datasheet, which is about 500V for a 12AX7 [EDIT: 550V, to satisfy any pedantic readers]. More than most of us ever use! (Both circuit may need heater elevation of course).

An LTP has the advantage that it is less prone to crapping out under overdrive conditions than the cathodyne, but with a little care both circuits can be made immune to this.

[MWJB]

I think MTBs post might be a little misleading. Although an LTP can be used under higher voltages, a cathodyne can still be used up to the max value given on the datasheet, which is about 500V for a 12AX7. More than most of us ever use! (Both circuit may need heater elevation of course).

500v for a 12AX7? Usually data sheets specify 300-330v (above the cathode) for most of the 12A#7 series. Typically, even these limits are not explored (just because they typically haven't been, doesn't mean they can't). But I take your point.

[Merlinb]

500v for a 12AX7? Usually data sheets specify 300-330v (above the cathode) for most of the 12A#7 series. Typically, even these limits are not explored (just because they typically haven't been, doesn't mean they can't). But I take your point.

Yeah 330V above the cathode, but the anode-cathode voltage in a cathodyne is no different from any other stage, so there are no particular reasons why the supply voltage has to be lower than normal.

[MWJB]

"Yeah 330V above the cathode, but the anode-cathode voltage in a cathodyne is no different from any other stage, so there are no particular reasons why the supply voltage has to be lower than normal." Indeed, but I'm thinking more with regard to the prior gain stage, in every amp I have seen this stage is fed from the same B+ supply node as the inverter stage, and 300-330v would be the limiting voltage for this stage (OK, plus whatever is deemed safe above data sheet recommendation, plus cathode voltage of a few volts). But as I said before, you could easily include an addidional B+ dropping node so that the inverter was fed from a higher B+ source than the preceding gain stage.