Tube Amps / Music Electronics
For current discussions, please visit Music Electronics Forum. The sunn still shines online!

ampage archive

Vintage threads from the first ten years

Search for:  Mode:  

Split-load vs. Longtail PI - Why/When one rather than the other?

7/20/1999 1:41 PM
Split-load vs. Longtail PI - Why/When one rather than the other?
Looking for information (in layman's terms) why/when these two different PI designs would be used. Is there a "rule of thumb" or is it strictly a matter of taste? Is one better than the other, sonically, tonally, etc.?  
7/20/1999 2:39 PM

The tonality issue doesn't come into play until the PI stage is overdriven (has its grid bias exceeded by incoming signal). There are noticeable differences, but I'm not any good at describing them.  
One main difference between the two types is the stage voltage gain. The split-load (also called cathodyne) inverter has no voltage gain. The cathode coupled long tail inverter has a voltage gain usually in the range of 10 to 25, depending on choice of tube and component values.  
Another difference is that if you don't need extra gain in your circuit, the split-load type uses only a single triode. The long tail requires two.  
Another difference is the resultant output voltage swing, what you need to drive a particular set of output tubes. The long-tailed type has a greater voltage swing capability before overloading, especially if the tail is connected to a negative supply (it's usually just returned to ground, such as in guitar amps).  
In the audio range, the simple split-load type is more linear in that it has lower phase distortion. At the high frequency end, there is a difference in output impedance between the plate signal and the cathode signal. This deficiency can be overcome/equalized with a simple driver stage between the PI's outputs and the driven power tubes. Look at the classic Williamson circuit.  
In guitar amps, it seems that any type of phase inverter gives satisfactory operation when driving the lower power tubes (6V6, EL84). For higher power amps, using tubes that require a greater signal swing (6L6, EL34), the long-tail is usually employed. Some bass amps go with the split-load followed by low impedance drivers or cathode followers to achieve large clean power levels (6550s).  
So there's some technical and practical reasons for choosing one or the other. You don't *need* a long tail to adequately drive EL84s. So why did Vox and Matchless use a long tailed inverter? I don't know for sure. One thing, you can take advantage of the signal gain, and not need as large a signal going into the PI to get sufficient output level. You don't *have* to overdrive every stage. Some folks don't like the sound of the PI stage clipping before the power tubes saturate. They use the long tail stage as a clean link between preamp/tone altering stages and power stage.
7/20/1999 3:30 PM
Carl B.
Hey, Doc,  
Does the cathodyne have lower output impedance than the long-tail (for guitar frequencies, anyway)?  
That would definately come into play when transitioning from AB1 to AB2.  
Thanks for any insight,  
- Carl B.  
7/20/1999 5:06 PM
I believe the output impedance is often considered one of the "deficiencies" of the split load driver, in that the impedance is high taken from the plate and low from the cathode. I'm still not totally sure how to describe what effect this has...  
I think it becomes most apparent when the driver goes into overload. The impedance imbalance contributes to assymetrical signal distortion.(I wish I could explain this better, Carl, but anything else I say I'd be pullin' it outta my...)  
7/20/1999 5:30 PM
Carl B.
Ah, I think I get it. Thanks!  
Doc wrote, "At the high frequency end, there is a difference in output impedance between the plate signal and the cathode signal." I took that to mean that the different impedance shows up at higher-than-guitar frequencies, but instead, he means that at high frequencies, the omni-present impedance imbalance starts to cause phase mis-match of the signal-split.  
That cathode has to be the low impedance point due to local negative feedback (no by-pass cap around on the cathode resistor causes that, a bypass cap would "kill" the signal on the cathode so you can't use one here), while the plate doesn't get a dose of nfb anyway (poor thing), and so stays at the higher impedance, even though the plate and cathode resistors are the same value.  
Hmmm. I wonder if a compensation nfb resistor/cap series pair from the plate to the grid could balance-out that cathode nfb (that pair would induce nfb from the plate to the grid), and balance the impedance of both?  
- Carl B.
7/20/1999 7:37 PM

I think you twisted a couple things. I did mean at higher than guitar frequencies. But yes, the cathode side has a lower impedance than the plate side. It doesn't mean much except at the frequency extremes. As you get up toward 20khz, with equal plate & cathode resistor sizes which are needed for equal output signal sizes, and equal size coupling caps, there is a slight phase shift in the output waveform between the two signals. And for high power applications, the cathode (low impedance) side is capable of driving the output grid with some current, but the plate side becomes unable to support grid current, allowing waveform distortion in one-half of the p-p setup.  
You really don't need perfect balance for pleasing tone in a guitar amp. I don't think that you would need to try to linearize the simple splitter. By adding another tube stage between each output tube and the inverter, the impedance mismatch differences from plate side to cathode side disappear(a major feature of Williamson's circuit). So now it's suitable for wide bandwidth or high power applications.  
The split load inverter really does have excellent signal balance within most of the audio range. It just wanders at the extremes. It's hard to get a long tail inverter to have, at the same time, perfect signal balance/ stage gain each side and equal plate operating voltages for both tube sections in the inverter. Also there's the matching of mu's for both sections of the tube. But the drive impedances are about equal for each side, from one end of the audio band to the other. With either type, better balance is achieved using low-mu tubes, like 6SN7 or 12AU7.  
There are some really good articles about the various PI circuits, their basic differences, along with their textbook equations, in back issues of Glass Audio magazine. Some are reprints of excellent articles from circa 1950 Wireless World magazines.  
Both inverters work well in general application. The split load was an earlier development than the Schmitt/differential/long tail type. (There are a few other types seen in audio amps.) The split-load is used successfully in many hifi amps, classic Dynaco, Fisher, Scott, and in guitar amps like Fender and Sunn. I think a princeton reverb sounds just fine, don't you? The long-tail was prominent in typical Mullard lab circuits, and can be seen in various high quality Eico amps. At one time, everybody was in a contest to produce an amp with more undistorted power and wider power bandwidth than the next guy. The long tail inverter came into vogue during those times. Most of the Fender higher power circuits, you know, the ones for the serious professional musician, employ the long tail type. Using low value load resistors (like the dreaded SF circuits - 47k) actually improves balance and large signal drive capability.  
You can't really come out and say, hands down, that one type is absolutely superior to the other type, unless you qualify the situation with particular circuit design goals.
7/20/1999 6:44 PM
J Epstein

Look at this link :  
It features a comparison (from a hifi amp designer's point of view) of various phase splitter options.  

  Page 1 of 1