Phase inverters

[erichayes]

Hi All.

As threatened earlier, I'm posting a thread on the three basic styles of phase inverters. I'm going to be semi-technical and, unfortunately, will be devoid of visual props; you'll just have to imagine the circuits in your mind (kinda like listening to old tyme radio dramas), or refer to printed matter that I'll try to provide at the end of this first post. I would encourage specific questions and responses--I'm going to try to avoid math in this overview, but some queries can only be amswered mathematically.

There are four postulates that need to be recognized when dealing with PIs:

POSTULATE #1: COMMON CATHODE VOLTAGE AMPS INVERT PHASE 180°

POSTULATE #2: COMMON GRID AND PLATE VOLTAGE AMPS DON'T

POSTULATE #3: CAPACITORS SHIFT PHASE 90°

POSTULATE #4: PIs, ALMOST WITHOUT EXCEPTION, HAVE 2 STAGES

PI #1: Voltage divider. The first stage is a standard common cathode circuit with the plate being capacitively coupled to one of the output tubes' grid. However, there is a voltage divider consisting of two resistors that's also connected between the first stage plate and ground. The junction of the two resistors is connected to the grid of the second stage, also a common cathode circuit, and the values of the resistors are chosen so the signal voltage on the second stage grid is equal to the signal voltage on the first stage grid. Because the phase is inverted (see #1), the output of the second stage will be inverted again, thus putting the phase in its original form, which is then fed to the second output tube's grid. Simple.

The problem with this scheme is that coupling caps have to be used to isolate the various grids from the plate voltages of the previous stages. #3 says we're going to have serious phase issues. I've seen variations on this circuit to eliminate the coupling cap on the 2nd stage grid, but, overall, this approach to phase inversion doesn't lend itself to high fidelity.

PI #2: The Cathodyne phase splitter, AKA Concertina, Hot cathode, and other aliases. This circuit can be regarded as a half-assed cathode follower. The first stage has a large amount of gain, usually provided by a pentode, and is, again, usually directly coupled to a triode second stage. The plate and cathode resistors of the second stage are either of equal value (cap coupled) or of equivalent value (direct coupled) so that the signal voltage is equal, but out of of phase. The output tubes' grids are capacitively coupled to the cathode and plate of the phase splitter. Again, simple and quite effective.

The main problem with this scheme is getting enough gain in the first stage to overcome the inherent laziness of the splitter: the spltter wants to be a common cathode amp and will thus work one output tube as if it were a single ended stage. It's only after you almost overwhelm the splitter with grid drive that it will start behaving properly. I've read reports of DIYers having problems with parasitic oscillation using the Cathodyne circuit, but I attribute said problems to constructor ignorance.

PI #3: The long tail. When Ed and I were young'n's in the '50s, there was a fad known as "Mommy, Mommy" jokes. Sick, perverse humor at the time. The one that stuck with me was "Mommy, Mommy, I'm getting dizzy!//Shut up or I'll nail your other foot to the floor." Weellll, the long tail PI is basically a differential amplifier with one foot nailed to the floor. More specifically, one grid is nailed to ground, as far as AC signal is concerned. Again, we have a first stage coupled (usually directly) to one grid of a pair of triodes that have a common cathode resistor. The other grid of the pair is capacitively connected to ground, which clamps it to no signal voltage swing. So, if the cathodes are tied together, one grid's nailed to the floor, and the other grid's being fed signal, what does the plate of the grounded grid stage do? Well, hell, it has to follow the cathode current, but because the cathodes are tied together, it can only do the reciprocal of what the "normal" half of the PI is doing. Hence, phase inversion. Complex to the point of being mind boggling, but...

The long tail is, with the exception of a balanced differential amp, the phase inverter of choice, IMHO. The key to a good LTPI is that grid-grounding cap. If it's not prime grade, you're gonna have an inferior PI, and, by extension, an inferior amplifier.

Let the discussion begin...

[TerrySmith]

Will a LTP/CCS PI be too much for a EL84 amp? I currently have a Heath AA-151 on the bench with all circuitry cleaned out except the preamp section. I was thinking of trying a LTP circuit to replace the 6AN8 thats stock, And also do away with the 6AU6 stage and the tone controls and the stupid mode switch.

The reason I am keeping the stock case is I like the look of it, they are very well made, and I don't feel like drilling and punching a blank chassis!

Any ideas or SWAG's are welcome and encouraged!

[erichayes]

Hi All,

Terry, if you will Pmail me your snail mail address or fax number, I'll send you the schematic of my 1773 amp, which uses a 12AX7 driving a pair of 6GU7 LTPIs, which in turn drive two pairs of EL84s. This circuit has buckets of gain, so there's a good chance you'll be able to eliminate some of that upstream amplification. Although I haven't used CCS on this particular amp, I have used a discrete CCS on one of the prototypes Bud and I are working on, and it's killer. It's essentially the same PI, so I'd feel comfortable saying that CCS will work fine with the 1773 circuit.

[EWBrown]

The VTA-70 board or Shannon's ST70 board should work fine with EL84s, though some minor modifications to resistor values may have to be made.

I'm considering a PPP EL84 amp, using the 1773 OP trannies I bought a while ago, and using fixed rather than cathode biasing for the EL84s.
Driver / PI will be either of the two aforementioned boards.

The VTA-70 uses all 12AT7s or 6201s, the ST70 board uses one 12AX7 and two 12AU7s.

For power, I am considering using "dual mono, with one PA-774 for each channel and a separate small trannie for the VA/PI driver board.

The end result will essentially be two VTL "Tiny Triode" monoblocks, sharing the same chassis. Pentode mode should be around 40WPC, UL mode around 35WPC and pseudo-triode about 20-25WPC.

/ed B in NH

[tubes4hifi]

Eric,
thanks for a great and simple explanation of phase splitters. Well done!
Roy

[erichayes]

Hi All,

Here ya go, JB.

[EWBrown]

This, from Claudio Bonavolta's site, covers the basic (and a coupleof esoteric) phase splitters, complete with circuit doagrams, it's a good "go-with" with Eric's tutorial:

http://www.bonavolta.ch/hobby/en/audio/split.htm

Besides, I like his "banner" for his audio section:




/ed B in NH