Here's a quick test of input section balance I've shared with Don and the team at Spatial:
Get a signal generator and feed 1 kHz to the direct XLR input, but with (+) and (-) phase getting the same signal. Use the normal input level of 0.5Vrms or preferably less ... this is not a test of clipping. Use a dummy load to protect the output transformer.
The output of the amplifier should be about 30 dB down from normal level due to cancellation in the input section. Keeping input level constant, sweep up to 40 kHz, noting level and looking for presence of distortion. If the faults that Zigzagflux found are present, it will be grossly obvious on the output ... sudden loss of cancellation, a sudden jump in level, and gross distortion.
Take it to the next step and play music instead of test tones. Again, it should be 30 dB down, but undistorted and neither tipped up nor bass heavy. If it is not, balance and cancellation is failing.
By listening to the error terms (failure to balance), that tells us a lot about that first stage. Is there a frequency dependence? Does one half have a different distortion spectra than the other side?
Zigzagflux, a quick question. Does the shared cathode resistor on the input 6SN7 benefit from a cathode bypass cap, or not? Any sonic difference?
Get a signal generator and feed 1 kHz to the direct XLR input, but with (+) and (-) phase getting the same signal. Use the normal input level of 0.5Vrms or preferably less ... this is not a test of clipping. Use a dummy load to protect the output transformer.
The output of the amplifier should be about 30 dB down from normal level due to cancellation in the input section. Keeping input level constant, sweep up to 40 kHz, noting level and looking for presence of distortion. If the faults that Zigzagflux found are present, it will be grossly obvious on the output ... sudden loss of cancellation, a sudden jump in level, and gross distortion.
Take it to the next step and play music instead of test tones. Again, it should be 30 dB down, but undistorted and neither tipped up nor bass heavy. If it is not, balance and cancellation is failing.
By listening to the error terms (failure to balance), that tells us a lot about that first stage. Is there a frequency dependence? Does one half have a different distortion spectra than the other side?
Zigzagflux, a quick question. Does the shared cathode resistor on the input 6SN7 benefit from a cathode bypass cap, or not? Any sonic difference?
I use the Fletcher-Munson curves, and the later Dolby noise perception curves, as an index telling me where distortion is most audible. Distortion below 50 Hz is very difficult to hear, but distortion in the 1 to 5 kHz band is very serious and easily audible.
An additional weighting has to be added for the parts of the musical spectrum that carry the most energy ... typically, this is around 300 Hz, with average energy tailing off above and below this.
A worst-case scenario is energy in the 300 Hz band going through an amplifier that has high-order distortion terms that fall in the critical 1 to 5 kHz band. The single greatest benefit of multi-amping is getting these high-order distortion terms out of the speaker driver.
An additional weighting has to be added for the parts of the musical spectrum that carry the most energy ... typically, this is around 300 Hz, with average energy tailing off above and below this.
A worst-case scenario is energy in the 300 Hz band going through an amplifier that has high-order distortion terms that fall in the critical 1 to 5 kHz band. The single greatest benefit of multi-amping is getting these high-order distortion terms out of the speaker driver.
What is interesting with Fletcher-Munson curves graph, is that statistically is correct, but individually may be not correct.
PS. I do not like statistics
PS. I do not like statistics
I will have to give that test a whirl next time I drag out the test equipment from the garage. Simple but comprehensive.
The capacitor was removed when I went from the IT and Western Electric harmonic equalizer to CCS fully differential. It never occurred to me to leave it in. With infinite CCS plate load and equal/hand matched driver stage load, I wonder what the capacitor would do? Maybe differences in each SN7 transconductance is the parameter to ponder. For that matter, maybe a string of LED's to establish bias instead. Worth trying.
Never thought of it that way, makes perfect sense.
Lynn,
Thanks for your statement:
"Does one half have a different distortion spectra than the other side?"
Now I can tell everybody why I do not like Paraphase Phase Splitters:
The input stage has intrinisc 2nd harmonic distortion; that output is attenuated and sent to the other stage of the paraphase phase splitter.
But, the other stage has its intrinsic 2nd harmonic distortion, which cancels the 2nd harmonic distortion of the input stage (serial stage cancellation).
Conclusion: the input stage sends 2nd harmonic distortion to the "push" output tube.
And . . . the other phase splitter stage sends no 2nd harmonic distortion to the "pull" output tube.
Paraphase Phase Splitters: Lots of gain, but un-balanced harmonic distortion.
Hmm, not the 50, 60, 100, or 120Hz type, just a thoughtful moment.
Is this "microphone" on?
Comments, please?
Thanks for your statement:
"Does one half have a different distortion spectra than the other side?"
Now I can tell everybody why I do not like Paraphase Phase Splitters:
The input stage has intrinisc 2nd harmonic distortion; that output is attenuated and sent to the other stage of the paraphase phase splitter.
But, the other stage has its intrinsic 2nd harmonic distortion, which cancels the 2nd harmonic distortion of the input stage (serial stage cancellation).
Conclusion: the input stage sends 2nd harmonic distortion to the "push" output tube.
And . . . the other phase splitter stage sends no 2nd harmonic distortion to the "pull" output tube.
Paraphase Phase Splitters: Lots of gain, but un-balanced harmonic distortion.
Hmm, not the 50, 60, 100, or 120Hz type, just a thoughtful moment.
Is this "microphone" on?
Comments, please?
Come to find out Steve Bench already looked into different bias arrangements (R/RC/diode)
https://diyaudioprojects.com/mirror/members.aol.com/sbench/dio_bias.html
https://diyaudioprojects.com/mirror/members.aol.com/sbench/dio_bias.html
A paraphase inverter is a standard inverting stage with the "non-inverting" side a type of plate follower with 100% feedback. It's basically an error amplifier, sensing differences between the paired resistors. It would have its own 2nd harmonic structure, not much like the other side, due to the 100% local feedback. I would guess its distortion would be several times lower than the inverting side, and not the same distribution.
Moving on, long-tail phase splitters are asymmetric as well. Redraw it a bit, and the inverting side is a normal plate-loaded amplifier, but the non-inverting side is a cathode follower driving a grounded-grid stage. At high frequencies, there is Miller capacitance on the inverting side, but the non-inverting side has almost none.
As for harmonic distribution, and more important, phase, that has to be measured to confirm. To cancel harmonic distortion, not only does the harmonic structure have to be the same, but the phase of the harmonics as well. Looking at it another way, the transfer curves have to be complements of each other. (The phase of the harmonics has a major effect on the shape of the transfer curve. For example, both square waves and triangle waves have the same distribution of harmonics ... the only difference is the phase of each harmonic.)
I am not sure any of the vacuum tube phase inverters has full cancellation of their own even-order harmonic distortion. I think a more significant difference is their ability to deliver linear current into the power-tube grid capacitance. Pentodes have negligible capacitance, ultralinear tubes have fractional Miller capacitance (no free lunch), and triode-connected tubes have the full Miller capacitance.
What does the distortion spectrum look like into that load? This consideration is usually washed out by the typical 20 dB of feedback, but it re-emerges with zero-feedback, hard-to-drive DHT power tubes.
Last edited:
As for Fletcher-Munson and individual differences, that region of peak sensitivity is created by ear-canal resonance, something we all have. Evolution favored that resonance to assist survival, giving peak audibility to spoken consonants and quiet forest sounds. Humans evolved in very quiet surroundings, far quieter than the world we live in now.
That ear-canal mechanical-acoustical resonance is also the reason that hearing is lost in the same frequency region. Deafness in old age, or earlier, is caused by repeated trauma to the hair cells in the cochlea, not by age. People who live in quiet tribal areas have nearly perfect hearing throughout life.
Our hearing is closely linked to emotional and survival centers, which is why music creates the emotional response it does. In addition, our long tribal evolution, going back at least 200,000 years, favored group bonding over a shared common meal, with story-telling and noise making/musical expression. This is why movies work: darkness, flickering lights, musical noises, and a good story will drop nearly all humans into a receptive quasi-hypnotic state. We are the descendants of the tribes that survived, the ones that were the most social, and had the best communication skills. As Ursula LeGuin put it, we are all shaped by the Lathe of Heaven.
(Less poetically, we are shaped by evolution, and our senses and emotional responses to them are shaped by whatever favored group survival over a very long period, and in very different environments of jungle, forest, savanna, coastlines, and several Ice Ages. We are not trapped by evolution; in reality, it gave each of us an extraordinary toolkit we can draw on any time we want.)
That ear-canal mechanical-acoustical resonance is also the reason that hearing is lost in the same frequency region. Deafness in old age, or earlier, is caused by repeated trauma to the hair cells in the cochlea, not by age. People who live in quiet tribal areas have nearly perfect hearing throughout life.
Our hearing is closely linked to emotional and survival centers, which is why music creates the emotional response it does. In addition, our long tribal evolution, going back at least 200,000 years, favored group bonding over a shared common meal, with story-telling and noise making/musical expression. This is why movies work: darkness, flickering lights, musical noises, and a good story will drop nearly all humans into a receptive quasi-hypnotic state. We are the descendants of the tribes that survived, the ones that were the most social, and had the best communication skills. As Ursula LeGuin put it, we are all shaped by the Lathe of Heaven.
(Less poetically, we are shaped by evolution, and our senses and emotional responses to them are shaped by whatever favored group survival over a very long period, and in very different environments of jungle, forest, savanna, coastlines, and several Ice Ages. We are not trapped by evolution; in reality, it gave each of us an extraordinary toolkit we can draw on any time we want.)
Last edited:
Lynn,
I look at it a little differently . . .
For an LTP phase splitter, the Miller effect, versus the signal driving impedance can cause a change of phase of the input grid at higher and higher frequencies; True, OK.
But if the input tube grid has a change of phase versus the input signal; then the input tube cathode [following] also tends to have very nearly the same change of phase.
Now consider the direct connection from cathode to cathode, so that both cathodes have the same phase.
If the grid currents are essentially insignificant, then the cathode currents are divided between the plate currents, and all is well.
Except for some lower order effect, I do not see a phase problem with this phase inverter.
Can you list one?
I can not think of any vacuum tube phase splitters that are more intrinsically balanced than the LTP phase splitter (and the very low gain concertina).
I am sure there is one, but I have not yet analyzed any circuit that has got my attention so as to make me favor it instead.
Unfortunately, the concertina phase splitter does have a problem, high frequency leakage from cathode to filament, and noise leakage of AC filament to cathode.
Perhaps there is a very complex, super compensated vacuum tube phase splitter out there.
But I generally subscribe to: "You should make things as simple as possible, but no simpler" - Albert Einstein
The LTP with a real good Constant Current Sink, is still my personal preference.
I look at it a little differently . . .
For an LTP phase splitter, the Miller effect, versus the signal driving impedance can cause a change of phase of the input grid at higher and higher frequencies; True, OK.
But if the input tube grid has a change of phase versus the input signal; then the input tube cathode [following] also tends to have very nearly the same change of phase.
Now consider the direct connection from cathode to cathode, so that both cathodes have the same phase.
If the grid currents are essentially insignificant, then the cathode currents are divided between the plate currents, and all is well.
Except for some lower order effect, I do not see a phase problem with this phase inverter.
Can you list one?
I can not think of any vacuum tube phase splitters that are more intrinsically balanced than the LTP phase splitter (and the very low gain concertina).
I am sure there is one, but I have not yet analyzed any circuit that has got my attention so as to make me favor it instead.
Unfortunately, the concertina phase splitter does have a problem, high frequency leakage from cathode to filament, and noise leakage of AC filament to cathode.
Perhaps there is a very complex, super compensated vacuum tube phase splitter out there.
But I generally subscribe to: "You should make things as simple as possible, but no simpler" - Albert Einstein
The LTP with a real good Constant Current Sink, is still my personal preference.
Lynn,
How many Paraphase splitters use capacitors in the divider network that drives the out-of-phase tube.
Is the miller effect thereby compensated properly, usually not.
I never got my paraphase splitter to work properly at high frequencies, too much phase shift.
Going for lots of gain is not the only criteria.
Even at mid frequencies, the input tube has perhaps 0.5% 2nd harmonic distortion, while the other phase tube cancels to near 0% 2nd harmonic distortion.
I can see why many designers use expensive phase splitter transformers, instead of vacuum tube phase splitters.
A totally different direction . . .
About 4 years ago I purchased a CD player that has balanced XLR outputs.
Then recently, I finally built 2 mono-block balanced amplifiers.
The input grids use 7 resistors including the Shunt volume control (about a 3 dB loss of input signal, but that is easily fixed, and soon).
The input triodes use a common self bias resistor, un-bypassed.
The extremely well matched output triode wired beam power tubes use a common self bias resistor, also un-bypassed.
And what about the rest of the topology of my balanced amplifiers . . .
Negative feedback in my balanced amplifiers? . . . well sort of, input tubes un-bypassed self bias resistor; output tubes un-bypassed self bias resistor.
The 100 Ohm screen to plate triode wired connection.
Then finally there is . . .
The output plate to output plate negative feedback (If you consider that one tube changes current more than the other tube, then the triode wired mode plate rp, tends to correct the other tube (I know, nobody looks at it this way; please just consider it. A smart person's simulation program can prove it, just consider un-equal rp, or un-equal Gm, or un-equal u, and put that into the simulation).
Think of one tube's extra current [error] pulling down too far, in push pull, that makes the other tube's plate voltage go too high, so in increases its current essentially equally (Class A region).
Go ahead, simulation fans!
How many Paraphase splitters use capacitors in the divider network that drives the out-of-phase tube.
Is the miller effect thereby compensated properly, usually not.
I never got my paraphase splitter to work properly at high frequencies, too much phase shift.
Going for lots of gain is not the only criteria.
Even at mid frequencies, the input tube has perhaps 0.5% 2nd harmonic distortion, while the other phase tube cancels to near 0% 2nd harmonic distortion.
I can see why many designers use expensive phase splitter transformers, instead of vacuum tube phase splitters.
A totally different direction . . .
About 4 years ago I purchased a CD player that has balanced XLR outputs.
Then recently, I finally built 2 mono-block balanced amplifiers.
The input grids use 7 resistors including the Shunt volume control (about a 3 dB loss of input signal, but that is easily fixed, and soon).
The input triodes use a common self bias resistor, un-bypassed.
The extremely well matched output triode wired beam power tubes use a common self bias resistor, also un-bypassed.
And what about the rest of the topology of my balanced amplifiers . . .
Negative feedback in my balanced amplifiers? . . . well sort of, input tubes un-bypassed self bias resistor; output tubes un-bypassed self bias resistor.
The 100 Ohm screen to plate triode wired connection.
Then finally there is . . .
The output plate to output plate negative feedback (If you consider that one tube changes current more than the other tube, then the triode wired mode plate rp, tends to correct the other tube (I know, nobody looks at it this way; please just consider it. A smart person's simulation program can prove it, just consider un-equal rp, or un-equal Gm, or un-equal u, and put that into the simulation).
Think of one tube's extra current [error] pulling down too far, in push pull, that makes the other tube's plate voltage go too high, so in increases its current essentially equally (Class A region).
Go ahead, simulation fans!
Last edited:
What about one phase goes through one active component and the other phase goes through two? The quote by Lynn below suggested just that.
I'm sticking with split-load/cathodyne/concertina or the best of all, input transformer (not necessarily with secondary center tap but using two matching resistors to ground).
Last edited:
directdriver,
1. Yes, a good phase splitter, the interstage transformer.
But the interstage phase splitter is not a tube.
The transformer does the phase splitting, not the tube(s).
You said:
"What about one phase goes through one active component and the other phase goes through two? The quote by Lynn below suggested just that."
I do not say any statement from Lynn that says that.
Would you explain, please, directdriver?
1. Yes, a good phase splitter, the interstage transformer.
But the interstage phase splitter is not a tube.
The transformer does the phase splitting, not the tube(s).
You said:
"What about one phase goes through one active component and the other phase goes through two? The quote by Lynn below suggested just that."
I do not say any statement from Lynn that says that.
Would you explain, please, directdriver?
Last edited:
I strongly disagree about the paraphase. Especially when sand, CCS' and other stuff like that were not an option, it was often the "best" solution.
The performance is really down to the actual implementation and requirements. I have found the ECC40 is a very good tube for this application and the performance reported in datasheet is more like a worst case scenario. Considering the distortion is relative to 30 V rms output on each side, it is quite good (see at pag 327):
https://frank.pocnet.net/sheets/046/e/ECC40.pdf
I would guess the 6SN7 is a good tube too and 30 V rms would not be needed in the 3-stages 300B amp.
The performance is really down to the actual implementation and requirements. I have found the ECC40 is a very good tube for this application and the performance reported in datasheet is more like a worst case scenario. Considering the distortion is relative to 30 V rms output on each side, it is quite good (see at pag 327):
https://frank.pocnet.net/sheets/046/e/ECC40.pdf
I would guess the 6SN7 is a good tube too and 30 V rms would not be needed in the 3-stages 300B amp.
1, I said INPUT transformer, not interstage so therefore there is no tube phase splitter needed.
2, From his description of the LTP, it implies the one phase goes through one tube and the other phase goes through two tubes. The advantage of LPT over paraphase is that the two tubes can be direct coupled, from cathode to cathode. It is this reason I stay with the split-load/cathodyne/concertina circuit or using input transformer.
Lynn,
directdriver in Post # 51 quoted you.
directdriver in Post # 54 said:
You [Lynn] are said to be implying that:
"From his description of the LTP, it implies the one phase goes through one tube and the other phase goes through two tubes."
Huh?
I am not convinced you meant that.
Please clarify.
And, I am still wondering what is untrue about my statements in Post # 49:
"For an LTP phase splitter, the Miller effect, versus the signal driving impedance can cause a change of phase of the input grid at higher and higher frequencies; True, OK.
But if the input tube grid has a change of phase versus the input signal; then the input tube cathode [following] also tends to have very nearly the same change of phase.
Now consider the direct connection from cathode to cathode, so that both cathodes have the same phase.
If the grid currents are essentially insignificant, then the cathode currents are divided between the plate currents, and all is well."
and in Post # 49, Where I also said:
"Except for some lower order effect, I do not see a phase problem with this phase inverter. [edit: LTP phase inverter]
Can you list one?"
directdriver in Post # 51 quoted you.
directdriver in Post # 54 said:
You [Lynn] are said to be implying that:
"From his description of the LTP, it implies the one phase goes through one tube and the other phase goes through two tubes."
Huh?
I am not convinced you meant that.
Please clarify.
And, I am still wondering what is untrue about my statements in Post # 49:
"For an LTP phase splitter, the Miller effect, versus the signal driving impedance can cause a change of phase of the input grid at higher and higher frequencies; True, OK.
But if the input tube grid has a change of phase versus the input signal; then the input tube cathode [following] also tends to have very nearly the same change of phase.
Now consider the direct connection from cathode to cathode, so that both cathodes have the same phase.
If the grid currents are essentially insignificant, then the cathode currents are divided between the plate currents, and all is well."
and in Post # 49, Where I also said:
"Except for some lower order effect, I do not see a phase problem with this phase inverter. [edit: LTP phase inverter]
Can you list one?"
Last edited:
Let's leave Lynn out of this. All I am saying is that in long tail pair, one phase goes through one tube and other other phase goes through two tubes. That's it. I didn't say there is a problem. (Maybe the exit time is different? I don't know.) But I just prefer split-load/cathodyne/concertina for philosophical reason. I WANT both phases pass through the same number of tubes. Hey, everyone has their preferences when it comes to phase splitters!
directdriver,
There are 2 triodes in all of my CCS cathode coupled phase inverters (Constant Current Sinks).
There are 2 phases relative to each other, and only 2 phases.
How close they are 180 degrees out of phase, versus frequency is a complex discussion and analysis; but for any given frequency, there are only 2 phases.
One phase in one triode, and another phase in 2 more triodes, not in my phase invertors.
Where did that 3rd triode come from?
You added a third triode to my phase invertor?
It is not on any of my schematics.
By the way, I addressed Lynn's listing of problems of a LTP phase inverter, in one of my postings (he did not object to my answers to his listing of problems).
Post your schematic, please, so I can analyze it.
I would also like to see your concertina schematic, including the power supply, etc.,
and please tell me how you completely eliminated the following:
1. If AC filament heating is used,
how is there no hum transferred to the floating cathode?
how is any common mode noise on the AC filament power not transferred to the floating cathode?
2. How did you deal with the capacitive reactance from the filament to the cathode that is in parallel with the cathode load resistor . . .
(High frequency amplitude un-balance of the cathode and the plate outputs)?
According to the different loads of the cathode and plate (at high frequencies).
3. Is the cathodyne phase splitter inside of a global negative feedback loop?
Thanks!
Please, directdriver and post readers . . .
Do not get me wrong.
I defend cathodyne phase splitters, based on the following rules:
Any cathodyne phase splitter works great, as long as it is implemented properly, and that includes paying attention to details.
As always . . . Tradeoffs, tradeoffs, tradeoffs.
There are 2 triodes in all of my CCS cathode coupled phase inverters (Constant Current Sinks).
There are 2 phases relative to each other, and only 2 phases.
How close they are 180 degrees out of phase, versus frequency is a complex discussion and analysis; but for any given frequency, there are only 2 phases.
One phase in one triode, and another phase in 2 more triodes, not in my phase invertors.
Where did that 3rd triode come from?
You added a third triode to my phase invertor?
It is not on any of my schematics.
By the way, I addressed Lynn's listing of problems of a LTP phase inverter, in one of my postings (he did not object to my answers to his listing of problems).
Post your schematic, please, so I can analyze it.
I would also like to see your concertina schematic, including the power supply, etc.,
and please tell me how you completely eliminated the following:
1. If AC filament heating is used,
how is there no hum transferred to the floating cathode?
how is any common mode noise on the AC filament power not transferred to the floating cathode?
2. How did you deal with the capacitive reactance from the filament to the cathode that is in parallel with the cathode load resistor . . .
(High frequency amplitude un-balance of the cathode and the plate outputs)?
According to the different loads of the cathode and plate (at high frequencies).
3. Is the cathodyne phase splitter inside of a global negative feedback loop?
Thanks!
Please, directdriver and post readers . . .
Do not get me wrong.
I defend cathodyne phase splitters, based on the following rules:
Any cathodyne phase splitter works great, as long as it is implemented properly, and that includes paying attention to details.
As always . . . Tradeoffs, tradeoffs, tradeoffs.
Last edited:
@6A3sUMMER I am not really sure any of the issues you mention about split-load are really relevant or important for audio. In the article below you can see measurements tell that your second point is quite irrelevant and back in the days people used to build cathodyne splitters that worked up to 3 MHz! The only real downside of the such splitter is the output voltage. I have to agree with such statement. Just don't drive 300Bs with a cathodyne splitter. Global feedback is not really an issue in 300B amps as most of the time there isn't any or very little, especially if these are PP amps.
45,
If by split load inverter, you mean the Concertina, I have a personal experience with an extremely bad one . . .
It had terrible noise, sounded like drill motor interference that the global negative did not fix, perhaps even made worse.
The amplifier was a commercial product, not my design.
Experience sometimes makes your preferences for you.
But even so, years later, I gave a Concertina another chance. It worked just fine.
But none of the amplifiers I use today have a Concertina phase splitter.
All good engineering designs work very well, if they are implemented properly.
Pick your preference of phase splitter, and then make it work properly.
What I do not really like is when someone says a particular kind of phase splitter is perfect (some are properly implemented, and some are not).
The bad ones of that phase inverter circuit give the good ones of that phase inverter circuit a bad name.
As I have previously said, some Paraphase splitters work very good, others not (some designs do not solve the problems of that configuration).
The transformer phase splitter can be good or bad. Quality of the transformer, and how it is driven, and how its output is loaded are the determinants of performance.
I have a preference for CCS cathode coupled phase inverters.
Yes, they only have 1/2 the gain and 1/2 the output swing of a Paraphase splitter (a generalization).
Lately, I prefer to use No phase splitter in my power amplifiers; I design and build balanced power amplifiers.
Instead, I let the CD player with differential XLR outputs "do the phase splitting".
If by split load inverter, you mean the Concertina, I have a personal experience with an extremely bad one . . .
It had terrible noise, sounded like drill motor interference that the global negative did not fix, perhaps even made worse.
The amplifier was a commercial product, not my design.
Experience sometimes makes your preferences for you.
But even so, years later, I gave a Concertina another chance. It worked just fine.
But none of the amplifiers I use today have a Concertina phase splitter.
All good engineering designs work very well, if they are implemented properly.
Pick your preference of phase splitter, and then make it work properly.
What I do not really like is when someone says a particular kind of phase splitter is perfect (some are properly implemented, and some are not).
The bad ones of that phase inverter circuit give the good ones of that phase inverter circuit a bad name.
As I have previously said, some Paraphase splitters work very good, others not (some designs do not solve the problems of that configuration).
The transformer phase splitter can be good or bad. Quality of the transformer, and how it is driven, and how its output is loaded are the determinants of performance.
I have a preference for CCS cathode coupled phase inverters.
Yes, they only have 1/2 the gain and 1/2 the output swing of a Paraphase splitter (a generalization).
Lately, I prefer to use No phase splitter in my power amplifiers; I design and build balanced power amplifiers.
Instead, I let the CD player with differential XLR outputs "do the phase splitting".
Last edited:
That's what it is known for, at least in the USA, as "split-load." That's why I included all the names into one lump: "split-load/cathodyne/concertina."
We all had bad experience with certain circuit that inhibits us from giving it a second chance. As they say, there's no second chance on first impression. I had good and bad impressions on all the phase splitter circuits but that doesn't stop me from have personal preference. There's nothing "wrong" with the other circuits either. I just like split-load for myself, period. This is purely subjective and I want to say that split-load/cathodyne/concertina is the only PP circuit that has the focus and imaging quality close to single ended amps. Don't ask me why! As for the criticism of no gain with split-load/cathodyne/concertina, then don't use it as a driver, use it in the front end or in the middle like Williamson amps.