LTP phase splitter as output stage?

[RollE2k]

I'm thinking about ideal for futuristiska project as dac and phono amplification stages, and starten wondering how to get balanced output without the need for OPT or extra amplification stage. So what i started thinking about is if it would be a dum idea making say a phono stage with stepup-d3a(se)-riaafilter-ltp phase splitter with low rp tubes (like 6n6p or even lower rp ones).
Will there be any trouble drivig the pre-amp there after (50k phase to phase impedance), is this a bad idea? If so, please explain

 

[petertub]

What is against it is that LTP is never totally symmetric, it needs NFB to even out ( or adjustable resistors which makes it a support nightmare). A catoodyn is symmetric, but might have problems with the load.
A lunddahl might be the best solution, and at the same time, show a swedish solution!

 

[RollE2k]

But with tail ccs and gyrator loading the output should(?) be "perfectly" symmetric? I already have a linestage with ll1676 as a 4:1+1 phase splitter at the input, but since the linestage is balanced it would be great to build a balanced source feeding it, plus i could go a bit lower gain on that source since it bypasses the input transformer and that balanced transfer is less sensitive.

Am I wrong or am I thinking right?

 

[petertub]

I'm not saying you are wrong, but it seems more complicated then a transformer...
By all means, try it, and please show us the outcome!

 

[JMFahey]

Why not use a dual Op Amp and have a perfect, stable, predictable phase inverter?

WAY more accurate (both in transfer function and frequency response) than any Tube or Transformer.

 

[petertub]

Why not use a dual Op Amp and have a perfect, stable, predictable phase inverter?

WAY more accurate (both in transfer function and frequency response) than any Tube or Transformer.

I did not hear this, was it some kind of noice on the line ??

 

[GoatGuy]

No, actually JMFahey was being practical… if it weren't for the absolute quixotic irrationality of people who insist for “all tubes” at all stages, if given a simple directive: “use the right parts for each stage, and try to use vacuum tubes for the outputs”, any competent designer would probably use FETs for the input, either op-amps or discrete high-gain semiconductor amps for the “middles” (including phase inversion), and sufficiently sensitive output tubes so that the voltage-swing limitations “of the middles” would still drive the output to full saturation on both sides of the signal swing.

As an example, I built a delightful full-range, well-powered amplifier about 25 years ago that had exactly the above design criteria.

It had matched J–112 JFETs (3 volt V_Gcutoff) for +3 dB input gain, unbypassed, degenerated source resistor…, which fed in turn another set of J–112 FETs in +3 dB gain on the anti-symmetric phase, to 'symmetrize' the front end triode-like gain. Used back to back 5 volt zeners to squash spikes and other input badness. Beautiful front end. From there, mixing both stages of output in a JFET input OpAmp, got another +20 dB gain. Being dual units, used the second one as the +0.0 dB inverter to get a the phase-inversion for driving a class A sextuplet of 6L6GTAs per channel. Ran the plates at 25 watts per, 150 watts for the sextuplet (per channel). Output power easily reached 50 watts per channel RMS. Excellent headroom for music-source dynamics.

The point? The point is … that JMFahey's advice is pretty keen. Even if because of design prejudice // bigotry // machismo … one chooses a nearly-all-tube design (valve inputs, valve middles, valve finals…) one can still drop in a near-perfect part that solves the problem at hand especially of one is going to talk about 'gyrators' and other exotics.

Anyway, petertub, I rather like your “ancient but excellent” solution of just using an interstage coupling transformer. Yes, the price(s) can be high for good ones. But they're also as quiet as mice, and as symmetric as one could ever want to have. Very few downsides except for cost.

GoatGuy

 

[Peleus]

Hi, guys,
About ten years ago, there is a I/V stage with LTP schematic as a part of DAC TDA1541 DIY kit in Taiwan. The website and the company, however, was gone and closed, and never reopen until these days. Fortunately, some diyers in Taiwan, Keep the schematics and discussed the kit these days. So, I think that it may help for your idea. Enjoy it!

 

[RollE2k]

But if we start to talk about opamps, then why are we even at the tube forum. i guess that is pretty much what petertub meant.

Anyway, i have read somewhere that people don't recommend ltp as splitter/driver, and why i raised this question was mere out of interest - because much ideas of how to build is based on old knowledge, and since we today often build with ccs and solid state plate loading, then i thought - why would it be a bad idea?

So, lets not turn this into a tube vs opamp/solid state thread, rather taking about the question i had - which was to get phase splitter AND output stage in one "stage", and with tubes.
The whole idea was to get less amplification stages/transformers in the signal and still have a tube option. If it is a bad idea, then just tell me why.

 

[GoatGuy]

But i fwe start to talk about opamps.. why are we even at the tube forum?

That's what I mean about irrationally prejudiced, friend. The irrational (but still quite understandable 'purist') notion that “if its not tubes, then we ought not to be talking about it”.

IT IS OK (both of our opinions). I think mixed mode conversations are just fine. I think mixed mode implementations are equally fine. I think without going “all crazy” that there are solutions having their own beneficial aspects, and sometimes there is a whole lot of work being done to work around other devices shortcomings.

Gyrators is one of those… somewhat classically, a semiconductor circuit to simulate a nearly perfect inductor, with the exception of having finite voltage swing. A lovely word that sneaks its way disingenuously into the whole “tube narrative”, as if in so doing it is 'normal' and 'not really a tube', and 'why should we care?'

Well… if you want to be really, really idealistic - then a gyrator circuit would have to be built from valves (it can be done. Not so easily, but it can.) Or, you kind of give up and use cascode (which is another great solution) for increasing linearity and isolating the variable V_AK that infests most nominal single-tube designs. Or you sneak in little semiconductors (as is popular these days) as cathode biases ("LED bias!") as if they're not really part of the signal chain, so why all the fuss? You can do the same thing with variable voltage regulators (LM317 and kin), and lash them into ''constant current'' configurations, which look surprisingly like ~gyrators~ after a fashion. But... wait... they're semiconductors, aren't they? And moreover, have a marked impact on the operating point, mu dominated gain curve and everything else of one's staunchly preferred tube types.

That's my point. If we're trying to be a purist about accuracy, then mixed-mode designs come to dominate. If we're less interested in that, and more about being Valve Purists, well … then a whole lot of work goes either into counteracting po' ol' tubes idiosyncracies, or in just living with them and calling the resulting distortions … that Sweet, Sweet 2nd Harmonic Tube Sound.

Moving on…
GoatGuy

 

[exeric]

Goatguy,
There is nothing wrong with having a specific philosophical orientation, such as most of the people here on the tube forum have. I think what is wrong is when individuals want to deny their own orientation to persuade others to join theirs. I think its obvious where your orientation lies with your signature of John Curl's golden rule. It seems a bit, (actually, more than a bit) hypocritical of you. I think it would be best to just accept that this isn't your crowd and leave it at that.

Rolle2k, as far as some disadvantages of LTPs I think you are onto something important. I think the "type" of distortion is very important to how a stage not only measures but how it sounds. An LTP stage has a lot of 3rd harmonic distortion compared to a cathodyne phase splitter. That's for the reasons you discussed. You can't really completely avoid distortion, even in a phase splitter, but a cathodyne has some advantages here.

Among the advantages, it has near perfect balance that depends only on the equality of both passive load resistors. When it does distort, which happen at overload into the next stage, then it causes primarily 2nd harmonic distortion. This is because there is a different effect of overload on the upper load resistance and the lower load resistance. This creates the 2nd harmonic distortion that sounds nice on overload. Finally, if you use two identical tubes in the voltage gain and cathodyne stages it will have twice the mu of the same two tubes in a LTP. That's cool.

I think the idea of using a Lundahl phase splitting transformer is a pretty good idea too. But to me it seems to me one is transferring the problems we are discussing to the manufacturer, Lundahl. I don't have any direct experience using that topology so I don't actually have any experience how effective they are.

 

[petertub]

Using transformers as phase splitters are as old as the tubes themselfs. Lundahl is
most likely the most knowledgeable maker of audio transformer on the planet today, thus
i would be confident that is a good solution, and a solution that uses a minimal of
parts for the task at hand. ( LTP or concertina with buffers would need 2-3 triode
equivalents).
This observation however does not preclude experimenting with other solutions, in fact
i would welcome other solutions and following test of it's properties. This is how
progress is made : someone tries a path that is deemed wrong, just to test. Who know,
maybe progress in other areas ( components? simulations?) has changed the tools
so much that a previously doomed solution is perfectly sane.

Go on! experiment!

 

[RollE2k]

petertub: Yes, i would rather use transformers for splitting rather than using cathode follower-buffer after the LTP. This because of the amount of triodes used - and the whole point with this thread was if it was possible to get a one-stage solution to work fine (which would mean one extra triode per channel) with help from ccs/gyrators. Also, the more different components you add, can mean you'll need additional psu and power transformers.

exeric: But isn't concertina alot more sensitive to loading than the LTP? Also, as far as i have understand (havent measured myself) balance/gain should be exactly equal between both halves if using a tail CCS for the LTP. I think it was SY, who demonstrated this earlier here in a thread, where he even tried to put two totally different triodes with tail CCS, and output was exactly equal.
Then of course depending on what typ of distortion there could be differences, because how we hear them.

Anyway, i think i'll try this later on when i have more time. Right now i don't have much time at all over for building - because we have gotten our second child, which also as the first one have Colic. So i try to spend the time thinking about what to try later on.