• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

How well do LTP driver tubes using CCS need to be matched?

The current through each plate load resistor has to be the same, no? If not where does the unbalanced current go/come from?
Yes the current is the same, but the impedances might be different, so the voltage swing also could be different. Triodes have lower Rp than pentodes, and there is no guarantee that Rp is equal in the two halves of a double triode.
 
If the CCS is perfectly stiff, then the sum of the two outputs is also perfectly stiff. An LTP works like a split load inverter, and for the same reason, that there's nowhere else for current to go. If loads are equal, signal will be equal and opposite. LTP internal impedance does not matter, because it's effectively another CCS, measured across the two outputs.

Imagine "unrolling" the LTP and grounding one load resistor, and we have a split load inverter.

All good fortune,
Chris
 
  • Like
Reactions: rayma and jderimig
This a simplified vision assuming the tubes are good and relatively close , if not , the CCS will raise or drop dynamically the voltage across it to maintain regulation with the signal swing , but after a point if the tubes are too different will lose regulation . And even before that , it is not indiferent if the voltage is symetrical or not on the positive and negative cycle . The common cathodes voltage would drift .
With different tubes or one good and the other worn out , the output max voltage swing will be limited and the distortion increased
 
Last edited:
And it is known from ages 🙂 that when you build a transistor amplifier , the input differential ( LTP ) stage should be matched transistors for beta and so on , you don't use any crappy different parts you can find and the CCS will resolve everything
 
Yes the current is the same, but the impedances might be different, so the voltage swing also could be different. Triodes have lower Rp than pentodes, and there is no guarantee that Rp is equal in the two halves of a double triode.
True, but our voltage 'signal' to the next stage is based on the voltage drop said current drops across the plate load resistances, not the Rp. Let the Rp be different - doesn't matter. What is critical is having matched plate loads. Very easy to hand match those, and as the tubes age, the plate loads do not.
 
And it is known from ages 🙂 that when you build a transistor amplifier , the input differential ( LTP ) stage should be matched transistors for beta and so on , you don't use any crappy different parts you can find and the CCS will resolve everything
Transistors are, as it is also known from ages, quite another thing than tubes. As you mention beta/hfe, you're referring to BJT's. BJT's draw base current, which adds to collector current to form emitter current. Matching a pair as the LTP input primarily serves to minimize output offset voltage, but also minimizes the emitter to base current differences, hence indeed minimizes distortion.

Best regards!
 
I think if you connect a resistor between the anode and cathode (simulating Rp) you'll get a different ac voltage on the anode. Otherwise you could replace that tube with a resistor, or not?
Not sure what you mean, but assume the following:
Tube 1 is 12AU7
Tube 2 is 12AT7
gm and Rp and mu are different between the two tubes.

Plate load resistors are MATCHED and CCS in the tail.

For practical input voltages and bias conditions (i.e. don't drive it with 10VAC) AC output voltages to the next stage are MATCHED.
 
  • Like
Reactions: rayma and 6A3sUMMER
I think if you connect a resistor between the anode and cathode (simulating Rp) you'll get a different ac voltage on the anode. Otherwise you could replace that tube with a resistor, or not?
Yes, if the input tube has enough dynamic range and B+ is high enough and the CCS is stiff enough you could replace the other tube with a conductive carrot (as an earlier poster suggested) and the currents would match.
 
Not sure what you mean, but assume the following:
Tube 1 is 12AU7
Tube 2 is 12AT7
gm and Rp and mu are different between the two tubes.

Plate load resistors are MATCHED and CCS in the tail.

For practical input voltages and bias conditions (i.e. don't drive it with 10VAC) AC output voltages to the next stage are MATCHED.
Max AC output usable voltages are greatly reduced ( important advantage of tube LTP ) , or if you prefer , going beyond the point where the CCS has enough voltage to regulate , the output voltages will be not matched . Distortion is for sure much higher at any output voltage . For what would you want degraded performance ...

And yes , any active device is a controlled resistor , you can calculate or imagine for how long a CCS can maintain regulation using tubes with different Rp
The issue is the CCS available voltage , not how stiff or good it is
 
Last edited:
A two Triode CCS phase splitter:

There are 2, and only 2, current paths from the CCS: current of the two cathodes that have a direct wire connection from cathode to cathode.
Consider a CCS that is a perfect 4mA current Sink.

Then, unless in some weird condition, both cathodes are cut off at the same time (*), the perfect CCS will always Sink 4mA
Both tubes cut off, (*), is not a working condition, and not a real world condition of the phase splitter.

The current from each cathode can only go to 2 places:
The grid
The plate

Consider both the quiescent condition, and a limited signal voltage swing to one grid (the other grid with no signal voltage swing);
where neither grid has any current, during the whole limited signal voltage swing.
That means no cathode current goes to either grid.

Suppose the quiescent plate currents are within 20% of each other, one plate is at 2.1mA and the other plate is at 1.9mA (a total of 4mA).
Not a real good match, but good enough to make a real good phase splitter matched signal output (within the dynamic range of voltages that are left).
With perfectly matched plate load resistors, RL, the quiescent plate voltages will be 20% different (Example: 210V and 190V).

Apply a peak signal voltage to one grid which causes its peak plate signal to draw 0.5mA more than its quiescent current; and no peak signal voltage to the other grid:
2.1mA + 0.5mA = 2.6mA.
The CCS sinks 2.6mA to that cathode.
4mA - 2.6mA = 1.4mA
So the CCS also sinks 1.4mA to the other cathode.
We have "Conservation of Current" (I always count on that, but you might not believe it; so check your physics-major's physics book).

One plate current is Increasing by 0.5mA, so that plate's voltage goes Down by 0.5mA x RL
At the same time . . .
One plate current is Decreasing by 0.5mA, so that plate's voltage goes Up by 0.5mA x RL

The voltage changes of the two plates in opposite phase, and have Equal [Absolute] voltage changes versus signal voltage to the one grid.
[ ] (Absolute value)

if you can read through all of the above, then I challenge you to try to prove it wrong . . .
If it is wrong, then there is no such thing as "Conservation of Current".
 
Last edited:
  • Like
Reactions: Kay Pirinha
Depanatoru,

I said a Perfect CCS (unlimited voltage range, equal current at any voltage).

Current is Electron(s) in motion.

Electons do not normally disapear, yes there are exceptions:
Nuclear explosions, a black hole in the near vicinity of your vacuum tube amplifier, or a linear accelerator, etc.
Solid state hole theory (the electron leaves one spot that becomes a hole, it goes elsewhere to fill another hole, so it did not really disappear)
I leave it up to you to list the other things that cause electrons to 'dissapear'.

All of my vacuum tube amplifiers do not experience any of the above items and events.
Do your vaccum tube amplifiers experience missing electrons? Do they experience electron motion that does not cause a current?

All the vacuum tube amplifiers I know of have a limit to dynamic range; beyond that it becomes a switcher or some other device.
I am talking about phase inverters that are used within their dynamic voltage limit (this is Tubes / Valves thread(s) for Hi Fi Stereo amplifiers.
It is not the Instruments and Amps thread(s) of Guitar amplifiers.
Different rules for differnt objects.

How many amplifiers have you designed, built and measured, that have phase inverters that use dual triodes with hard wired parallel cathodes and a CCS?
 
Last edited:
Depanatoru,

You are correct!
The original post asked about using 2 Pentodes and a CCS to make a phase splitter.
That is much more complicated, as many of the posts in this thread discussed.

I would never use pentodes wired as pentodes in a 2 pentode + CCS phase splitter.
I will consider using the expensive EF806S Pentodes in Triode Wired mode.
If I have to go from a 2 stage amplifier to a 3 stage amplifier, in order to use a triode or triode wired pentode CCS phase splitter, I am willing to add a driver stage to get the gain and dynamic range I want or need.

I am talking about practical phase inverters, that Do have a limited dynamic range, but that work really well (well enough so that most people who design, build, own, and listen to them are not worried, so they can actually sleep at night).

I am still waiting to hear about your dual triode, cathode connected, CCS phase inverters.
Or if you prefer, with pentodes instead of triode.
 
Last edited:
The following illustration is about staying within the limited linear dynamic range:

IF the screens draw the same quiescent current, and draw the same [delta] signal current.
(Good luck with that).

Consider the division of CCS current from one cathode to the other:
If one cathode draws more current, its plate has more current. (Delta current #1)
If the other cathode draws less current, its plate has less current. (Delta current #2)

Total Current #1 + Total Current #2 + (Total Screen currents) = CCS current.

The major problem of different Gm of the two pentodes: it causes Early dynamic range limits (but there is still a small linear range, no matter how small).
Increasing the signal level beyond the linear dynamic range of the phase splitter always causes problems (understood and assumed by most designers).
"Man has got to know his limits" - Dirty Harry

Down with pentode wired pentodes in CCS phase splitters.
Up with Triode Wired pentodes in CCS phase splitters (equal signal voltage [amplitudes] with equal RLs).
 
Last edited:
  • Like
Reactions: jcalvarez