Triode vs Pentode

[neon indicator]

I've not taken much notice of valves for the last 40 years. So I'm a bit bemused reading about people "converting" or using EL34, KT66 or EL84 etc Audio amps in "triode" mode. Sometimes calling this erroneously "Ultra Linear".

Triodes have problems.

1. The characteristic is roughly a square law for output vs input, so generates harmonics
2. Less gain than equivalent Pentode
3. They are variable output impedance. Pentode output impedance above the "knee" is ideally infinite, so effectively the load impedance, Triodes are not "constant current" at a particular Vgk as Pentodes are.
4. RF performance limited by Anode/Grid capacitance, "Miller Capacitance" which is thus negative feedback at high frequencies.

Official 1p24b Triode curve


So the Tetrode was developed by adding a "screen" grid between the control grid and Anode. If this is "earthed" at AF/RF the RF performance is greatly improved. It also results in a constant current type output response, but with a "kink", an area of negative resistance just past the "knee". So the next development was the Pentode and Beam Tetrodes which remove the "kink". K in KT66 = Kinkless, i.e. Beam Tetrode. The Pentode adds the "suppression" grid (Kink Suppression I suppose!)

Official 1p24b Pentode curve

Note "knee" is the curve connecting upward slope from zero HT and zero Anode to flat part from 30V approx upwards.
Note G2 current is a maximum at Anode currents below the knee. On the Russian tubes the G2 current is very low as long as Anode volts are above the knee. On the Pentode Curves it's assumed G2 is NOT varied but at a fixed HT voltage (in this case G2 is a fixed 125V)

Disadvantages of Pentode (and Tetrode etc)

• Minimum Anode voltage limits output swing, the Knee of each graph line above. Raising G3 slightly can move the "knee" to a slightly lower voltage, but the Anode impedance likely drops.
• It's a constant Current source. But maybe you wanted that to drive a loudspeaker coil.
• It won't work at low voltages.
• The Screen grid can on some models and situations dissipate 1/2 the power.
• Poor gain at low load impedances
• G3 wiring or connection can cause oscillation if Cathode not at RF earth. Types with G3 internally connected to Cathode are more limited in RF application (can't be used in Common Grid Mode)
• Often needs higher HT for RF than Triode.

Advantages of Pentode/Tetrode

• It's a constant Current source.
• More gain than equivalent Triode at high load impedance. Gain is gm (Siemens or Mhos) x Load resistance.
• More linear, less distortion, less harmonics.
• DC or AC+DC Negative feedback by adjusting G2 voltage and "load" resistor.
• Ultralinear operation by feeding part of output into G2 (auxiliary windings on o/p transformer is one method)
• Extended RF response. A pair of Triodes connected in Cascode is equivalent. But this needs highest HT & Anode volts, thus lower Anode load and lower gain unless very high HT.

Triodising a Pentode.

There are several reasons to do this:

• Operation at HT volts below the Pentode "knee".
• Desire output swing closer to 0V, and distortion doesn't matter. i.e. More Power!
• Operation at low values of Anode load
• Want a "valve" sound or increased harmonics.

How is it done?
The simplest method and most common is to tie G2 to Anode instead of HT. This works best at HT volts above the "pentode" knee. This is really Pseudo Triode Mode. This allows highest o/p power. At voltages from start of "knee" to end of "knee" this mode gives highest sink current (g2 + anode currents)

A more advanced technique is to use the screen grid (g2) as Anode and load the Anode with a resistor from the g2 instead of to HT. Since on most valves the Screen Grid (G2) power dissipation is more limited you can't have so much power. But you get a bigger "triode" curve voltage swing.

The third mode only applies to "Space Charge" valves (very low electrode spacing) and low HT volts below the Pentode Knee. In this case the g2 is anode and the "real" Anode can be connected to 0V or used as a "gammatron" grid. Gain is very low, but current sink is better than in 2nd mode at HT voltages below the knee. Examination of the 1p24b Pentode curve shows g2 and anode current inverse relationship.

Gammatron mode (in whatever configuration) is using an electrode not between cathode and the electrode used as Anode to control current flow. The voltage gain might even be only 1/4! But you do get signal inversion and it's essentially an Electrostatic mode, so current gain can be millions. It will work with extremely low "anode" voltages and works best with very small electrode spacing. But even a dual Anode Diode rectifier can be demonstrated to have it. Since the spacing of some electrodes in the Pencil valves is only 700nm, the Gammatron modes work better than other valves.

Triode Mode Power
Also only the 1j29b and 1p24b appear to have "plate" like anodes. None have wire or mesh or gauze electrodes, all is rods. So the g2 current and load is likely limited like normal anode current to the maximum "Cathode" current, i.e. the failure mode is filament destruction.

Since the 1p24b has 2W Anode dissipation and 1W g2 dissipation, in simple Triode mode (g2 tied to anode) it then is only limited by Maximum Cathode current. So they have "official" Triode curves for 1p24b.

Triode curves (g2 and anode tied) are here http://www.techtir.ie/blogs/watty/miniature_valves_2 for 1p24b and 1j29b up to 45V. I will add curves for 1j18b, 1j24b and 1j37b all for HT up to 45V.

Sometimes you want gain. Then Pentode mode and HT above knee and as high a Anode load as voltage swing allows (smaller signals you can have higher load). Sometimes current drive is more important (output stages). Then you might even have less than 1 or even 1/10th voltage gain if HT is low and load is low impedance. Then "triode" mode might be better.

Effect of G2 on Pentode Operation.
On most designs you see g2 (screen grid) either connected to HT or with a resistor and a decoupling capacitor. As you increase the series resistor on g2, the voltage on g2 will vary inversely with Anode voltage due to inverse relationship of Ig2 and Ia below the knee. This is thus negative feedback and if the g2 is decoupled with a capacitor only affects DC. Without a capacitor the Audio or RF gain is reduced and signal is more linear also any hum or noise on supply rail has less effect on the load.

This where I need to test the "regular valves" such as EF86, EL86 or pentode in ECL82. Because on the Russian tubes I'm testing strange things occur as the g2 to HT resistor load is increased or the G2 is decoupled with a capacitor.

Up to about 1.5x the Anode load, the gain is indeed reduced and as a result the frequency response is better. Transient response is better. As Rg2HT is increased the gain of edges stays roughly the same as at the 1.5x load. But lower frequency gain continues to decrease. This means with transients there is severe "overshoot" About 50% when R on G2 to HT is about x3 or x4 the Anode load. So.. Add a decoupling capacitor? Well that gives terrible transient performance at larger signals! Maybe all pentodes behave like this? I dunno. Literature suggests up to x5 resistor for G2 load than Anode, but that doesn't work at all on these Russian tubes. Yet at 45V to 90V HT the screen (g2) current is much lower than many valves with normal Anode / Grid load/bias point.

I need to connect up my ECL82 and compare Anyone got any DF96, DL96, DKk96 or DAF96 and a B7A socket they want to donate for comparative testing?

[G8HQP Dave]

Triodes have lower distortion than an equivalent pentode, in most circuits. This is because of the feedback from the anode. The lower output impedance can also be an advantage, espeically when driving a loudspeaker. Remember, most speakers are designed for voltage drive and a triode is a better approximation to that than a pentode. Drive most speakers from a high impedance like a pentode and you will get a very lumpy frequency response, especially around the bass resonance. The force on the cone may be set by the voice coil current, but the displacement of the cone is affected by lots of other things too like effective mass and restoring forces from the suspension. We hear the displacement, not the force. A constant current source is only an advantage when you actually need a constant current source.

The generic triode characteristic is not square law but 3/2 law. Ultralinear and triode conection are two different things, although related. Most people know the difference.

[GrimJosef]

Originally Posted by neon indicator

I've not taken much notice of valves for the last 40 years. So I'm a bit bemused reading about people "converting" or using EL34, KT66 or EL84 etc Audio amps in "triode" mode.

It's not a new idea. Williamson used triode-connected KT66s in the output stage of his amp in 1947.

Some of the 'disadvantages' of tetrodes and pentodes can be drastically reduced using negative feedback of course.

Cheers,

GJ

[kalee20]

In fact, depending what you mean by the 'triode characteristic', you could even argue that triodes are essentially linear.

Although the Ia - Vg characteristic (for constant Va) is roughly a 3/2 law, the Ia -Va (for constant Vg) is 3/2 in the opposite sense. So µ is reasonably constant, and if you operate a triode into a load considerably greater than ra, distortion is pretty low.

For power amplifiers, this often isn't the case I admit, and moreover triodes have a limit as power amplifiers in that you usually can't drive them hard enough to get reasonable efficiency before running into the nuisance of grid current. But up to that point, distortion should be fairly low (and there is the benefit that what there is, is cancelled out in push-pull).

There are other benefits of triode vs pentode - lower noise, reduced transit time at high frequencies (because the anode can be brought much closer to the cathode), but although on-topic, these clearly weren't in the OP's mind when he started the thread.

[neon indicator]

Triodes are usually lower noise for RF, I've seen Cascode (2 triodes, 1st is common cathode, 2nd is common grid) described as "Pentode" equivalent. But it's not really and I think Pentode invented 1st. The Cascode is a good RF circuit and used today for very low noise ICs that work 100MHz to 6GHz.

For anything other than very small signals you need negative feedback. This can be a cathode resistor (that's not decoupled) on Triode or Pentode and various methods of wiring HT, Anode, G2, Transformer etc on Pentode.

I thought it curious that there appears to be no triodes in the Russian Sub-miniature Pencil valve / Rod Electrode series. Though they do appear to have used the 1p24b Power Pentode as "Triode". The very small dimensions do give very short transit times and they appear to handle a substantial percentage of "normal" Anode power (Pentode mode) on g2 (in Triode Mode). As to what the noise figures are? I don't know. I have a 18dB gain < 0.5dB NF 100MHz to 2GHz amp and a 100MHz to 1GHz noise source. I suppose I could test at about 110MHz to 140MHz and see what attenuator setting on noise source makes 6dB increase on amplified o/p by 18dB of DUT compared with a 50 Ohm termination at source.

What exact wiring is "Willamson's Triode Connected KT66"? Simply tying Anode and G2, using G2 as main output and resistor from G2 to " real Anode" or what?

I think people originally thought a constant current source for a voice coil was a "good idea". Later in the Bipolar/Mosfet eras it was regarded as "good" to drive from very low impedance.

As soon as you add negative feedback on Pentode you don't have a constant current source.

(BTW I wondered what happened to the initial post of this thread as my musing was in context of the Sub-miniature Russian Rod Electrode Pentodes.).

The Sub-miniature Russian Rod Electrode Pentodes obviously can give either a Pentode or Triode set of curves (as shown in official data sheet in 1st post). But it's doubtful that they are "really" Pentodes. The electrode spacing means that "space charge" effects must be significant at filament and G1. The "grids" are each multiple rods more like a number apertures that are VERY long compared to inter-rod spacing of same electrode and other electrodes. The filament -> G1 rods -> G2 rods -> G3 rods -> Anode pair of plates or rods on opposite sides don't even appear to be in line. Also some have a "shield" or "screen" wire in addition to G3 wire. There is nothing visible in the tube that looks other than a rod!

Photos

1. 1j24b is 12mA Heater @ 1.2V, 60V rated, 8mm x 40mm (ex. top exit)
2. 1p24b is 200mA heater @ 1.2V, 160V rated 1W G2, 2W Anode, 10mm x 40mm (ex. top exit) (two views at right angles)
3. Older 1j17b
4. Strange dual g1 Pentode 1j37b. "anodes" could be mistaken as "Rods"

The "Anodes" on the 1j24b (lowest power member) are not really larger than the rods. Just flat & skinny instead of round.

The "Anodes" on the 1p24b (Insane 800W peak ratings in Datasheet, 2W continuous Anode dissipation + 1W G2! ) are obvious pair of strips.

Nothing resembling a "shield" or "screen" even though as well as G3 the 1j17b, 1j24b, 1j29b and 1p24b has the external connection as well as external G3. The 1j18b has only 4 wires on base. G2, G1 F+ and F-, though claims to have a G3 and Shield according to datasheet, connected to Filament (f+?).

So why no actual Triodes?

[John M0GLN]

Originally Posted by neon indicator

What exact wiring is "Willamson's Triode Connected KT66"? Simply tying Anode and G2, using G2 as main output and resistor from G2 to " real Anode" or what?

Just looked at the 'Williamson' circuit I have, and its a couple of 100ohm 1/2 watt resistors from each KT66 screen grid to each anode. If you want I can send you a pdf of the complete Wireless World article.

John

[neon indicator]

no need. Thanks. Yes, that makes perfect sense for a Power Output stage, as the Screen Grid (G2) as the "main" Anode would not handle the power.


This article suggests that when you do that the Screen grid should have the Anode load. More applicable to lower power "triodization" (EF86 as "triode" instead of Pentode). But on 160V to 300V Mains driven Radio or Amp why would you "triodize" a Pentode for RF/IF/AF amp/pre-amp rather than output stage. You'd use an ECC82 or ECC85?

http://www.radiomuseum.org/forum/1j2...nnections.html

FET version (replacing RF or AF pre-amp Triode with FET?)
http://www.radiomuseum.org/forum/the...te_triode.html


(More here... http://www.radiomuseum.org/forum/rus...d_id=200277#13 Too much to take in at one reading!)

[DangerMan]

The EF86 is specifically designed as a low noise pentode, and so has good internal screening, including of the control grid lead out, plus a heater designed to give low hum field when fed from AC. It is also constructed for low microphony.

When connected as a triode, all these attributes are retained but the partition noise that comes from pentode operation disappears, giving a very low noise audio triode indeed.

Pete

[neon indicator]

That would be the theory. So if a Triode is such a good idea why use a Pentode in the 1st place? Why not just use a Triode?

I found this interesting, but it's only at a simple level http://www.freewebs.com/valvewizard/pentode.html

But the Partition noise is related to 1/f so at IF and RF it doesn't matter. So if the dimensions are small enough (which they are on "pencil valve" Rod Pentodes) so that transition time is no issue what is advantage of Triodizing a Pentode or simply using a Pentode? Also a low spacing Triode for short transition time will be limited in gain and RF upper limit by Cag (miller capacitance) which is why the Tetrode and Pentode have a G2.

Some other writings on the subject
http://myweb.tiscali.co.uk/g8hqp/radio/vmixernoise.html (multiplicative, frequency changer mixing, not Audio desk additive mixing is what that's about).
http://en.wikipedia.org/wiki/Pentode
http://en.wikipedia.org/wiki/Triode

So...

• Triodes might be lower noise at Audio Bass frequencies
• Triodes lower output impedance but lower gain and efficiency compared to similar spec Power Pentode, when there is no negative feedback.
• Pentodes have more supply variation / hum on supply rejection (depending on G2 circuit)
• Triode mode vs Pentode mode for the same valve allows much lower HT (below the "knee" operation
• Above the knee (without negative feedback) Pentode is like constant current source. Voltage gain = gm x Rload only above the knee.
• Given enough HT volts, Pentode is much more gain.

I'm not trying to prove Triode (or Triode strapped pentode) is better or worse than a Pentode. Just trying to understand it all.

[DangerMan]

You could say the EF86 is a special case.
Mullard's given reason for using a pentode in the first place was that they wanted the extra gain it gave as well as its inherent microphony and hum reduction. This gave them chance to use heavy overall negative feedback in order to reduce distortion but still end up with designs that gave full output from about 200mV input.
It is noticeable that Leak went another way with the Stereo20 and its sisters, which used triodes in the voltage amplifier and yet had much the same sensitivity and distortion., and no doubt if all EF86's other attributes had been available in a triode, or twin triode, it would probably have found use just for its screening.

As an aside, a common minimal-changes modification for the Mullard 5-10 is to convert the input stage to a triode connected EF86, which will reduce the open loop gain, then reduce the closed loop gain to end up with a less sensitive amplifier, but the same feedback as previously. (This is described in Morgan Jones)
The result is likely to be more appropriate to "modern" signal sources, and really is audibly quieter.
Increasing the overall feedback without reducing the gain would mean reconsidering the loop stability... hence the minimal-changes criterion above.
As it happens, similar gain alterations also can be usefully applied to the Stereo20 for use with today's preamps. (runs for cover..... )

It's pretty doubtful you could get enough gain from a triode of the era for use in a preamp with feedback-applied RIAA or tone controls set anywhere other than "flat". Not with decent distortion and hum/noise figures from AC heaters, anyway.

I'm not disagreeing, but I think that's why the EF86 was not your typical pentode !
(and probably 6BR7 too, to be absolutely fair )

Pete

[ricksteptoe]

As far as I am aware there is another factor in that triodes produce odd harmonics and pentodes produce even harmonics ( I think that is the right way round! ). The first harmonic of a musical note is one octave above, i.e. double the frequency, the next (even) is a fifth above that, the third is two octaves above the original, the sequence continues and gets more complicated. Anyway if you play the original note plus its first and third harmonics they blend together perfectly, whereas the original, second and fourth "beat" against each other.

In a hifi amp, or even a table radio the difference between triode and pentode is not only audible, it can change the way you think about a piece of music.

Pentodes are great for power, triodes give a purer tone, beam tetrodes are a good compromise at audio frequencies. At rf the harmonics don't matter as they are so far away from the wanted frequency they fall outside the passband of the circuit in the vast majority of cases.

Hope this is of some help,
regards,

Rick.

[murphyv310]

Think back to the EF37A, a classic Philips/Mullard design. It really was the fore runner of the EF86. Both are very good valves. When run as a pentode (as it was designed) partition noise is lower than overall noise in an ECC83 triode.

Where the little pencil valves fall down is the fact they are directly heated filament valves without a separate cathode. Due to this a very smooth and noise free DC supply is required whereas the EF86 is happy with an AC heater which is so well designed that hum ingress from the heater is I believe virtually impossible to measure.

[kalee20]

Triodes produce even harmonics, pentodes odd. But I think there's some confusion in the posts just above, as the first harmonic is mentioned.

The first harmonic is the fundamental; 2nd harmonic is 2x fundamental frequency; 4th harmonic is 4x fundamental, etc. The fundamental, 2nd, 4th do not beat against each other (although they don't sound particularly 'interesting' either).

The third harmonic is indeed an octave and a fifth (musically) and does sound 'interesting'. But add successive odd harmonics and you soon get unpleasant dissonance. (Even harmonics do the same, but by the time you get dissonance the frequencies are much higher and generally get lost, or are above upper hearing frequency range).

Where confusion may exist is that an alternative nomenclature of frequency multiples is to refer to them as overtones. The first overtone is 2 x fundamental frequency (ie 2nd harmonic); second overtone is third harmonic; etc. So odd and even change places depending whether you are referring to overtones or harmonics.

[neon indicator]

With RF you can easily filter out the 3rd and higher harmonics. The 2nd harmonic is trickier as that limits the Band for the low pass filter to less than 2:1. So for RF Push-Pull AND Pentode (or tetrode) helps with variable frequency transmitters. (180 degree push pull cancels 2nd (1st even) Harmonic if well matched. In RF amplifiers sometimes a 90 degree phase shift is used with Push Pull on narrow band MOSFET amplifiers (pentode like curves, thus higher in 3rd harmonic than 2nd). At 3rd harmonic the signals cancel as the two 3rd harmonic signals on the Push Pull output are in phase, as the 2nd harmonics are with a 180 shift at fundamental.

The Harmonics matter very very much at RF and are a PIG to filter out. Beam Tetrodes are essentially identical to Pentodes for most purposes. Not the same as "regular" Tetrodes, which have a Pentode like Ia/Va but with a negative "kink after the "knee". Didn't Philips market a particular Beam Tetrode as a Pentode?

Crystal Oscillator Overtone modes at higher frequencies confusingly are not exact harmonics of the fundamental as it's a different mode of vibration. Not to be confused with a combo oscillator / multiplier where the output is actually a harmonic of the real oscillating frequency of the crystal, which if oscillating at 100MHz, is likely an overtone mode rather than the mechanical fundamental frequency...

The Russian Pencil "rod electrode" valves that are 1.2V are of course intended for Battery powered (even if trickle charged from aircraft or truck generator) NiCd supplies. I believe there are indirect heater versions using 6.3V heater. The 1.2V direct filament valves are definitely needing very clean DC if not battery powered. I have no idea about the characteristics of the 2V family.

Where the "pencil valves" don't fall down is the low price and possible battery operation! Also no metal bashing!

[dominicbeesley]

Originally Posted by neon indicator

I thought it curious that there appears to be no triodes in the Russian Sub-miniature Pencil valve / Rod Electrode series. Though they do appear to have used the 1p24b Power Pentode as "Triode".
...
So why no actual Triodes?

Well they did make some 6V ones...I've just ordered some...will report back in a few weeks when they turn up.

I think you should read Morgan Jones's (link below), he has quite a good non-bs exploration of triodes vs pentodes for audio and when each should be used...with decent objective reasoning.

Filtering out harmonics at audio frequencies is a non-stater as most of us can hear 10 or 11 octaves. However, the human ear is a funny instrument. It does seem to be less offended by 2nd harmonic (and even orders) rather than odd harmonics. I've played around with this in audacity and it seems to be true...I seem to recall from physics at school that the ear introduces some 2nd order distortion anyway but that may be apocryphal!

http://books.google.co.uk/books?id=e...page&q&f=false

Dom

[neon indicator]

I know there are 6V ones (6.3V?). It's the 1 series (1.2V) pencil type Rod electrode I was thinking of.

Are the miniature 6 series "rod electrode" or "normal design"?

[G8HQP Dave]

Don't get harmonics and overtones mixed up. A harmonic is an exact integer multiple of the fundamental frequency. An overtone is not, whether we are talking about quartz crystals or musical instruments.

It is not true, but often stated, that triodes produce even order (2nd etc) and pentodes odd order (3rd etc.). Both produce both. It may be that some pentodes produce proportionately more odd than a similar triode, but second-order will usually dominate until clipping sets in.

At RF the harmonics might not matter, but the intermodulation (caused by the same nonlinearity) can be a serious limit to the performance of equipment. Odd order is particularly bad because it is in the same frequency range as the wanted signal so can't be removed with filters.

This thread is getting rather confusing, because it seems to contain a lot of (mostly eventually corrected) misinformation!

[ricksteptoe]

My apologies to Kalee and everyone else for the incorrect numbering, (as I said I was not sure of which way it went).

All I know is that, to me, triode amps sound the best, tetrodes next, and pentodes last, and to get a transistor amp to be as good as a valve one generally costs a lot of money.

I believe Pythagoras did the early research into harmonics and I am sure he could explain it much better than me.

Rick.

[neon indicator]

Nature of Internet. It's a wonder it's still on subject. Started as post on another thread though! Overtones are mechanical vibration patterns in a crystal or sound board or exceptionally a string. Common on drums (according to where you hit it). They are not usually direct integer related to fundamental as Harmonics are.

You can make Triode, Tetrode, Beam Tetrode, Pentode and Transistors amps that are not distinguishable. The thread isn't meant to argue one or other makes "better" audio amps. That's like OS-X vs Windows vs UNIX vs Linux.

My idea was to discuss and understand the differences for designing circuits, when to Triodise, when to use Pentode normally, when to use a real Triode, what the Issues are for low power RF, IF and AF, what the issues are for headphone, loudspeaker or R.F. P.A. Aerial driving...

I'm an ignoramus on valves.

[G8HQP Dave]

OK, broad brush --
small signal:
For audio and VHF upwards, use triodes (but for different reasons).
For RF use pentodes.

large signal:
it gets complicated!

When to wire a pentode as a triode:
when you want a triode but can't find/afford a suitable one