Featured
Latest
Beam Tetrodes vs Pe...
 
Share:
Notifications
Clear all

Forum 141

Beam Tetrodes vs Pentodes

30 Posts
5 Users
0 Likes
5,854 Views
Dr Wobble
(@dr-wobble)
Posts: 702
Honorable Member Registered
Topic starter
 

I've followed your replies with interest. One question comes to mind; I'm curious as to why the pentode is used almost exclusively by guitar amp makers (ECC83 - EL84 - EL34 all Mullard/British) . Especially by Marshall when RCA/USA had comparable valves, but beam tetrodes. (KT range).

I know its not as simple as that-IE beam tetrodes = USA, Pentodes= Britain. Maybe its down to physical size.
Its a shame the evolution of the valve didn't continue; audio technology is prone to fashion-who could have predicted the recent popularity of valves? Wish I had, Andy.

Curiously curious

 
Posted : 02/07/2012 11:51 am
Anonymous
(@anonymous)
Posts: 16832
Group Deactivated Account
 

In technical terms, classic guitar amplifiers are a bit rubbishy, but the distortion and overload characteristics are thought to be a desirable part of the sound. The muddy tones of Orange amplifiers and all the rest of it.

http://en.wikipedia.org/wiki/EL34

Says that US makers tended to use BTs and the EL34 was driven to distortion more easily. I can see there'd be an audible difference in this type of application.

I do believe the Mullard EL34 was very well made and would stand up to use in extreme conditions a lot better than some of the modern counterparts although generally, I think that agonising over various particular makes and subtypes of valves is misguided.

Pete.

 
Posted : 02/07/2012 3:42 pm
Kalee20
(@kalee20)
Posts: 256
Reputable Member Registered
 

Anyway, your comment about the 6K8 reminded me that Langford-Smith had something to say about this, and that it was definitely “different”.

Certainly for frequency changers, most triode-hexodes (and triode-heptodes) arranged the oscillator injection to be on the outer grid, with RF on the inner grid (g1 = RF signal; g2 = positive screen-grid; g3 = oscillator injection; g4 = positive screen, usually commoned with g2; g5 = suppressor if it exists).

The 6K8 however reversed this, with oscillator injection on g1, RF signal on g3. The construction was a single cathode, a single grid, then one side of this assembly had another 3 grids plus an anode (forming the hexode), and the other side had just an anode (forming the triode). This of course dictated that the hexode section had the oscillator voltage on g1. I don't believe there was any beam-forming, critical-distance geometry, however, to eliminate the kink.

America did seem generally to like its pentagrid-type converters, with g2 being effectively a minimum-area 'grid' just sufficient to intercept enough electrons to sustain oscillation. And Europe liked them in battery valves - DK92 / DK96 (though the 2 volt TP25 is a triode-pentode with no screen-grid between g3 and anode). It seems to me that the 6K8 is a half-way house between these and a full triode-hexode. The 6A8 is an octode - a pentagrid with suppressor.

My own gut-feeling is that outer-grid injection is 'better' - the signal on g1 is acting on a steady electron stream which hasn't been modulated by a high-amplitude oscillator, so there is less oscillator radiation due to space-charge coupling. The presence of oscillations on g3 alters the distribution of current between anode and g2, but this happens on the outside of g2 so the signal on g1 doesn't 'know' anything about it. Of course, in the pentagrid-type of FC, this isn't possible. F Langford-Smith does have some observations.

There are definitely tubes with SAME model but some are "real" Pentodes and some Beam Tetrodes. Often different makes though.

Can you give some examples? All the DL92's I've seen - which includes Mullard ones - have been beam tetrodes. On the other hand, I've never seen a beam-tetrode EL84!

One question comes to mind;I,m curious as to why the pentode is used almost exclusively by guitar amp makers (ECC83 - EL84 - EL34 all Mullard/British) . Especially by Marshall when RCA/USA had comparable valves,but beam tetrodes. (KT range)

Dunno - http://vintagetvandradio.myfreeforum.org/viewtopic.php?f=3&t=2246 is an example of a guitar amplifier using BT's.

 
Posted : 03/07/2012 6:56 pm
Synchrodyne
(@synchrodyne)
Posts: 519
Honorable Member Registered
 

I have now had a chance to read through Langford-Smith’s chapter on frequency changing, and quite a complex subject it is too. The distillation seems to be that on balance, outer grid oscillator injection is preferred to inner grid, and that implies a separate oscillator. Still, one has the impression that the 6BE6, with its elliptical (in planview) grids 1 and 2 and grid 2 shields was probably about as good as it got with self-oscillating pentagrid converters. Whether the 6BA7 also had these features I do not know. As an aside, Fink shows colour television chroma synchronous demodulators using the 6BA7, but with signal on g1 and reference on g3, the opposite way around to an externally-driven pentagrid frequency changer. (And Carnt & Townsend show a similar circuit using the sharp cutoff 6BY6 dual-control heptode, originally developed, like the 6CS6 (aka EH90) for noise-gated sync separator applications, in which demodulated noise went to g1 and signal to g3.)

Re the Eddystone heptode choices, if one takes account of chronology, then there does seem to be a preference for the ECH81 in the later models, starting with the 888A, which used two. The 888A also appears to have introduced the EK90 (6BE6) as a self-oscillating SSB and CW demodulator. So one might reasonably infer that it was a case of “horses for courses”, with the ECH81 and EK90 each deployed to best advantage. The ECH81 was also used in the 830, 850 and 940 series.

Back to the original theme, something I came across recently was the fact that the UL84 and EL86 both had what Mullard called “shadowed” grids, which I think was its term for aligned grids. When I mentioned this on another forum recently, a well-informed contributor outpointed that the screen current-to-anode current ratio for these was a lot lower than for the EL84. So this looks like a case of Philips/Mullard borrowing a kinkless tetrode technique, perhaps to keep total cathode current within manageable bounds for what were low anode voltage 12 W pentodes in B9A envelopes.

They were two equivalent ways of doing the same thing, avoiding the tetrode kink caused by secondary emission from the anode. Philips/Mullard held the pentode patent and I think RCA developed the beam tetrode and held the patent.

I was under the impression that GEC-MOV developed the kinkless tetrode with aligned grids, etc., but assigned initial manufacturing rights to RCA because it though that the precision required was beyond its immediate capability.

Conversely some differently marked valves from SAME maker are actually the IDENTICAL tube not even selected on test (ECH81 for Mains Radio and ECH83 has 12.6V data sheet for car radio and they are the identical tube). But some tubes from different makers or the same maker at different time have different "insides" but same part number.

I wonder about the assumed identicality between the ECH81 and ECH83, though, even though that is the established wisdom. Before the ECH84 (sharp cutoff) became available, Philips used the ECH83, not ECH81 as a noise gated sync separator in some of its European TV receivers. This was a low anode voltage application, around 20 V, I think. Given that the ECH81 and ECH83 were equally available to Philips, one should reasonably impute a technical reason for choosing the ECH83, namely that it worked better with low anode voltages than the ECH81, which is what would be expected given its primary role as a car radio (12 V HT) frequency changer. From that comes the inference that the ECH83 was a selected ECH81. Now maybe most ECH81 examples met the ECH83 requirements, and the selection process was simply one of weeding out the minority that did not.

Another sidebar here; at the beginning of the UK dual-standard TV receiver era, Mullard positioned the ECH84 as a noise-gated sync separator, but the EH90 (which as the 6CS6 was originated for the same function) as a locked-oscillator FM demodulator (with signal on g1 and self-oscillator on g3). The ECH84 was also used in some Mullard prototype FM stereo decoder circuits as demodulator with signal on g1 and 38 kHz on g3. Such is – or was – the complex world of heptodes.

Back to pentodes again; the 6AK5/EF95 was one of the better pentodes for VHF work; I wonder did it have aligned grids, kinkless tetrode style, to help reduce partition noise. The dual-control 6AS6 was a derivative. In the UK, Mazda offered the dual-control 6F33, which seemed to be alone, with nonesuch from Mullard and GEC.

Cheers,

Steve

 
Posted : 19/08/2012 9:57 am
turretslug
(@turretslug)
Posts: 567
Honorable Member Registered
 

Mention of dual-control pentodes reminds me that ordinary pentodes were pressed into service as seperate electrode (presumably multiplicative?) mixers as well as single electrode (additive) mixers. E.g., the USAAF BC348's 6J7 (cathode LO injection) and Marconi CR150's EF50 (suppressor LO injection). In the case of the CR150, the suppressor is operated markedly negative of k/g1 with a high value cathode resistor and LO injection is markedly high level- presumably, the suppressor is of typically coarse pitch and thus struggles to influence the electron stream. It's noticeable that the quoted sensitivity on the highest (32-60 MHz) band is markedly lower than on the other bands, LO output is also much lower- an under-driven mixer is one with notably poor conversion conductance, perhaps?

Pentodes, again- the CR150 also combines "straight" (EF50) and variable-mu (6K7) valves in the AVC loop, the X66 2nd mixer and subsequent 6K7 IF amp get the full AVC volts whilst the two RF stage EF50s get around one-fifth (470k/2M2) of this voltage. I wonder how well things like gain distribution and signal handling fare over the full AVC excursion?

Colin.

 
Posted : 24/08/2012 1:26 am
Synchrodyne
(@synchrodyne)
Posts: 519
Honorable Member Registered
 

Very interesting subjects both, dual-control pentodes and AGC.

There is a good commentary on dual-control pentodes at: http://www.radiomuseum.org/forum/dual_c ... todes.html

I have read somewhere (I can’t remember where now) that the 6AS6 dual-control pentode had a much finer suppressor grid pitch than the 6AK5 from which it was derived, presumably to increase its influence on anode current. Even so, according to Fink, the 6AS6 required around 30 V p-p reference on the suppressor/second control grid when used as a colour TV colour subcarrier synchronous demodulator, about the same as the 6BA7 heptode required in its second control grid when used for the same purpose. With regular pentodes, I should surmise that an even bigger voltage swing was necessary to get a regular pentode suppressor to function as a modulating electrode. And that non-negligible negative bias was required to precent the suppressor grid from going too far positive and so putting a kink into the anode curve, tetrode-style.

And talking of the 6AK5, this was used by Eddystone in both mixer stages of its 880 series receivers, in both cases with cathode injection of the local oscillator. That I think would make it additive mixing, but I have never figured out quite what the advantages were over a heptode. Lower noise, perhaps, but then the quiet, high gain cascode RF stage might have rendered mixer noise contribution negligible.

Re the Mazda 6F33, it is curious that this appears in some of the transmitter coding schematics shown in Carnt & Townsend, Volume 1, amongst a sea of GEC valves such as the Z77 and D77!

As well as mixing and gating, and agc, Lowther found another use for output valve (e.g. EL34) suppressor grids by including them in the feedback circuit. It called this “Lowther Linear”, and I think presented it as a step beyond ultralinear. But I have never seen a detailed analysis as to how it actually worked.

A rather unusual variant was the American 6BU8 dual, dual-control pentode, albeit with common control and screen grids, the common g1 used for noise input. One pentode was used as a noise-gated sync separator (in much the same way as would be an EH90 or the heptode part of an ECH84), the other as a noise-gated and line-gated agc rectifier. My own odd way of looking at typical negative modulation TV agc valves that are anode fed with line flyback pulses is that they are effectively grid-controlled shunt rectifiers whose AC input (for rectification) is the line pulses. Conceptually this aligns them with customary radio practice which is to use a shunt rectifier in this case fed with IF signal as it input, and usually without grid control, and makes it easy to accept that the negative agc voltage comes from the anode end. I just find this easier than looking at them as pulsed amplifiers. Back to the TV case, the grid control voltage which varies the negative DC output comes from the video signal. The line pulses thus serve as both an AC source for the agc rectifier and as timing pulse, ensuring that the rectifier operates only during the back porch (or sync) period, which means that its gain is controlled by video signal level at that time, which is black (or sync) level. In the case of the 6BU8, there is the addition of noise gating as well.

As to agc voltage distribution, and even actual agc voltage ranges for radio receivers, I have not found too much that goes into detail here. But I do have a couple of Mullard TV agc articles that give some pointers. In one case the valve line-up is a mix of variable mu and sharp cutoff, namely PCC84, PCF80, EF85, EF80, EF80, with agc applied to the PCC84, the EF85 and the 1st EF80. The nominal maximum (magnitude) agc voltages are -10 V for the PCC84, with 3.5 V delay, -13.5 V for the EF85, and -5 V for the EF80. The EF80 agc voltage is simply proportioned from that of the EF85, so that it reaches -5 V when the EF85 reaches -13.5 V. Another, all-sharp cutoff line-up is given as PCC88, PCF80, EF80, EF80, EF80, with agc on the PCC88 and the first two EF80s. For all controlled valves the maximum (magnitude) agc voltage is -5 V. The PCC88 has a 2 V delay. But the point is made that with sharp cutoff valves, all must reach maximum agc voltage at the same time in order to avoid cross-modulation. So the PCC88 needs to reach -5 V at the same time as the EF80s, which requires quite a fancy resistor network, with -12 V agc available at its input to give the required outputs. I guess that the point is that once a sharp cutoff valve is pulled down to the bottom of the more-or-less linear part of its curve, any more agc voltage is highly undesirable because of the risk of cross-modulation, so reaching that point must be simultaneous for all controlled valves.
I imagine that the same considerations applied to radio receiver agc systems when variable mu and sharp cutoff valves are mixed. If there is a single agc line, then the sharp cutoff valve voltage will be a fraction of that for the variable mu valves, as you have said is the case for the CR150. In the radio case agc voltage magnitudes seems to be higher than in the above TV examples, maybe going to -20 or even -30 V in some cases, and some valves, such as the EF89, seem to be able to go a “long way south” as it were without getting into major non-linearities. So quite a low fractional voltage might be chosen for the sharp cutoff valves so that these didn’t spoil the range available from the variable mu valves. Again, this accords with your CR150 observations.

Valve era HF receivers with RF agc delayed with respect to IF agc do not seem to have been all that common, although the GEC BRT400 and the Marconi Atlanta come to mind as examples, both I think with all variable mu controlled valves. More common seems to have been a single agc line, with delay, feeding FC, FC and IF stages. So the extra complication of matching sharp cutoff valve agc magnitudes when there is a delay probably did not arise all that often. Sometimes, though, IF or final IF stage, even though variable mu, has limited agc voltage to avoid modulation rise distortion with large signals, which can reach a point where the signal swing goes beyond the linear parts of the curve as the valve is, so as to speak, chasing itself trying to generate an even bigger agc voltage at a point when the agc voltage increment per IF voltage increment is getting smaller. An example is the Quad AMII/AM3, which is designed for low distortion and has only something like a quarter of the agc voltage applied to the EBF89 IF stage as compared with the EF89 and ECH81 RF and FC stages. (The EBF89 is also neutralized to minimize passband tilt and the demodulator diode-to-pentode capacitance is also neutralized. John Collinson of Quad appears to have done a text book (Langford-Smith) job on this one.) Of course, with multiple IF stages, the final can be without agc without much compromising the agc range. On the other hand, simple LF/MF receivers with just FC and one IF available for agc probably need full agc on both to have an adequate range. The earlier (G. Horn-designed) Quad AM was like this, but the IF stage (EF93) had negative feedback (degeneration), presumably as an offset to any tendency to modulation rise.

A perhaps more elegant solution was that adopted by Chapman (and by Rogers for the short time that it offered an AM tuner), which was to have an AGC IF sidechain (EBF80 in the Chapman case) that provided amplified delayed agc, and allowed full agc voltage on the final IF stage without running into modulation rise, because the agc IF amplifier, fed from the final IF anode, was taking the strain. This also avoided differential distortion - which would otherwise require a third diode - although the usual palliative here for domestic receivers was to limit the agc delay to around 2 or 3 volts, so that differential artefacts occurred at relatively low signal levels, or with the advent of FM-AM receivers, abandonment of the delay for AM altogether. With HF communications receivers, distortion was likely less of an issue and I think that quite large agc delays were often used.

What I haven’t found, and which would help a lot I think, are worked examples of actual radio receiver agc systems, inclusive of graphs agc volts vs. RF signal input volts, overall and valve-by-valve, that cover a range of approaches. A quick wish list for such treatment would include: GEC BRT400, Marconi Atalanta, Eddystone 940, Chapman S6BS, Quad AM & AMII, Murphy TA160, Dynatron T139.

Cheers,

Steve

 
Posted : 24/08/2012 7:37 am
turretslug
(@turretslug)
Posts: 567
Honorable Member Registered
 

Erratum: I should have quoted the CR150's AVC division ratio as approximately one-sixth (470k/470k+2M2). It was obviously far too late in the evening for even simple maths!

I strongly suspect that you're spot on re. cathode injection resulting in additive mixing: I need to keep the term "wrt" a bit higher in my conciousness.

Not strictly a "beam tetrode" but the 7360 beam deflection mixer valve seems to have established a good reputation for both conversion gain and low noise, though it needed considerable drive- surely, the success of beam-guiding tecniques in the BT would have given encouragement to its designers? I believe that it featured in US (NTSC) colour TV modulators/demodulators, though by time CTV came to the market in the UK/Europe, HF signal transistors would have been thoroughly established and specialised chroma-handling ICs would have been at least a twinkle in their designer's eyes. I'm sure that a few 7360s would have figured in the PAL breadboarding and maybe even in the BBC's famed "405 NTSC" experimentation. I have encounterd early professional PAL modulators/demodulators that were awash with "1496"-type "transistor tree" modulators- again, perhaps, the 7360 would have figured in their designers' minds as a fore-runner.

Colin.

 
Posted : 24/08/2012 10:32 am
Anonymous
(@anonymous)
Posts: 16832
Group Deactivated Account
 

The 7360 is 1/2 a Gilbert cell (the Doubly Balanced mixer / analogue Multiplier Gilbert didn't invent but made more linear for Analogue multiplication and used today in almost every IC based radio and loads in every phone).

It's a true switch based multiplier like used to down convert for today's DSP based radio designs. It's not actually related to a Beam Tetrode or Pentode amplifier at all, but basically a single axis electrostatic deflection CRT with two anode targets.

As such it was used for SSB TX modulator to IF frequency (The RX IF used to block unwanted sideband on transmit as it will generate DSB-SC). On RX used as down converter to IF or for BFO to audio.

You'd need 8 of them though to produce or demodulate FM, AM or SSB without any other detectors and minimal filters!

 
Posted : 24/08/2012 10:57 am
turretslug
(@turretslug)
Posts: 567
Honorable Member Registered
 

I find that the Gilbert cell has an appealing elegance, every now and then when I'm sifting through the valve stash looking for something or other and I realise how many ECC82 and ECC91 I have, I'm tempted to make a rainy-day "thermionic 1496"! Years ago, I aquired a few dozen NOS complete PCBs for Perdio trannies that were being disposed of, I had the loony notion of making a "Ge1496" with a load of AF117s but rapidly realised that such a crackpot scheme would be doomed to failure by leakage and hopeless thermal tracking.
Not to mentiom tin-whisker MTBF concerns...

I'm also intrigued by the Russian rod pentodes, especially the trode-pentode mixer you concocted, I have a Pye handbag portable that's been neglected too long- I'm always inherently reluctant to xxxx around with quaint old things like this but it's not as if they're especially rare or exotic sets,

Colin.

 
Posted : 26/08/2012 11:27 pm
Synchrodyne
(@synchrodyne)
Posts: 519
Honorable Member Registered
 

Could one approximate the MC1496 function with a pair of 6BU8 valves? Signal input in antiphase to the respective common first grids of each valve, and switching signal in antiphase to the individual third grids in each case? A large enough switching signal would swing each pentode section from zero anode current to unimpaired anode between negative and positive excursions.

If you look at say a pair of ECC91 sitting on top of a single ECC91 and direct coupled, is not a 6BU8 approximately what you would get if all three were placed in the same bottle with the anode-to-cathode connections replaced by electron streams in vacuo.

Talking of beam deflection valves, Zenith originally proposed the use of the 6AR8A in (time division multiplex) FM stereo decoders. However, it seems to have used more conventional methods in its production decoders.

It would be interesting to know to what extent beam deflection valves were used in the 405-line NTSC colour television experimental work. Carnt & Townsend Volume 1 show example circuits from GEC, Murphy and RCA receivers, none of which use beam deflection valves for chrominance demodulation. On the transmitter coding side, the example circuits are mostly of GEC origin judging by the valve lines-up , but as mentioned, the Mazda 6F33 features where mixing and gating are required, although one can also find the Z77 with suppressor grid used as a second control grid.

Cheers,

Steve

 
Posted : 27/08/2012 7:30 am
Page 2 / 2
Share: