Featured VTs: The 6BQ6

 

The 6BQ6 as an audio final came to my attention as a chance result of finding an old article from a Portuguese (Brazilian? -- it was written in Portuguese) ham magazine. This described a simple AM plate modulator that claimed an output of 30W (fixed bias) or 25W (cathode bias) that used a push-pull pair of these VTs. You would expect to see 6L6s or 807s used in this particular application. Why 6BQ6s, and what were they?

The 6BQ6 is a large signal beam former. It has no audio pedigree whatsoever, and the spec sheet makes no mention of its use as an audio final. During the 1950s, screen sizes and deflection angles increased, giving a larger viewing area, and a shorter CRT for more compact TV sets. This development meant that the usual audio finals and RF types became increasingly unsatisfactory for horizontal deflection duty. New types more suited to the task were developed, and one such type was the 6BQ6, in several different iterations for large screen, B & W TV sets.

The 6BQ6, as with all HD types, is capable of pulling big currents through the horizontal deflection coils. This was accomplished by cathodes much larger than those found in comparable audio and RF finals, such as the 6L6-oids. Compare heater voltages and currents:

6V6: 6.3V / 0.45A (2.835W)
6L6: 6.3V / 0.9A (5.67W)
6BQ6: 6.3V / 1.2A (7.56W)

This makes possible a maximum cathode current spec of 400mA. In terms of solid state, that isn't very spectacular, but for a nominally high voltage, low current device, it's pretty good. Other design features include a low voltage capability, made possible by the close proximity of screen grids to the cathode, and that's what you want for drawing big currents through deflection coils, and for operation from transformerless power supplies that derive the working DC directly from the AC mains. The latter consideration has spawned versions of this tube with odd heater voltages for operation in series heater strings.

As an audio final, the high current and low voltage capabilities make for lower load impedances, and therefore, for easier to design OPTs. The lack of any audio final usage requires one to draw up loadlines to find suitable operating points. Here is one such audio loadline:

As you can see, the load is just 1K1 per phase. This is a Class AB loadline, so you would need an OPT with a primary that can match 4K4 (P-2-P) to your speaker load. It's is a good deal easier to manufacture an OPT with excellent high and low frequency performance than it would be for the 10K (P-2-P) load a PP pair of 6V6s need. It's also a convenient value since there are off-the-shelf OPTs that were designed for Class A, PP, 6L6s that match this value. The only drawback to using one of these OPTs is that they are a bit under powered for the 6BQ6.

As for why no audio uses are specced, the plate characteristic tells all. It is not possible to use the 6BQ6 in Class A. The most linear part of that characteristic lies well within red plate territory. You are committed to Class AB, push-pull only. That isn't necessarily a bad thing, though.

If you stick with the usual convention for determining the static plate bias current, you are deep into Class AB. The closer you come to Class B conditions, the more cross-over distortion you have. The estimated H₃= 5.0% is barely acceptable. There is something you can do about this. The 6BQ6 was specced very conservatively, since its main purpose is to output max RMS power for hours a day, all the while maintaining a reasonable service life. Audio amplification is a good deal less demanding. If you increase the bias, and bust the plate dissipation spec, the performance improves greatly.

Doubling the bias current to 50mA per plate, makes for 17.5W of static dissipation: exceeding the spec by 6.5W (the 6BQ6 is rated close to the 6V6). This is of no consequence, as you don't get any color on the plates. The RCA gray plates can handle up to 70mA, with just a trace of color. The Sylvania black plates aren't showing any color even at this extreme 24.5W of static plate dissipation. 50mA of bias won't harm this type, even if it's outside of the spec. The increased bias current, and the much lower H₃ estimate, does make a noticeable -- and improved -- sonic difference in practice.

As to the sonic performance of the type that had no audio pedigree, in a word: excellent. The 6BQ6 easily matches the performance of the legendary 6V6, and it pumps out more than twice the power. Running open loop, the 6BQ6 doesn't have any of that expected "pentode nastiness" until it's nearly at the point of clipping. There's just an "edginess" or "aggressiveness" to the sound. That would account for why the Portuguese (Brazilian?) hams elected to use it in their plate modulator project (it was an open loop design). 6L6s tend to sound much worse open loop.

The Twin-T test also bears this out. After nulling the fundamental, the residual is almost a perfect sine wave at three times the frequency. There is little higher order harmonic components. Though high fidelity isn't legal on the ham bands, the 6BQ6 plate modulator would indeed sound better. For excellent sonic performance, all that's required is enough gNFB to take off the edge. The 6BQ6, like the 6V6, doesn't require the additional help of local NFB. That is a good thing in that the low screen voltage spec pretty much eliminates Ultralinear. Operate them in full pentode mode with a regulated screen supply. As with any pent that can't run the plate and screen at the same potential, the best sonic performance results when the screen voltage is tightly regulated, and supplied from a source with a very low impedance, as is the case with an active regulator.

The 6BQ6 also seems to be a good deal more stable in operation. It doesn't produce the snivets that are seen with other types, even if you don't include plate and screen stoppers. Even if not strictly necessary, it does no harm to include these anyway. Screen stoppers can be 680Ω/0.5W C-comp resistors. Plate stoppers can be made from ten turns of #18 wire, space wound 7/16ths inch ID, with 100Ω/2W (or four 470Ω/0.5W C-comps in parallel) de-Qing resistors mounted inside the coil. That should take care of any possible instabilities, or tendencies to make RF, due to lay-out issues.

6BQ6-oids:

This type appeared early on in TV development. The first iterations were in ST glass. As TV screens got bigger, new versions appeared. The 6BQ6GTA has the small profile, tubular glass envelope. This type has smaller cathodes, and is not quite so robust as later iterations. As such, it will red plate badly if used at the Q-Points mentioned previously.

The 6BQ6GTB has the same small diameter envelope, but does work as described above. The 6BQ6GA has a larger bottle, but otherwise biases identically to the 6BQ6GTB. As for sonic performance, the 6BQ6GA may have a slight sonic edge, but I'm not 100% certain of that. Either version sounds great, and will stand up to the spec busting for long service life in audio amplification applications, which is what we're interested in here. Of course, audio amplification is not the same as horizontal deflection duty, or "brick on the key" RF modes (FM, Packet) either.

The 6BQ6 really is a tougher, higher powered, 6V6. The main difference is the greatly reduced screen voltage spec that precludes Ultralinear, and makes pseudotrioding problematic. As a pseudotriode, since the screen voltage never exceeds the plate voltage, you may be able to get away with running at higher than specced screen voltages without either poofing the screens or plate current run-away. That remains to be seen, if you're so inclined to experiment with it.

The advent of TV sets running transformerless power supplies has also led to the development of *BQ6s that have some odd heater voltages:

12BQ6: 12.6V / 0.6A
25BQ6: 25V / 0.3A

These types were intended for use in series strings, as there was no PTX to provide heater power. That such types exist is beneficial in case the 6.3V version should become harder to acquire, and/or the price becomes excessive. Five years ago, you could get 6BQ6s for under a dollar a pop. Due to inflation, and probably the publication of designs that use the type, the prices have been going up. Still, it's not like audiophool expensive, even for quality that's dubious, of finals that do have that audio pedigree. Availability should not be a problem, since the 6BQ6 was widely used in a great many brands of wide screen, B & W TV sets.

6AV5 and Other HD VTs:

This type is very close to the 6BQ6 in terms of specs and sonic performance. The main difference is that the 6AV5 lacks a top cap connection to the plate. This would be a plus for designers who're leery of plate top caps in audio amps that potentially expose end users to dangerous voltages.

If you can wind your own OPTs, or don't mind paying extra to have custom OPTs wound, other possibilities present themselves. As B & W screens increased in size, new types that could handle the increased current demand were developed. This would include types such as the 6CB5 and 6DQ5 -- both capable of even higher output power.

With the advent of color TV, the demands put upon horizontal deflection subsystems grew even greater: color TV CRTs operate at much higher voltages, require more cathode current, and also had the large screens and deflection angles. To meet these demands, some truly awesome power tubes were developed for increased current sourcing, and power handling.

This beast is the 36LW6: Pd= 40W, Imax= 1.75A

It was one of the last generation of Octal HD finals before the advent of the Duodecar (12 pins) and Novar (9 pins) all glass tubes that began to appear during the early 1960s. Part of this development was to allow for putting more tubes in the same envelope. You had a single bottle that might include two RF pentodes for TV IF amps, or combination power triodes and small signal triodes for vertical deflection duty, or triode/pentodes for sync sep/horizontal deflection time bases. FM demodulator/audio voltage amps and power pentodes were likewise included in the same bottle. It was common to put the HD final and the damper diode in the same bottle.

The 6.3V version has become extremely difficult to find these days since a good many were diverted to ham use as RF finals. Even worse was the diversion to illegal CB linears during the mid-1970s CB craze. These illegal rigs were all too frequently badly designed, and poofed the finals too quickly. However, there are quite a few still available if you go with the odd heater voltages, and the prices aren't exorbitant either. You'll have to either wind your own heater PTXs, or have them custom wound. The characteristics remain the same. This type was also designed for operation from a DC rail derived from a transformerless voltage doubler. The spec sheet mentions that it needs just 280V_DC to work, and the published plate characteristic doesn't extend beyond that particular voltage. That would account for the variety of heater voltages. This is a definite benefit for audio design: low load resistance, and the avoidance of high plate voltages.

This looks promising indeed. That's nearly 95W of audio power from a single pair of PP finals, running conservatively, and with a Vpp= 300V_DC. It is helpful that the grid voltage swing isn't out of line with the more common audio power pents, as this allows for lower distortion from the front end. A good many power finals for RF use in particular would need to run at twice -- or more -- the DC plate voltage and likely Class AB₂ operation. Getting that much audio power would require PP-parallel operation of the more usual audio power finals running Class AB₁.

The estimated H₃ for this loadline comes in at about 0.2%. That bodes well for sonic performance. Given the low screen voltages, there is another possible way to use this: screen drive. Use a MOSFET source follower to drive the screen grid, with the control grid bypassed to AC ground, and used only to set a Q-Point current. This becomes a type of Class AB₂ operation that can boost the output power while maintaining a reasonable plate dissipation. It is also helpful in that the screen current remains low until the V_pk is taken to very low levels indeed.

I'm definitely going to have to give this a go some day.