• 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.

Those Magnificent Television Tubes

Ex-Moderator
Joined 2011
Unfortunately, no. Also a page is missing from the 2014 article as well, just our luck... But I think the use of the large shunt capacitor between the control and screen grids may be optional, since it was not connected in the schematic in the 2014 version.
 
One -WILL- have to address the output impedance in some way, since the Twin/Crazy Drive is high output Z by itself. Getting sufficient CFB to do much Z lowering though will incinerate the driver stage when combined with grid 2 drive requirements. Like a total cathode follower output stage running on grid 2!!! Suggest some other (global) approach. Or "Schade" back to the driver cathode.
In this case (conversion to crazy drive) I guess i would need a different output transformer?
 
There should be no need to change the OT as far as power matching goes. Your schematic in post 931 shows a global N Fdbk loop, which should take care of the output Z, as far as speaker damping is concerned. (no difference between grid 2 drive and Twin/Crazy drive as far as Zout goes)
 
Last edited:
Ex-Moderator
Joined 2011
Variation on the same theme, another "4DC" amp from Susumu Soya (MJ 2014 July), this time with 36LW6:

Screenshot - 11_30_2017 , 11_02_05 AM.png

Note the lack of shunt capacitor across the control and screen grids.
 
I did something like this with a 6CB5:
- used a mosfet source follower with about 60VDC on G2
- 100R in the cathode, 70mA, so about 7V over the cathode. This wasn't bypassed however.
- 1 resistor between G2 and G1, and 1 between G1 and Gnd,
- G1 was at about 4VDC, so below cathode potential and didn't pull any current

exact values for both resistor I cant remember.

Objective was to create a low distortion amp with high output impedance (transconductance amp) and a mate used it quite a while to drive his Supravox speaker (15 inch woofer and 8 inch midrange), 1st order Xover - with very good bass :)

well, reading through it, my thingy was quite different actually - but sounded very nice and will be remeasured one day!
 
G1 was at about 4VDC, so below cathode potential and didn't pull any current

This would give "Dual Drive", like George (Tubelab) was experimenting with earlier.
If G1 doesn't draw current, then the grid resistors maintain a fixed voltage divider ratio on the grids. Typically arranged near the internal Mu to get each grid to share 1/2 the work. Only requires 1/2 the drive voltage swing of G2 only drive.

--------------------------------

Jazbo8's posted MJ 36LW6 "Crazy/Twin Drive" schematic above is using a tube I have here, so I'm going to try those 10K/10K resistors with the 36LW6 on the curve tracer here to see what they are doing.

I've also determined that the 12G-B7 tube used in another "Crazy/Twin Drive" schematic (below) is close to a 6GB5 or 21HB5 tube, which I have also. So I'll try those tubes with 14K/10K on the tracer also.
 

Attachments

  • 12GB7SE-screendrive.gif
    12GB7SE-screendrive.gif
    19 KB · Views: 657
UH-Oh!

I just noticed that both of those SE schematics have grid 1 negative, with respect to cathode at idle, according to the posted voltages on the schematics.

SE would be operating in class A mode. So at best, Crazy/Twin mode could only occur for well positive signal voltage swings. Seems unlikely they would be operating with two modes of operation, so these two designs probably never even reach grid 1 current mode for Crazy/Twin drive operation.
 
Last edited:
I just remembered about power law from these modes I've measured and seen here.
Crazy Drive tends to linear law, and Dual Drive tends to a perfect square law.
Will be a good oportunity for compare the measurements and sound.
I cancelled my "multi-mode amp" project due to various factors (priority to others unfinished projects etc.) but someday...
 
42KN6 (old style) = 2X parallel 38HE7 beam pentodes

I just got my 1st 42KN6 Sweep tube in off the Bay, and it's the old style type, with two paralleled beam pentodes inside. I've been wanting to measure one of these, since the newer 42KN6 type, with one big pentode inside, is equivalent to the 6KD6, 6LX6, 26HU5 sweep tubes.

I curve traced it next to a 26LX6 and they are fairly close. I can get identical matched curves with the 42KN6 g2 at 85V and the 26LX6 g2 at 75 V, which matches with the datasheet internal Mu's of 4.5 and 4.0 reasonably well. Same grid 1 stepping for both, so matched equiv. gm 1.

Some while back I had speculated that the individual beam pentodes in the old style 42KN6 might be identical (datasheet wise) to the single beam pentode in the 38HE7 tube. So I connected up two 38HE7 pentodes in parallel to curve match against the 42KN6, and they ARE identical, Exactly! Same 85V on g2 and same grid 1 stepping for identical curves.

About 80% of the 38HE7 tubes I've seen bring out the heater separately for the pentode, on pins 10 and 12. Runs on 21V at 0.46 A. The 42KN6 runs at 42V at 0.46 A, so we have a match there. The 42KN6 is rated at 400 mA max DC, the 38HE7 pentode at 230 ma max DC each, so two would be 460 mA.

The 42KN6 is rated at 30 Watts plate diss. and the 38HE7 is rated at 10 Watts, but that is with the 6BJ3 equiv. damper diode operating along side in the bulb. The 6BJ3 damper uses 7.5 Watts of heater power and is rated for 4 Watts of diss. for a total of 11.56 additional Watts in the bulb. Without the damper powered up (using the 21V pentode only heater on pins 10 and 12), the beam pentode should be good for 15 Watts each I figure. So 2X in parallel would give the 30 Watts of the 42KN6. (the 21HB5 beam pentode is very similar in specs to the 38HE7 except bigger fin extensionss on the plate supports, and is rated 18 Watts alone in a bulb)

38HE7 are still on the $1 list. So you can make a 42KN6 (or a near clone of 6KD6, 6LX6... ) for $2 and two sockets.

------------------------------------------

I also got in a 20LF6 Sweep tube to try out (clearly marked, a GE tube with the extra grid 1 cooler fin up top). (came in a 26LX6 box I ordered, so by mistake I think) It curve tests VERY close to the 26LX6, just rated for more Watts. 40 Watts plate, 500 mA max DC, versus 33 Watts and 400 mA for the 26LX6, 6KD6 etc.

I also curve tested a 6MH6 tube I got in, and it is VERY close to the 26LX6 and the 20LF6 too. Rated 38.5 Watts and 500 mA max DC.

I then tried two 18 Watt 21HB5 tubes (which used to be on the $1 list) in parallel to see how they might compare with the 20LF6. Took 85 V on g2 to get the same knee current as the 20LF6 at 75 V. And the combined gm1 of the paralleled 21HB5 tubes was about 15% more than the 20LF6 etc for the same grid stepping.

(21HB5 has like 15% higher gm than the 38HE7 pentode, so was expected) Can get very similar curve tracings to 26LX6 and 20LF6 however, WITH the corrective compensations to g2 and g1. So not -exactly- an equivalent, but close enough, with those compensations. (and 36 Watts versus 40 Watts P diss.)

The 21HB5 plate is about 75% as big as the 20LF6 or 26LX6 plates. It usually doesn't have the grid 1 cooler fin up top like the 20LF6, but some manufacturer models do. (6HJ5 plates are the same size as 21HB5, and have the grid 1 cooler up top, for a 24 Watts rating, instead of 18)

Looks like I won't need to buy any $50 20LF6 tubes now at least. And no more 26LX6's either. Just sockets. (back during those old sale days, I bought a boxed deal of 138 38HE7 tubes for around $35 as I recall, and a big bunch of 21HB5's too. )
 
Last edited:
Tony:
Here are the 21HB5A triode curves. 50 mA/div Vert., 50 V/div Horiz., and 5V grid 1 steps from 0 V. The datasheet Mu for the 21HB5A is 4.8. (21HB5 non A is 4.7)

The gm and rp would depend on the plate current too. Rated gm is 9000 uMho at 46 mA and 130V. Pentode rp 9900 Ohms. Triode rp would be 1/gm2 = Mu/gm1 = 1000000 x 4.8 / 9000 = 533 Ohms at 46 mA. gm1 varies roughly as Ip to the 0.6 power. Can also measure off from the curves in the picture.

I suppose the triode gm1 might be 4% higher than the pentode rated gm if it includes the screen current.

2nd pic are the recent tubes: 38HE7 on bottom, next up 42KN6, then 21HB5A and 26LX6 up top.

As can be seen, the 38HE7 pentode plate is the same length as the dual 42KN6 plates above it, but the 38HE7 plate is wider, more than 50% of the tube diameter. The 21HB5A plate above those is also the same length, but mostly fills the tube width.
 

Attachments

  • rsz_21hb5a_50v_50ma_5vstep_t.jpg
    rsz_21hb5a_50v_50ma_5vstep_t.jpg
    77.2 KB · Views: 482
  • HE_KN_HB_LX.jpg
    HE_KN_HB_LX.jpg
    67.4 KB · Views: 478
Last edited:
thanks Don, that is a big help...triode rp at 533, and you have 5 in parallel and that makes around 100+ ohms equivalent to a 6c33 and more mu....yummy indeed...i am planning to maka a circlotron next year with those, i have 20 on hand all from the same International brand, and i would love to have more...got them when they were a dollar each, i should have gotten more from Rogalski brothers...can you point me to your source for those?

triode curves seem too lovely to ignore...
 
Sure.
ESRC still has a $1 list here (below):
With 38HE7 still on it. But 21HB5A is long gone now. I think Vacuumtubes.net (Rogalski Bros.) has them for $6 now. You might inquire with them for a quantity pricing.

6GB5/13GB5/PL500/XL500 tubes are nearly identical in curve tracing to the 21HB5a tubes. If only someone hadn't bought up 30,000 of them. They were on the $1 list for a long time.

Summer Dollar Days - Vacuum Tube Sale - $1.00 Vacuum Tubes

Now if they would just put the 26LX6 on a $2 or $3 list, since I can "build" them for 2 x $1!! Well it does take two sockets, guess we could increase that price to $6. :D ( I have gotten 36MC6 off Epay occasionally (rarely) for $8 or $9, nearly the same tube except for the (feared) Novar base)

I don't know how well any of these sweep tubes will stand up to triode mode over extended periods (and high plate voltages) The grid2 and grid 1 are close to each other in the Mica insulators. I would try one using the lowest plate V practicable. Screen V in pentode for the 21HB5 is rated 220 V, triode mode should allow some V extension, but run-away is a possibility at some point. George had run-away issues with 6LW6s, and those are rated for 280V on grid 2. 38HE7 is only rated for 150 V on g2 !

.
 
Last edited:
George had run-away issues with 6LW6s

I have seen a couple of 6W6GT's die a toasty red death while operated exactly at the edge of their published ratings.....and I have come up with a possible reason.

The 6W6GT is rated for audio output, and TV vertical sweep. It's identical twin the 12L6 was also used at the horizontal sweep tube in some old TV sets (Motorola Golden View for one). There are ratings for "class A1 amplifier" service with a maximum plate voltage rating of 300 volts and a screen voltage rating of 150 volts. The plate dissipation spec is 10 watts. There is no mention of triode operation.

There a separate rating for TV "vertical deflection amplifier" service where the ratings for plate and screen voltage are still 300 and 150, but the plate dissipation is lowered to 7 watts. There is also a triode connection rating of 300 volts plate + G2, with a dissipation rating of 7.5 watts plate + G2.

So, take a 6W6GT connect it to a power supply with 300 volts on the plate + G2, and adjust the grid bias so the tube dissipates 7 watts, about 22 mA.....Let it cook for a while, a long while. I set a power supply up to a tube and a current meter. Whenever my workbench was on, the tube was cooking. The first tube, a crispy old looking tube with a worn getter, fried in about 2 weeks. I put in a new tube about 4 months ago, and it is beginning to get "bias creep." I have to set the grid voltage such that I read about 18 mA so that it will not be over 24 mA after 8 hours or so. This is the first sign that a tube is going to run away.

I wanted to continue this experiment until the tube fried, but I just needed to steal the power supply and test socket for some real work.....a kick a$$ amp using 6550A's in triode for the moment (70 watts out), but bigger stuff is planned. You can follow along here, lots of tubes are going to get "tested."

Tubelab Universal Driver Board, 2015 version

I was planning to test the 6W6GT in triode as an audio output (the reason for the "cook" test), but it doesn't look like I will be able to squeeze big power without the runaway demon returning.

Why do the tube docs show 300 volts in triode for TV sweep use, but they blow up when we try this for audio? When does a TV sweep tube go into runaway, when being cranked beyond it's ratings? No, they blow up at IDLE. Every runaway I have seen happen to me has happened when the amp was left on without any music playing. This pertains to TV sweep tubes, and conventional audio tubes. Guess what a TV sweep tube DOESN'T do? IDLE, it runs full crank all the time. It doesn't get a break, you can't turn down the volume on the TV sweep, unless you want your picture to shrink.

I may still visit the 6W6GT in pentode mode. I have seen 40 watts with 400 volts on the plate, and 120 volts on the screen, but those were short term tests.
 
>>
I was planning to test the 6W6GT in triode as an audio output (the reason for the "cook" test), but it doesn't look like I will be able to squeeze big power without the runaway demon returning.
>>
That's sad news, I was planning on making a one on particularly 6W6GT, well, a PP with 50% quiescent will be ok, what do you think?

Any observed difference in such behavior between gray vs black plated ones? (maybe a lame question)

My setup often plays music on weekends from dawn to duck, several times I put it on mute and forgot to turn it off over the night and following monday. .. that's why I added the standby timers in all 3 that I lately made.
 
hey George, i made a 12EN6gt pp amp B+ at slightly lower than 300 volts G2 regulated at 150 volts cathodes at around 30mA...

and i liked the sound better than my 6V6 pp amp...

i made a 6LU8 pp amp in a mullard 5-20 topology, G2 and G1 both regulated, fixed bias, ang B+ slightly below 400 volts....sounds awesome too...

my fascination with TV tubes lingers on.....