Connecting 6L6 cathode as source for 12AX7 filament heater

[jecars]

Question for the experts:

Looking at a Silvertone 1433 schematic, I noted that 6L6GB tubes have cathode bias connected through two 12AX7 heaters. It struck me as a potentially simply way to have DC filament heaters. However, I got curious about all the pros and cons of this approach.  I started a list and would like help completing it.

[Edited based on comments as of 2/10/17]

Pros:
* Simple circuit for DC filament heaters for preamp
* An option to reduce heater hum
* Others?

Cons:
* Power tube configuration must have cathode current of 140-160mA.
[Per PRR: This rules out 2x6V6 and 2xEL84.  4x6V6 or 2x6L6 can be made to work, though not at very high B+ ]
* AC heater circuit is still necessary for power tubes, etc, so perhaps other anti-hum options are "better"
* Others?

Toss-ups / Shrugs / Personal Preference:
* Slow, soft-start of preamp tubes
* "Fixed bias" in this case means "fixed grid voltage and variable cathode current"
* Coming out of standby (typically, B+ is cut) will be slower by 10+ seconds, due to preamp tube heater warm-up. However, come-alive will be smoother and more gradual without pop/thump
* Consequences of preamp tube heater failure is that it could be misdiagnosed as power tube problem. If cathode cap is used, be safe and use 100V rating to handle this (rare) failure case.
* Others?

Thanks

[clyde]

Eico HF87 did the same, most recommendations are for a dedicated supply in this circuit. 

[terminalgs]

I think it was a radio practice in the 1940's.  It also provides a slow start  (feature or detriment? you decide).


Here's another example with four 12AX7 filaments in series with a 75ohm resistor under a pair of 6L6's.


http://www.magnatoneamps.com/schematics/Magnatone-M-212-8.pdf

[PRR]

It was also done in an early mono Fisher I was fond of. And many more.

> dependent upon number and type of preamp tubes

Once designed and built, that won't change. If you have too many small tubes, only a few need real-clean heater wires (inputs, post-Volume, reverb recovery); the rest can eat the usual AC.

The main "constraint" is that the power stage cathode current *must* be 140mA-160mA. That rules out 2x6V6/EL84.  4x6V6 or 2x6L6 can be made to work, though not at very high B+. And "fixed bias" is fixed *grid voltage* and too-variable cathode current.

A hidden "mis-feature" is that it takes twice as long to start playing. First the 6L6es must hot-up to near full current, then the 12AX7es will begin to warm. Nominally 22 seconds, in practice 15-20 seconds. It is very "smooth" because the power stage is very stable before any signal comes through the preamp, and the preamp comes-up at heater heat-speed which can be slower than its thump. Fine, as long as the user does not get antsy at the 12 second and start hammering at it.

In a Standby amp which kills power tube B+, the 12AX7 will cool down, so it takes 10+ seconds to come out of standby. Mostly this does not matter. If you have a conductor, barkeeper, or preacher who needs you dead-silent yet ready to rock instantly, some other SB scheme is needed.

> Consequences of preamp tube heater failure...???

How often does that happen?

Power tubes shut-down. Cathode voltage roughly doubles. If there is a cathode cap, be generous on voltage rating. Here (315V B+) I would expect ~~62V, so use 100V cap. The amp stops playing, of course, but the preamp tubes are stone-dead so it is likely nothing for the power amp to play. The main drawback is that NO technician knows this trick so diagnosis will take forever. (Dead power tubes? Replace a preamp tube..... huh??)

[jecars]

Great inputs!  I have updated my original post to summarize the feedback so far.

[HotBluePlates]

Question for the experts:

Looking at a Silvertone 1433 schematic, I noted that 6L6GB tubes have cathode bias connected through two 12AX7 heaters. It struck me as a potentially simply way to have DC filament heaters. However, I got curious about all the pros and cons of this approach.  I started a list and would like help completing it.

[Edited based on comments as of 2/10/17]
...

Toss-ups / Shrugs / Personal Preference:
...
* "Fixed bias" in this case means "fixed grid voltage and variable cathode current"
...

This statement from PRR was not meant to be a "feature" of the plan; rather, it's a reason this plan is for a Class A (and cathode-biased) amp with steady average cathode current instead of a fixed-bias Class AB amp with widely-varying cathode current.

...
Cons:
* Power tube configuration must have cathode current of 140-160mA.
[Per PRR: This rules out 2x6V6 and 2xEL84.  4x6V6 or 2x6L6 can be made to work, though not at very high B+ ]
...

Connect more of the dots:
Your posted schematic shows 2x 12AX7 heaters, using pins 4 & 5, connected in-series from the 6L6 cathodes to ground.  When you series-operate the 12AX7's heater-halves, the heater requires 12.6v at 150mA.  We have two such tubes, so the heaters are looking for something like 25.2v at 150mA.

This define your bias and idle cathode current for the output tubes:  Whatever output tubes you use, the balance of the plate-to-cathode voltage, screen voltage, and allowable idle dissipation must result in a total cathode current (sum of all the output tubes) near 150mA (PRR places this number sensibly-around 150mA, at 140-160mA, since the 12AX7 heater will still operate fine in that area).

1.  In practical terms, the design starts with the choice to use the "preamp tube heaters as cathode resistor" then you know the required cathode current for the output tubes.
2.  The number of preamp heaters placed in series then provides your known bias voltage.
3.  Since we'll almost certainly use pentodes/beam power tubes, we find a screen voltage that permits the desired cathode current with the dictated bias voltage (likely near 75mA at 25.2v bias in this case).
4.  Class A operation of the output stage will be essential/assumed.
5.  Output tube plate voltage can be well above the screen voltage if we accept power supply complexity; however, this defeats the simple elegance of "free d.c. heat" for the preamp tubes (because we could just as well shift the complexity towards making a nice d.c. supply for the heaters).  Plate voltage near screen voltage seems implied.
6.  OT primary impedance implied to be near V_plate / Current_Idle (of 1 output tube);  Primary Z a little lower than this calculated value would be typical.

[jecars]

Connect more of the dots:
Your posted schematic shows 2x 12AX7 heaters, using pins 4 & 5, connected in-series from the 6L6 cathodes to ground.  When you series-operate the 12AX7's heater-halves, the heater requires 12.6v at 150mA.  We have two such tubes, so the heaters are looking for something like 25.2v at 150mA.

Thanks for the info. This leads me to a related question.

* I understand from 12AX7 data sheet that the heaters (2 per tube since it's a dual triode) are 12.6V/150mA for series hookup. Two tubes in series (i.e. four heaters) would therefore give 25.2V. Got it.

* However, on the Silvertone schematic in original post, it shows PT cathode nominal voltage of 19V. This discrepancy makes me wonder if 12AX7 heater current and/or resistance is actually different in real life. In any event, approximately 25% difference between schematic and data sheet seems like too much to ignore.

Do you guys have experience that explains this discrepancy?


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

Notes on my math...V=IR stuff

Data sheet of 12.6V/150mA implies that series resistance is 84 ohms per 12AX7. Incidentally, this implies that each heater is 42 ohms, since there are two heaters in a 12AX7.

Scenario (A): Is actual current draw more like 19/(2x82ohms) = 113 mA, vs 150mA

Scenario (B): Is actual resistance more like 19/150mA = 127 ohms (total) or 64 ohms per tube vs 84 ohms.

[sluckey]

Nothing wrong with your math but either case is based on an assumption that may not be correct. Who says that Sears put the real correct voltage on the schematic? Sears is pretty sloppy with their consumer electronics documentation. They do a great job with replacement parts though.

Data sheet of 12.6V/150mA implies that series resistance is 84 ohms per 12AX7.

Tube filament resistance is low when cold and increases as temperature increases. So what temp do you want to assume?

Don't overthink this. 19V should light those tubes just fine and also provide a nice safe bias voltage for the output tubes. Probably last a long time. Remember, this is not a high performance amp.

[jecars]

Don't overthink this. 19V should light those tubes just fine and also provide a nice safe bias voltage for the output tubes. Probably last a long time. Remember, this is not a high performance amp.

Thanks, I'll take your advice. I'm considering a breadboard mock-up so I suppose that I will see for myself at some point.

And, I realize that Silvertone is not an arena-rocker by any stretch .  Actually, that's my motivation behind researching these circuits.  For my playing, I find that I really like power tube clipping with it's compression and harmonic content. But, managing sound level to <100 dB in-studio is my key constraint. Something simple with 320V B+ & ~20V cathode bias driving 6L6GB or 5881 reissue has me curious.

Right now, my everyday player is Class AB 320V B+ with grid bias for 5881 RI.  I use a homemade simple/cheap L-Pad to knock off 3-6 dB to manage EPNdB and preserve my hearing (tinnitus sucks, so I'm told ). Mostly, I just like the character of the guitar to come through the amp with the power section doing it's thing. 

[PRR]

> cathode nominal voltage of 19V

The cathode current typically increases 10%-30% at full output. We do not want to over-power the heaters.

19V/2= 9.5V. That's low for a nominal 12.6V tube, true. The small tubes will not be optimally heated. But 12AX7 is a 10mA cathode (from 12AU7 types) never used over 2mA. It may be OK, It may not be "best" in some sense. It may be good-enough for Silvertone.

And of course the factory data was often taken hastily.

> tinnitus sucks

Two 6L6, even under-fed, will damage hearing.

[HotBluePlates]

...  When you series-operate the 12AX7's heater-halves, the heater requires 12.6v at 150mA.  We have two such tubes, so the heaters are looking for something like 25.2v at 150mA.

...
* However, on the Silvertone schematic in original post, it shows PT cathode nominal voltage of 19V. ...

I'm not certain, but suspect the current on a 12AX7 heater matters more (really, the "heater power," voltage * current, is the key to warming the cathode, but...).

Cathode bias always settles to whatever voltage/current the tubes want to see.  That means that assuming the Sears numbers are right, whatever cathode current  is that created the 19v drop across the 12AX7 heaters, that 19v of bias was the voltage which would result in that amount of current with those tubes.

If this sounds like a circular argument, it's because that's the nature of cathode bias. Cathode current rises until the voltage drop created across the cathode resistance stops the current from rising further.

In this case, it's the 6L6 cathode current vs bias voltage which dominates the picture.  "Right values" for 12AX7 heater voltage & current are unimportant, except that the 12AX7 may not deliver "the best performance" or have "as long a life" as if proper values were present.

I posted my notes to help a new design settles on "proper values".  No doubt they'll settle to "improper values" in a rushed design, just as the Leaning Tower of Pisa is settling to improper lean due to a foundation problem.

[sluckey]

I'm not certain, but suspect the current on a 12AX7 heater matters more (really, the "heater power," voltage * current, is the key to warming the cathode, but...).

I've always thought the current was the more important factor too, but that's based on experience with high power radar transmitting tubes and may not hold for receiving tubes. Never really thought too much about filaments with small tubes. Just kinda take them for granted. But big transmitting tubes were another matter. The last transmitter tube I worked on had filaments rated at 5V and 50A. The filament transformer was on a big variac and we would bring the filament current up slowly to 50 amps watching a panel meter. We never even cared what the actual voltage was.

[PRR]

Power is the main thing.

How to specify? ALL small radio tubes are run from fixed-voltage battery or AC winding. The tube maker tells you that voltage. You ask "how much current do it suck?" and that number is available, for the nominal voltage. There's some tolerance but very small.

However at another voltage the current will not be the same, or in exact proportion (Ohms "Law"). A hot filament is not a perfect resistor. For "small" changes like 11V-12V, proportion may be good enough.

As the tolerance is quite small, you may force either V or I and let the other value fall where it will. Sometimes the choice hinges on fail-safe design. In Sluckey's big bottles, you could "have 5V" and no heat due to blown filament or loose socket. Seeing 50 Amps actually flowing is a real good clue the tube is taking heat. (The case of zero V and 50A, a short, is also possible, but would happen way too soon on the Variac, and he might notice.)

> If this sounds like a circular argument

You have two things which are sorta resistors but not really. The cathode impedance can be 200r at 100mA and 100r at 25mA. Two heaters is 84 Ohms at 25.2V but dropping toward 25r at low voltage. If we had both curves, the intersection gives the point it will settle at. However by diddling G2 we can in-effect change the cathode impedance.

It is generally wise to figure how to work near 25.2V 150mA, then scale-back toward 20V and perhaps 130mA-140mA to allow for rise of current at full roar. If we say two 19W plates, 130mA suggests 292V plate-cathode and 4K-5K load, a reasonable op-point. You could instead pick two 6550 @40W each, go 571V, though this leads to an awkward 9KCT OT.

[jecars]

Power is the main thing.

Looking at power makes sense to me, based on my technical background.   

I did some calculations looking at power from the power tube point-of-view.  Using schematic values for B+ & screen voltage (less the cathode voltage), and 5881 data sheet, here's what I calculate for bias voltage from 25.2V to 5V:

Bias V   Bias I   Power from two 5881 (watts) 25.2    0.033      1.66 19.3    0.066      2.55 10.0    0.131      2.62 5.0      0.168      1.68 Note: From 12AX7 data sheet, heaters in series = 12.6 x .150 = 1.89W. Double for two 12AX7 gives 3.78W Based on this, I conclude that cathode @19V is more plausible because power available for 12AX7 heaters is near the maximum power available from the cathode. In other words, balancing power between the tubes would tend to reduce cathode bias --> move to peak of idle power ---> that's all the power that 12AX7 heaters get What do you guys think of this?

[trobbins]

jecars, there is also likely to be a certain designed in underbias due to the tremolo output stage bias wiggle in that amp.  That, and the aspect of class AB bias increase with signal would mean that 19V idle is designed to avoid significantly exceeding 25.2V + 10%.

There may be advantage in bypassing the common cathode locally, so that most of the higher frequency signal current stays close to the output stage and doesn't loop back through the input stage wiring.

I'd also put a 150V 5W zener across the common cathode to gnd, as protection against stressing heater-cathode in the 6L6's if a 12AX7 was pulled, or had a dodgy heater contact.