[HBR] Re: Filament Voltage Dropping Scheme

[Walt Hutchens]

Dan said:
> I just measured the cold resistance of the 12SK7 plus the 50L6 to be
> 59 ohms. The time constant is then R times C or 229 microseconds.
> Hardly enough time to cause any thermal shock.

This scheme is actually a good idea and used commercially in certain
applications -- I have a plug-it-in-the-wall emergency light that
charges a couple of nicad 'AA' cells at 50 mA by running a cap to a
bridge rectifer and I have used the same trick for charging the battery
for a small test instrument.

Your case is a little more difficult because of the higher current.
While the caps will indeed run cold (no steady state power is actually
dissipated) the working *current( is 150 mA and the turn on surge could 
be close to 3A, depending on exactly when the switch goes ON.  (167V/59
ohms).  The connections inside caps are not necessarily rated to handle
such currents and I'd guess that failure is likely, one day soon.

You'd expect the failure to be open-mode but who knows where a bit of
vaporized metal or charred dilectric winds up?

The problem used to exist in the context of higher power transmitters,
in which the RF-carrying caps had to be rated by current, as well as
voltage.  Remember those big rectangular phenolic block micas?  If you
look closely there's often an amperage rating on them along with the
capacitance and voltage.

However, what you're doing is NOT an off-the-wall application.  You can
buy paper, mylar, and similar caps that are rated for high current
service.  They have internal connections that will take this treatment.
  Mouser is probably the best place to shop.

Or, if you're experimentally minded like most of us here, just be sure
there's a suitable fuse in series -- it might not protect the tubes but
should prevent anything worse -- and go for it with what you've got.

There's some suitable junk for such experiments in an average computer
monitor.  Any cap that's directly on the AC line is probably okay, since
these caps have to handle a turn-on surge.

It would be conservative to go for a higher voltage rating too, if this
equipment will be on very much.  'Line rated' capacitors are all 1000
WVDC and up, more often 1500 WVDC.  Surges can and do happen.

> In fact the capacitive reactance seems to cause the filaments to come
> up to temperature more slowly than with a resistor.  I need to do an
> actual experiment to verify this, though.

When you do the calculation for the capacitance, remember that the 
voltage across the combination will lag the current by just about 45 
degrees due to roughly equal voltages across the cap (90 degree lagging) 
and filaments (in phase).  So it's not just a matter of getting a 
reactance equal to the correct series resistance.  The easy way is to 
calculate that number and then measure the voltage across the tubes (not 
the cap!) and adjust.  The tubes are resistive; if they have the right 
voltage they'll have the right current because it will be in phase.  I 
think you'll wind up with rather more than the expected capacitance -- 
consistant with your observation that the tubes come up more slowly than 
with a resistance.

Good idea -- very uncommon in vacuum tube practice because when vacuum
was king, we didn't have caps that could do such things -- but still a
good idea.

Walt Hutchens
KJ4KV