6.3VDC from 18V transformer?

[OneRoom]

I'm trying to generate 6.3VDC at 300mA (for a vacuum tube filament) using the 18VAC winding of a transformer I already have. Ideally, I'd like to be able to regulate it using a LM317 or similar. The 18V winding doesn't have a center tap, so it seems like my options are to use a FWB rectifier, which would output ~24VDC...which leaves me a lot of voltage to "throw away," or try to use a half wave rectifier, which I think would only yield around ~7.8VDC, which is cutting it too close to use a LM317.

Are my calculations correct? All of this is assuming a 0.7V drop across the diodes. What would be the best way to go about this? Thanks!

 

[KlausST]

Hi,
don´t recommend a linear regulator like LM317.
The problem is waste of energy that then becomes dissipated as a lot of heat,

As you say you get about 24V DC input .. to generate 6.3V. The difference of 24V - 6.3V = 17.7V is wasted. Multiply it with 0.3A of load current you get 5.3W of heat. This is a lot, you need a big heatsink.

I´d go with a switch mode buck converter. It should dissipate/waste less than 0.4W. No heatsink needed.

Depending on noise requirements, you may need some filters.

Klaus

 

[crutschow]

The peak output of a half-wave rectifier is the same as a full-wave rectifier, the half-wave just has a lot more ripple, so half-wave won't help reduce the voltage.
If the load is a constant current, then you can reduce the filter capacitor value to reduce the average output voltage, and thus the LM317 dissipation.

Below is the LTspice sim of such a circuit:
The relatively small C1 filter capacitor reduces the average DC input voltage to about 18Vdc due to the high ripple voltage (blue trace), compared to about 24Vdc with a large filter capacitor, saving about 1.8W, but that still gives about 3.7W dissipation in the LM317 (purple trace), requiring a heat-sink.

A switching regulator, such as Klaus suggested, would be much more efficient, but switching noise could be a problem if it's powering a tube in a sensitive circuit.

 

[betwixt]

I would suggest one of those postage stamp sized switch mode regulators available from several sources for very little cost. LM2596 based ones are good up to 40V input and if you can keep the input voltage below 24V, the MP2315 ones are smaller and slightly more efficient.

Brian.

 

[dick_freebird]

If you had three tubes and their filaments were the same you could
series-string them and be roughly right. Many old radios used AC
for filament power (maybe not the real audiophile stuff of the day,
but 60Hz hum doesn't bother old timey radios much).

6.3VAC transformers used to be everywhere and dirt cheap. Bet
you could find an ancient B/W tube TV for free, pull its xfmr and find
a filament voltage winding. If, as I suggest, AC filament power
doesn't negatively impact performance of the whatzit "unacceptably".
Probably not hard to find, cheap, online at surplus places. Shipping
is another matter. But you only need a little one.

 

[BradtheRad]

Capacitive drop is an option since tube heating filaments often receive AC.

Simulation of different arrangements:
primary loop cap (high V small uF)
secondary loop cap (low V but larger uF).

The last circuit illustrates power factor correction of the middle circuit.

 

[KlausST]

Hi,

From my understanding:
* solution 1 is not possible this way on real transformers. The series capacitor works against the (urgently needed) transformer internal inductance and thus causes high primary currents that may kill the transformer's primary.
* solution 2 may work
* solution 3. I'm not sure how it is meant to work on a real transformer. I expect the secondary side capacitive current to overrule the transformer's inductance ... so even without the primary side compensation I expect it already to be "capacitive". And one can not compensate capacitive currents with a capacitor circuit.

I once did try #1 ... and resulted in tranformer killing.
I did try #3 only on circuits that had indictive character.

Klaus

 

[BradtheRad]

Klaus's experience exceeds my own and I believe his cautions to avoid using capacitive drop with real-life transformers.

Unexpected resonance easily creeps in. I had to give up on one (simulation) approach of putting series caps on each side of the transformer. (I wanted to see if it corrects power factor error.) However I was unable to adjust values so that the system behaved properly. Just when Ampere waveforms appeared to cooperate, their phase shifted and sometimes rose to high levels.