Some thoughts on a filament power supply for the Aikido.
AC filament power is worth considering. Indirectly heated tubes don't pick up too much noise from the heaters. I'll be using AC heaters when I first power up my Aikido boards.
DC will reduce noise. My initial idea was to use a STEPS board to provide unregulated power around 14 V. I'd leave out the regulator. A series connected diode string would cut the voltage down to something very close to 12.6VDC. If I ever switched to 6 volt tubes, I'd just swap the primary to series connection, cutting the output voltage in half. Just one problem. The largest transformer that will fit on a STEPS is 25VA. At typical tube complement needs just over 30W.
I decided to step-up the STEPS and the MEPS was born.
The unregulated power is derived from Tangent. Matching Vunreg to the output is critical at the higher currents. If Vunreg is too high, we will dissipate too much power in the regulator. If Vunreg is too low the regulator will drop out. There needs to be a margin to allow for line voltage fluctuation. I want to allow for post construction Vunreg adjustments without having to swap transformers. The Aikido will work with a wide range of tubes. The filament supply should support that with a minimum of work.
MEPS keeps the original over spec'd line filter. The transformer is an Amveco 35 or 50VA unit. I plan to have jumpers for both the primary and secondaries, allowing for the maximum possible output choices from a single transformer.
Rectifier diodes can be any TO-220 units that can handle the current and power. Two or three optional series diodes follow the bridge. You would use these when needed to drop the unregulated voltage a little. I can see needed this when swapping to tubes that use less power. With 6Volt tubes, these diodes could dissipate a few watts, so I will leave room for heat sinks.
I have several ideas for the regulator. The above schematic is idea #1.
An LM317T is limited to 1A. That isn't going to work. Moving up to the LM338 gets us 5A. That is enough for 12V tubes but not 6. Power dissipation is a problem. The dropout voltage is around 2.5V. To get a safe margin, Vunreg is going to have to be 3 to 6V higher than the output.
6V*2.4A = 12.8W
12.8*4 = 51C
That is with a perfect heat sink! I decided to split the heat across two LM338s. The increase in cost isn't much. You can't just put two 3 pin regulators in parallel. If you do that, one ends up getting all of the load, then goes into thermal or current limit. This regulator circuit, updated from the LM338 data sheet, uses the two .1 ohm resistors to sense the current going two the two regulators. The op-amp adjusts the top regulator to keep the load balanced.
National has an ancient application note, LB-51, that describes an alternative way to do this. It is worth a read.
This will work for 12V tubes. 6V tubes draw twice the current and will dissipate twice the power in the regulator. With a regular NPN linear regulator, the amount of power dumped into the regulator is nearly as much as what goes into the tubes. That's not a good design.
That leads us to idea #2.
A LDO regulator, like an LM1084 has a 1.5V drop out. Some of the LT parts from Linear are 1V. Thats much better but we need to keep the input voltage less than 3V over Vout. One way to do this is with a switcher feeding a linear regulator. Since there are some interesting 6V tube choices, I'll draw up a #2 MEPS over the next few days.
Comments are welcome.
Andrew
