0
\$\begingroup\$

i'm in need to reduce the 9 to 12V from my power supply to 6.3V, for a current consumption of 300mA. Here's the requirements:

  • Cant be a switching solution (so Buck converters are not an option)
  • Doesn't have to be a very clean 6.3v, as long as it stays within 5% tolerance
  • Can't evacuate a lot of heat (small closed enclosure). (my first idea was the usual LM317, but it does dissipate a LOT of heat with this requirements)

I know it's not an easy solution (considering i eliminated the 2 most logical options) and eventually i'll have to compromise (i'll probably end up using the LM317 and hope i don't cook everything). Any ideas would be apreciated (oh, i already thought of 7 nasty diodes in series, i was really hoping it didn't came to that)

(for those who didn't already guess what could use 6.3V/300mA, it's a vacuum tube heater)

any ideas would be welcome

\$\endgroup\$
  • 3
    \$\begingroup\$ A "Cant be a switching solution" in combination with "a vacuum tube heater" sounds like an audiophile. "Cant be a switching solution" without any reasons means that the question belongs on puzzling.SE. Your diodes will drop as much heat as an LM317. \$\endgroup\$ – pipe Jul 6 '16 at 1:55
  • 1
    \$\begingroup\$ oh, i actually really like using buck converters, but the location i'm currently located i have to order even a simple LM2596 and takes a while to get here, and wanted to solve this rather quickly with components that are easy to find (say in local electronics store where "led light" would be a monumental discovery). \$\endgroup\$ – ndelucca Jul 6 '16 at 2:08
  • 3
    \$\begingroup\$ You can build a switching regulator from a LM317. See the TI datasheet for the circuit. \$\endgroup\$ – Ignacio Vazquez-Abrams Jul 6 '16 at 2:13
  • 2
    \$\begingroup\$ I like to keep some (known) toroidal cores on hand so I can wind my own if needed. \$\endgroup\$ – Ignacio Vazquez-Abrams Jul 6 '16 at 2:25
  • 4
    \$\begingroup\$ No switching + can't dissipate heatmeans you need to change a requirement, really. You have ~6W of linear loss (somehow) or you have much less loss and switching noise to deal with. \$\endgroup\$ – user2943160 Jul 6 '16 at 3:11
1
\$\begingroup\$

I'd really like to post this as a comment, but I don't have the rep. I'll try to make it a reasonable "answer".

You didn't specify your tube complement, or what the circuit does. Would you care to share this information?

If you are flush with tubes, you can always add a second tube and put the two filaments in series across twelve volts. 6 V vs. 6.3 V won't make a discernible difference. Maybe the second tube could be used in the circuit?

\$\endgroup\$
  • \$\begingroup\$ The variable 9 to 12V input seems to indicate a poorly-regulated supply or different possible supply sources. This would definitely prevent series tubes from being practical as a regulation method. It would also have about the same amount of additional thermal dissipation as a linear regulator with none of the regulation advantages. \$\endgroup\$ – user2943160 Jul 7 '16 at 13:30
  • \$\begingroup\$ @user2943160 He doesn't specify why the input can vary from 9 to 12. My solution does assume that he can choose to use 12 V. Tube filaments resistance is dynamic, much like an incandescent light bub. Resistance goes up as the filament is heated, being very low when no current is flowing. Putting identical tubes in series ensures voltage is split pretty evenly between them, but of course it is not regulated. And he could conceivably use that extra tube as another active device in the circuit (which the LM317 would not provide). \$\endgroup\$ – Michael Gorsich Jul 7 '16 at 14:36
1
\$\begingroup\$

There are a few specs that are mutually exclusive, and you're basically ruling everything out.

Let's see:

  • Doesn't have to be a very clean 6.3v, as long as it stays within 5% tolerance

How accurate must the rail be over load? Is that still 5%? This is why voltage regulators were invented; they provide a constant output voltage given a sweep of voltages and load currents.

A diode, resistor or any passive dropper "solution" have rather horrific line/load regulation specs, that will likely surpass 5%. A diode can drop anywhere from 0.3V up to 1.0V depending on the type, the load current, etc. These rules of thumb can work, but you have to remember these will only work for 1 load current and also for the particular diode you have chosen. Same really for resistance; 20 ohms may work at 12V, but maybe not at 9V. Is there a large difference in load current between a cold and a warm tube?

But then you also require:

  • Cant be a switching solution (so Buck converters are not an option)
  • Can't evacuate a lot of heat (small closed enclosure)

Any linear regulator/solution will dissipate (excess) voltage x current as a heat power output. 2W is plenty of heat, especially with no heatsink on a TO220 package. A switching regulator is a better solution, but I understand for logistics reason you can't use it.

Any "dropper" circuit using earlier said components will still dissipate a similar amount of power, however it may be spread across multiple components. This can be an advantage and disadvantage; i.e. more components to cool (if each of them are running excessively hot) or multiple components to spread the heat.

My advice: get a switching regulator OR a linear one with proper thermals (heatsink/ventilation). If that will not last, you could bodge something together but only expect it to work under 1 set of conditions.

\$\endgroup\$
0
\$\begingroup\$

Using linear regulators you are looking at a maximum of 2W heat output. If you are okay with the regulator dissipating that much heat you can safely go ahead and use LM317. If not you are pretty much stuck with a switching regulator like LM2596.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.