Switched Mode Supplies for Filaments

Henry's Cat · 5242

0 Members and 1 Guest are viewing this topic.

Offline Henry's Cat

  • Jr. Member
  • **
    • Posts: 24
on: December 10, 2012, 11:24:04 AM
Has anyone tried using switched mode regulators for filament supplies for directly heated triodes?

I've recently modified my "Single-Ended-Glory" 45 power amp. for use with a D.C filament supply as I was experiencing low-level hum at 100Hz from the A.C filament feed particularly when using a pair of Majestic balloon envelope 45's.  I tried using an all linear supply with a TS1084 2.5V regulator for each 45 tube filament.   But I found that too high a supply voltage to the regulators resulted in excessive power dissipation.  Too low a supply voltage and the regulators ripple rejection went out the window and an irritating buzzing could be heard in the background.

As overheating was a problem when supplying heavy loads like filaments I thought I would try switching regulators.  I've used them before and noticed how cool they run even when several amperes are being drawn from them.  I had not tried using them in a tube amplifier before because of the possibility of interference from switching noise.  After a search on E-bay I found some bargain ready-made modules, the hyperlink is below -

http://www.ebay.co.uk/itm/New-LM2596-DC-Buck-Step-Down-Voltage-Adjustable-Converter-Power-Module-Regulator-/150841422837?pt=UK_BOI_Electrical_Test_Measurement_Equipment_ET&hash=item231ed973f5

These are excellent regulators. They can supply up to 3 amps. and the output voltage is variable, set by adjusting a preset potentiometer.  The switching frequency is 150KHz.  An area underneath the board connects to the regulator tab via flow-through solder to conduct heat away from the chip and this is where I soldered a strip of copper to act as a heatsink.  One is normally not required but I decided to err on the side of caution.

The modules can be seen bolted to the case with extra heatsinking in the Amplifier Interior2 photograph.  Amplifier Interior1 photograph shows the interior wiring and the C4S active loads I adapted for the 6SN7G driver tube.  The photograph Power Supply shows the new housing for the off-board supply for the amplifier.  It's housed in a refurbished 1940's British Relay Wireless loudspeaker cabinet.  The front-end for the filament supply is linear, using a 6V 4.17A toroidal transformer, SB540 Schottky rectifiers and 6800uF smoothing capacitor.

The only difference to the amplifier's sound with the new filament supply is the inky-black background silence. The filament hum has gone completely and there is no audible signal breakthrough or interference from the switching regulators.
« Last Edit: March 31, 2013, 07:35:39 AM by Henry's Cat »



Offline grufti

  • Sr. Member
  • ****
    • Posts: 320
Reply #1 on: December 10, 2012, 03:23:20 PM
You can do it just don't tell anybody. Bad things happen when you talk about it to strangers. Just kidding of course, but I have used correctly built SMPS units for many "critical" supplies and have enjoyed the results.



Offline RayP

  • Sr. Member
  • ****
    • Posts: 301
Reply #2 on: December 11, 2012, 01:41:36 AM
I see you can get one with an LED voltmeter display.

http://www.ebay.com/itm/LM2596-Voltage-Regulator-DC-Buck-Converter-with-Blue-LED-Voltmeter-Display-/390511085141?pt=LH_DefaultDomain_0&hash=item5aec464655

I take it you do not do any more filtering between the device and the filaments.

Could you run two in parallel for increased amperage?

ray
« Last Edit: December 11, 2012, 01:53:04 AM by RayP »

Ray Perry


Offline Henry's Cat

  • Jr. Member
  • **
    • Posts: 24
Reply #3 on: December 11, 2012, 10:27:41 PM
I did think about applying more filtering but tried the regulators without any extra and there doesn't seem to be a need for any more as the amplifier sounds fine.

I wouldn't try to parallel these regulators as supplied.  I'm not even sure that they can be, the datasheet for the LM2596 doesn't feature paralleling in the application notes.



Offline earwaxxer

  • Hero Member
  • *****
    • Posts: 1336
Reply #4 on: December 12, 2012, 04:58:21 AM
I had problems with a switcher being anywhere near my tubes. Too much noise. I'm sure there are ways around that.

Eric
Emotiva XPA-2, Magnepan MMG (mod), Quickie (mod), JRiver, Wyrd4sound uLink, Schiit Gungnir, JPS Digital power cord, MIT power cord, JPS Labs ultraconductor wire throughout, HSU sub. powered by Crown.


Offline Henry's Cat

  • Jr. Member
  • **
    • Posts: 24
Reply #5 on: December 13, 2012, 06:08:03 AM
What type of regulator chip was it?  Maybe the switching frequency was too low and too near to the audio spectrum.



Offline earwaxxer

  • Hero Member
  • *****
    • Posts: 1336
Reply #6 on: December 13, 2012, 08:23:41 AM
oh, I have no idea - it was a large wall wort type. 30v output, its actually a laser copier power supply.

Eric
Emotiva XPA-2, Magnepan MMG (mod), Quickie (mod), JRiver, Wyrd4sound uLink, Schiit Gungnir, JPS Digital power cord, MIT power cord, JPS Labs ultraconductor wire throughout, HSU sub. powered by Crown.


Offline Henry's Cat

  • Jr. Member
  • **
    • Posts: 24
Reply #7 on: December 16, 2012, 10:54:34 PM
Ah, that would explain it.  Wall warts are usually cheap and nasty and it would be 100% switched mode were as my project has a linear D.C supply feeding the switching regulators.