Lessons learned building a 5VDC Power Supply for a 300B heater

[sgerus]

My 300B amplifier had to much hum for my needs, so I built a nasty looking C-L-C power supply by tapeing two 10,000uf caps on a chunk of plywood, wired in a 3mH 1.5A choke, two 22R resister is series served as a bleeder and the tap for the 300B cathode R&C.

Worked like a charm… hum went way, way down.

But I just couldn’t live with this crappy looking dc psu, so I decided to make my first PCB. It turned out professional looking, but my hum level went UP.

Here is the lesson learned part:

The location (order) of the electrical connections matters.
(I know… basic electronics)

Picture the typical power supply schematic, the diode full wave bridge is on the left, then the first capacitor, the choke, the 2nd capacitor. The + and – output are at the right.

What I did wrong (when I made the PCB) was to put the two diodes that provide the negative side of the dc voltage on the right end of my PCB (instead of the left end).

Ground is ground…..Right? Wrong! You have to think of electrons like plumbing, you want all the water (electrons) flowing in the same direction.

When I tested the psu on the scope, I found a very strange looking sine wave. I moved the diodes to the left side of the PCB. How my DC output has a normal looking sign wave with about 20mv ripple.

There are some results of psu’s I’ve been testing:
I’m using a 6.3V 1.2A transformer, 10,000uf caps and 3mH 1.5A chokes.
The diodes are 31DQ09, you drop about .8V over each diode at 1.2A

The goal is 5VDC at 1.2A, with just a C-L-C psu

1. 30,000uf no choke 5.42V, 190mV p-p ripple, saw tooth wave patter (not good)
2. 10,000uf ,3mH,10,000uf 5.8V, 50mV ripple, normal sine wave
3. 10,000uf ,6mH,10,000uf 5.4V, 25mV ripple, normal sine wave (better)
4. 20,000uf ,6mH,10,000uf 5.2V, 7mV ripple, normal sine wave (best so far)
5. 10,000uf ,3mH,10,000uf,3mH,10,000uf same as #4

At this point I don’t know how little ripple is good enough… my plan is to start with example #4, since that gets the voltage the closest and has the least amount of ripple.

[Doug Kim]

Thank you for sharing your story.
This type of post is double-helpful when accompanied by some pics.

Doug

[kevinkr]

Your transformer is probably a bit undersized based on the 1.2A rating you mention. Due the non-sinusoidal rectified charging currents and their phase relationship to the ac supply voltage you need to increase the transformer rating by at least 60%. I recommend a minimum of 2A.. The power factor of a bridge rectifier driving a large capacitor can be 0.6 or worse.

If you ever plan to use JJ 300B or others with 1.5A filaments you will need an even higher rating.

I use CCS based heating which allows me to get away with 4700uF filter caps and no choke. No hum either. (8V/3A secondary, 6A bridge, cap, LT1085, 1 ohm resistor, 470uF cap on out.)

[sgerus]

Your transformer is probably a bit undersized based on the 1.2A rating you mention

That's good info. Right now I'm just using these transforms for testing.

My plan (for the next amp) is to use Your typical Hammond 350-0-350vac 175mA power transformer which would have a 6.3vac 4 Amp section.

I'm THINKING of useing that 6.3vac 4A section for the 5vdc power supply AND the in put/driver with 6.3VAC!

I expect this arangement to bias up the input driver to the level of the 300B cathode..... can anyone confirm that?

I've only seen one schematic like this.... so we will test this on the breadboard before building the actual amp!

[gingertube]

Guys,
I've been mucking about with B2Spice looking at 300B Filament supplies (5V at 1.2 Amps).

The Simulator Schematic is attached.
V1/R1 models a 6.3V 2A transformer
D1-D4 are 1N5822 Schottky Diodes
C1 is Panasonic FC 2200uF/16V
C2 is 2 off Panasonic FC 4700uF/16 in parallel
Note: The 16V version used for lower ESR and higher ripple current rating than the 10V versions.

X2 is IRLZ44N which is TO220 Isolated Pack
Mosfet power dissipation is 0.3 watts
X3 is AD8031 Low Noise rail to rail input opamp.

R3 is the 300B filament

V2 is the reference which will need to be low noise "buried zener" type with divider resistors to suit.

Simulator says I'm getting +/- 160 microvolts (113 microvolts RMS) of ripple across the 300B filament.

The position of R10 is important - It gave a huge improvement by being on the -ve side of the supply (as opposed to an equivalent resistor between the caps on the +ve side or even one resistor of 1/2 value on each side) - A very usefull trick.

Cheers,
Ian

[kmaier]

Gee, small world. I'm also in the process of designing a small DC filament supply for a 300B monoblock. S/N referenced to 1-watt is only 52dB with an AC filament. Also looking at a C-L-C supply with a 7.6VAC/12.6VAC c.t 3A supply and a pair of 4A hexfred diodes for full-wave. Nice to see some measured specs on a few variations, thanks.

I'm leaning towards simple, ie, no active circuitry for VR or CCS. One failure might damage an expensive WE300B. Hopefully I get a prototype running over the weekend. I need a solid 30dB improvement.

Regards, KM

[jackinnj]

Want a great 5VDC supply? Subscribe to the Migratory Junk Box thread -- http://www.diyaudio.com/forums/showt...56#post1560156
there's peach of a Lambda supply for TTL in there and someone's bound to figure out that it would work wonders in this application.

[sgerus]

This type of post is double-helpful when accompanied by some pics

Doug,
I created a pdf file with pictures....

I'm leaning towards simple, ie, no active circuitry for VR or CCS. One failure might damage an expensive WE300B.

KM,
That's exactly my concern, and I really want to run this from 6.3vac. You need more AC voltage to use ccs or any kind of solid state voltage regulator.

Also, I try to keep the heat inside the amp as low as possible. A 7805 voltage regulator needs to drop about 2 volts to work correctly. That’s about 2.4 Watts of power (heat).

[kevinkr]

Originally posted by sgerus


Doug,
I created a pdf file with pictures....



KM,
That's exactly my concern, and I really want to run this from 6.3vac. You need more AC voltage to use ccs or any kind of solid state voltage regulator.

Also, I try to keep the heat inside the amp as low as possible. A 7805 voltage regulator needs to drop about 2 volts to work correctly. That’s about 2.4 Watts of power (heat).

No failures at all in 10 yrs (4 different 300B SE amplifiers - 3 in use with customers) so properly designed and heatsunk I don't think this is much of an issue. Also unregulated supplies are subject to variations of line voltage.. (Variations of +/-5% are not unusual during the course of a day.) A CCS gives you inrush current limited to the rated operating current of the filament which should in theory improve tube life. (My 300B are 6 yrs old and still fine)

The LT1085 has relatively low dropout voltage and even with the current setting resistor voltage drop will not be substantially worse than a 7805. A couple of additional watts of dissipation compared to what you are burning in the cathode bias resistors is not that significant. I use fixed bias in my designs for this and many other reasons. (Nothing better than no cathode bypass cap IMO.)

[SY]

Originally posted by jackinnj
Want a great 5VDC supply? Subscribe to the Migratory Junk Box thread -- http://www.diyaudio.com/forums/showt...56#post1560156
there's peach of a Lambda supply for TTL in there and someone's bound to figure out that it would work wonders in this application.

Too late.

But the next guys on the list will be pretty happy with the stuff I chucked in there.