300B filament supply question

[needtubes]

I know this question has been asked before, but I've found the hundreds of pages of discussion to be confusing...

I am wanting to rebuild the filament supplies in my 300B amps. Currently, they use the 5V tap into a bridge rectifier, into a Xicon 10,000uF 16V cap, into a 0.8-ohm resistor, and into another Xicon 10,000uF 16V cap.

I have the following limitations:

* There is limited space available in the amp chassis, so I cannot fit in a large inductor.

* I don't have enough voltage, nor space for another transformer, to allow a regulated supply.

I am hoping to use Schottky diodes into a higher quality capacitor, then into an inductor, followed by more high-quality capacitance. But, I don't know which parts would be best... Do you have any recommendations for the Schottky's, caps, inductors, etc?

Thanks all!

[ThorstenL]

Hi,

Originally Posted by needtubes

Do you have any recommendations for the Schottky's, caps, inductors, etc?

Links to vendors are provided for convenience and consist neither recommendation or endorsement in any way of the vendor, only of the specific part. Generally parts of comparable quality and rating can be substituted.

Diodes:

4 pcs of 31DQ04 assembled as discrete bridge.

In theory you can solder them directly across the old Bridge rectifier, their lower flow voltage will effectively "hog" all the current and disable the original bridge, doing it properly and using a small vero-board looks cleaner and gives more satisfaction of a job well done.

Capacitors:

1st Cap Nichicon FW 10,000uf / 25v

2nd Cap Nichicon UFW 33,000uf / 10v

Choke:

This needs looking at what parts you can get where you are.

The choke needs to be rated for 1.5A to 2A DC and reasonably low DCR.

I would suggest using 2 pcs, one in each line of the supply (Positive & Negative) and adding a resistor between bridge and the first capacitor to fine-trim the voltage.

For example the API Delevan 1,000uH/2.4Amps/0.235 Ohms has suitable ratings...

Two in both legs give effectively 2mH inductance with 0.47 Ohm DCR. It would be nice to have more inductance, but the needed chokes may be hard to come by...

You can easily sim the resulting PSU in PSUD, I did quickly and I get 9mV RMS (26mV Peak-Peak) remaining noise with a nice sine wave shape, indicating the absence of higher mains harmonics. The remainder should null out nicely using a normal hum bucker pot.

Ciao T

[needtubes]

Thank you!

I am wondering... Are there "rules" that apply when designing a filament supply? Meaning, should one not immediately follow the Schottky's by a large cap? Is there a reason for placing the RC filter before the choke, or can it also be placed after the choke?

I would suggest using 2 pcs, one in each line of the supply (Positive & Negative)

So a common-mode choke is acceptable here?

Is there any concern (hum, noise) with these small chokes being within an inch or two of my 300B and the output wiring?

[ThorstenL]

Hi,

Originally Posted by needtubes

I am wondering... Are there "rules" that apply when designing a filament supply?

Same as all other supplies.

Originally Posted by needtubes

Meaning, should one not immediately follow the Schottky's by a large cap?

Yes, this is usually a good idea for all power supplies. Big capacitors reduce ripple but increase peak currents and as a result all sorts of havoc ensues...

Originally Posted by needtubes

Is there a reason for placing the RC filter before the choke

Yes, there is. One reason why Tube rectifiers sound better than solid state is because they add some resistance. This limits peak currents and makes the charge current waveform "softer".

So in this case we kind of simulate this effect. So, if we have to trim the voltage down it is best down with a resistor between rectifier and first cap. In my commercial designs I tend to "design in" the DCR to the mains transformer...

Originally Posted by needtubes

So a common-mode choke is acceptable here?

Of course not. A common mode choke has practically zero inductance in series with the current loop, hence it is a common mode choke, we need a differential mode choke.

As these tend to get big quickly as currents and inductance goes up (you realise, the 1mH ones I linked are 33mm diameter and 16mm thick - this is not "small") and increasingly hard to find it makes sense to use two. In addition, for common mode noise they will of course also form a barrier equivalent to halve their individual value...

Originally Posted by needtubes

Is there any concern (hum, noise) with these small chokes being within an inch or two of my 300B and the output wiring?

Look at the orientation and orient them to they do not directly radiate towards the grid pin and coupling cap, then you should be okay. Output wiring will be no issue.

Ciao T

[needtubes]

Would a choke like the following work?

2x3.3mH 1.5A Current-compensated ring core double chokes

It claims to be a double choke, so I'd need only one to handle both lines, right?

[kevinkr]

Originally Posted by needtubes

Would a choke like the following work?

2x3.3mH 1.5A Current-compensated ring core double chokes

It claims to be a double choke, so I'd need only one to handle both lines, right?

This is designed for suppression of common mode noise so this is not what you want. (Data sheet) You need chokes on separate cores with some degree of physical separation.

[sawreyrw]

From the data sheet it looks like a common mode choke. This won't work in your application, because the differential inductance will be very small.

[ThorstenL]

Hi,

Originally Posted by needtubes

Would a choke like the following work?

2x3.3mH 1.5A Current-compensated ring core double chokes

No, "current compensated choke" is another way of saying "common mode choke". These are basically mostly expensive resistors in our application. You need real chokes. No point trying to cheap out, you might as well not bother in that case.

Ciao T

[v4lve lover]

some ATX power supplies have chokes in them perfectly usable parts for filament regulators

[needtubes]

Originally Posted by ThorstenL

"current compensated choke" is another way of saying "common mode choke"

I didn't know this... I'll look again.