DHT filament PSU: What would you do?

[Fuling]

Hi,

After chasing funny looking tubes and other stuff for two years it´s about time to get started with my 808 SE class A2 project.
The output tube filaments requires 7,5V @ 4A each, and AC heating is out of the question (µ=49).

Pretend you were in my situation and had the following parts to choose amongst, what would you do?:

2pcs 11VAC ~5A toroid transformers

2pcs 8,5VAC 4,8A " "

2pcs 10VAC 5A " "

6pcs 1,1mH 10A filter chokes with airgapped transformer cores

6pcs 27000uF 16V caps

4pcs 100 000uF 15V "


I´m leaning towards using the 11VAC transformers followed by LCLC-filters (1,1mH + 1,1mH - 27000uF - 1,1mH - 2x27000uF).

[danny_66]

This is what I would do: common mode chokes
See the second schema with two chokes on
http://www.dhtrob.com/impressies/com...e_choke_ae.htm

The article is in Dutch, but babelfish can help a little:
http://babelfish.altavista.com/babel...e_choke_ae.htm
smoorspoel = choke

[Dave Cigna]

If you don't need to make a decision ahead of time then try different combinations. In my experience the filament supply is at least as important as the B+ supply, maybe more so. Even if you think you know ahead of time which configuration will be best, you owe it to yourself to try them all just to hear the difference.

Make sure you try LCL. That's right, no final C after the last L.

-- Dave

[Fuling]

I´ve seen common mode chokes in various filament PSU schematics, but I don´t quite understand their purpose. Surpressing noise from the mains? Must read further about this, the babelfish translation didn´t make me much wiser...

Dave: I should have mentioned that the 11VAC transformers also have 110 and 220VAC windings suitable for the input and driver stages, so it would be great if I could use them.
What´s the point of removing the last "C"?

With LCLC-filtering as mentioned above I will probably have a volt or so to spare, so I guess I could add more inductance at the output if it´s beneficial.

[Fuling]

I made a quick prototype to run some basic tests and the initial results were not too impressive: with 8,35V out across a chain of 0,47R resistors (1,88R in total) there was a loud buzzing sound from the chokes.
Quite a surprise, as I´ve used similar chokes and caps before in a LCLC filament PSU for 811A and that time it worked just fine. This time I used two chokes in series at the input position to cope with the higher input voltage, so i didn´t expect this to happen.

The solution was easy though, I inserted 0,15R in series with the second "L" ( a single 1,1mH choke) and the buzzing was gone and the output voltage down to 7,9V. There is still 0,4V to get rid of, and I guess another 0,1-0,15R resistor would bring the output down to the desired 7,5V and add some "safety margin" for the chokes.

[Dave Cigna]

Common mode chokes: Ordinary (differential mode) chokes resist changes in current through them. They provide the obvious smoothing of voltage across the filament. Common mode chokes provide no smoothing. (At least not in theory. In reality there might be some leakage between the two windings which results in a small amount of smoothing.) What a common mode choke does is block voltages (noise) that cause both ends of the filament to go up and down together. i.e. common mode voltages.

Where do these voltages come from? Mostly they are capacitively coupled through the power transformer, I guess. Though I suppose there might be some picked up in the wiring to the filament, which suggests that the CM choke should be close to the tube.

The filament in a directly heated tube is especially sensitive to CM voltages because they appear as input to the tube. The entire CM signal is 'modulating' the grid-cathode voltage so it gets amplified by the tube.

As for filament supplies with an L as the last element, some people report that it sounds better that way without a C across the filament. We can only guess why, but one guess is that some of the signal current ends up going through that last cap if present.

On a more general level, some people feel that efforts to keep the signal currents out of the filament supply are well spent. Note that that is a completely separate issue from simply filtering a DC supply. With that thought in mind, it makes sense that a CM choke as the last element will block signal current from wandering up into the filament supply. Maybe that's the real reason that people report an improvement in sound when they add them. I don't know for sure. Although I am interested in knowing, I don't need to understand it to believe that there might be something to it.

[Fuling]

Thanks for a good reply!

Seems that a CM choke between the actual PSU and the filament might be a good idea, especially since I´m going to use cathode feedback in the output stage (the cathode will "move with the signal"). Isolating the filament from its PSU must be a good thing I guess.

BTW I have ordered another pair of 11+110+220V transformers, the ones I already have are going to find a home in a pair of OTL monoblocks. Unfortunately the (surplus) dealer was out of toroids but there was two EI transformers with the same specs left so I grabbed them.
I guess it would be a good idea to test them with the actual load before I start designing chassis, buzzing transformers are just as boring as buzzing chokes.

[Brett]

SLA batteries and a shunt reg to get the voltage? Then add a charger for off and/or standby positions.

[woody]

I just pluged some numbers in PS2 . A choke input filter
using your 10v transformer and 2 of your 1.1 mh chokes
then a 22,000uf cap came up with 7 v using a standard ss rectifier now a schokky would probably put you right at 7.5 v.
But a lot will depend on the seriese resistance of those transformers.

[Fuling]

Brett: SLA batteries has actually crossed my mind once or twice, but no. Too bulky, too expensive and the voltage drop x the current means quite a bit of heat.

Woody: I tested the PSU with one of the 10V transformers and got 7,2V out, and that´s with the 0,15R resistor in position.
One explanation might be that the transformers have 220V primary windings and the mains voltage here is 230V.
Also, the chokes are made with about 30 turns of very thick magnet wire so the DCR is next to nothing.