Hi. I'm looking at buying the Tentlabs filament supply for my amp project. The Tentlabs website says that each unit should have a separate transformer secondary though. My current PT only has one available 6.3V winding, but it's rated for 5A. Is there any reason I can't connect both units to this in parallel?
As an alternative, I also have a spare transformer with a centre tapped 12.6V secondary. Is it possible to use the centre tap to make this behave like two 6.3V secondaries?
As an alternative, I also have a spare transformer with a centre tapped 12.6V secondary. Is it possible to use the centre tap to make this behave like two 6.3V secondaries?
If this is for your 300B project, no. Each filament must float, meaning ideally as isolated from its power supply as possible. Separate transformer windings, rectifiers, filtering, yada yada, are the bare minimum for cathode bias. For really significant signal frequency isolation, you also need something like the Rod Coleman regulators.
300B's are like a snub-nosed revolver. Folks naively think that they're an easy solution, but actually, they're an expert's weapon, requiring great delicacy in operation.
All good fortune,
Chris
300B's are like a snub-nosed revolver. Folks naively think that they're an easy solution, but actually, they're an expert's weapon, requiring great delicacy in operation.
All good fortune,
Chris
The winding must be completely isolated from each other, as Chris says.
> As an alternative, I also have a spare transformer with a centre tapped 12.6V secondary. Is it possible to use the centre tap to make this behave like two 6.3V secondaries?
CT windings are just about always connected together, so that will not work.
A separate transformer is always better - sharing windings for high-current DC with the anode/plate supply's winding will lead to some cross-coupling, and degradation of performance.
> 300B's are like a snub-nosed revolver. Folks naively think that they're an easy solution, but actually, they're an expert's weapon, requiring great delicacy in operation.
They certainly don't take kindly to any corner-cutting in the driver design, nor in any of the power supplies.
> As an alternative, I also have a spare transformer with a centre tapped 12.6V secondary. Is it possible to use the centre tap to make this behave like two 6.3V secondaries?
CT windings are just about always connected together, so that will not work.
A separate transformer is always better - sharing windings for high-current DC with the anode/plate supply's winding will lead to some cross-coupling, and degradation of performance.
> 300B's are like a snub-nosed revolver. Folks naively think that they're an easy solution, but actually, they're an expert's weapon, requiring great delicacy in operation.
They certainly don't take kindly to any corner-cutting in the driver design, nor in any of the power supplies.
What about with fixed bias?
Not a 300B, but I'm using 26 input tubes with battery bias on the grids and I was told that a shared filament supply was OK in that situation.
I have no way to measure but there isn't any audible crosstalk between channels. Soundstage is excellent.
You can share the Raw DC (trafo, rectifiers, caps) in general with DC supplies that have grid bias{with many limitations}, but the Tentlabs unit has the rectifiers built-in. I don't think a shared winding will work well there either.
Sharing the actual regulator across two different filaments degrades performance
Sharing the actual regulator across two different filaments degrades performance
Last edited:
Thanks, I was asking a more general question, I'm not using the Tentlabs supply. To be more specific . . .
I've got a little Meanwell AC-DC SMPS with additional filtering (L-C) supplying the filaments of both tubes and a battery pack that provides -4.5v bias to both tubes. Plate supply is regulated with a VR tube and then each 26 is loaded with a cheap Hammond 156C plate choke (one 156C per tube).
Any glaring issues with this setup?
It sounds good.
Thanks for all the replies. I'll get onto finding another transformer to run the second supply and will get in touch with you, Rod, about ordering some of your regulators.
The difference with directly-heated tubes is that the filament voltage skews the effective bias, depending on where you look at the filament. so the bias is hotter at the positive end, colder at the negative.
This seems like no problem until you consider the gm (mA/V). With skewed bias, you get skewed gm, and this forces a music-signal to appear across the filament. This leads to a number of negative effects, depending of the type of power supply used to heat the filament.
- any kind of regulator that senses the output voltage will consider the music-signal to be an error. It "corrects" this error by throwing current into the filament. This current also includes distortion products from the voltage-regulator output (they are rarely designed for low-distortion). This unexpected current merges with the plate current of the triode.
- voltage regulators often show increasing phase lag with frequency - the bode-plot looks to be inductive. When a capacitor is added to the output, it forms an LC-like resonance in the output impedance, often in the audio band. This can add an unexpected EQ effect. See for example, the curves of output impedance of the LM317, with different capacitors.
The Line-to-DC supplies like the Meanwell will have a voltage-loop, so I would expect them to behave like this.
- sharing the supply between two filaments will of course give crosstalk. If the two filaments are cathode-biased, three will also be DC leakage current between the highest biased, and the lowest-biased side.
All these effects will add up to subjective degradation, compared to using a filament supply that does not interfere with the signal. There are plenty of reported experiences here on diyAudio, about this - but of course, one has to do a private comparison to see how it is for a given set-up.
Separately, Line-to-DC converters also leak common-mode current to safety-earth. The leakage can be 0.1 to 5mA depending on size and input voltage. This common-mode current is often quite noisy, up to high frequencies. Care must be taken that this leakage does does run through any audio cables, on its way to safety-ground. DACs do not take kindly to ground pollution. Connecting safety ground close to the negative side of the Line→DC unit helps.