Is constant filament current better way to build amps?

[Sunsun22]

I am working on a 6dj8 i/v part for a DAC and is looking for a constant current for filament. Since 6dj8 is using an individual filament, it is benefit to have DC? What about limiting the filament current 365mA? Does it have benefit to use constant current for filament?

[mach1]

A constant current filament source eliminates current inrush occuring at turn-on when the filament is cold, thereby extending tube life.

If using DC on a directly heated tube filament, a constant current source tends to sound better than constant voltage.

[Sunsun22]

In this respect, does it mean there is not much benefit in using constant current DC filament on 6dj8 since it is not directly heated tube? The only benefit is to limit the current when the tube is cold thereby extending the life of the tube, isn't it?

[vax9000]

I am curious about this claim "If using DC on a directly heated tube filament, a constant current source tends to sound better than constant voltage."

Is it because the filament/cathode is (AC) isolated from the filament power supply? I read that people claim AC powered DHT sounds better than DC powered DHT. Is it because of the same reason? Then, this opens some possibilities about wiring. Say, to reduce hum, One may use a pot between the two AC leads and the center tap, but not ground the center tap. Or, one can use DC to power the filament, but do not ground the DC supplier. Interesting claim.

vax9000

[EC8010]

AC heaters apply 50Hz (or 60Hz) plus about 5% distortion to the heaters. The problem with DC heaters is that rectification produces a spray of harmonics, starting with the second, and the trick is to stop that getting everywhere. I've seen a lot of very poor DC schemes - at the very least you need to follow the reservoir capacitor with a regulator. Another issue is that not only does the rectification noise go forward (to the heaters), it goes backwards, into the transformer and is coupled elsewhere, and bigger reservoir capacitors make it more of an issue.

[Eli Duttman]

Another issue is that not only does the rectification noise go forward (to the heaters), it goes backwards, into the transformer and is coupled elsewhere, and bigger reservoir capacitors make it more of an issue.

That's precisely why Schottky diodes should be used in DC filament supplies. Stop the "hash" before it starts.

[Eusebius]

I read that people claim AC powered DHT sounds better than DC powered DHT>>

Not necessarily - posts on this forum have stated a preference for a really good current source (multi-stage, Schottky diodes etc). This was referring to a 300b some time back.

[Wavebourn]

Constant voltage is needed. But if a voltage regulator starts increasingh a voltage slowly, it prolonges life of tubes. One way to bring it up slowly is to set up a current limit just above a normal working current, so while filaments are cold the current will be limited, but when they are hot the voltage is limited.

[poobah]

We are talking about some small picofarads between the filament and cathode. dV/dt, which 6 VAC has little of, is the noise source.

In the end, you either hear hum or you don't. I do agree that generating less diode hash in the first place is wise.

As far as Schottkys are concerned, yes, the waveforms look better. But the sharper corners imply spectral energy at higher frequencies. At the very freq's where all components begin to look like mongrels instead of R's, C's, and L's. Now jelly bean silicon diodes produce downright ugly waveforms, but, the spectral energy is distributed ats lower freq's... where normal components can "catch" them.

Not to counter Eli's advice by any means. Has any one here taken a look at the spectral output of different diodes? I would think that "soft recovery" might fair the best... after all, that is why they are built.

[Wavebourn]

Or buid it such a way so it brings up the voltage slowly (R1,C2).