12V regulator IC that can power 6 preamp tubes (900mA) FROM COLD STARTUP?

[Dave H]

It worked!?! KEwl!!

I think the explanation is simply that as OD said, the regulator senses that its output voltage is bigger than 0.000 - by about 0.6V, and that's enough to convince it that it's not trying to power a dead short. That makes sense to me.

Neat trick. I've not seen that before.

I think the reason Wombaticus didn't get it is because OD actually said 'diode junction resistance' not diode forward voltage drop.

[kg]

honestly I've never been a fan of hard regulation on heater supplies: 3055 based cap multipliers have worked very well to reduce ripple even in ridiculously high gain designs.

[Wombaticus]

It worked!?! KEwl!!

I think the explanation is simply that as OD said, the regulator senses that its output voltage is bigger than 0.000 - by about 0.6V, and that's enough to convince it that it's not trying to power a dead short. That makes sense to me.

Neat trick. I've not seen that before.

I'm unconvinced R.G. that the regulator could detect that...don't forget that I added another diode to the ground pin to compensate for the diode drop on the ouput. So the device is not referenced to ground to begin with. In fact there are now *two* 1N4148s on the ground pin, because I already had one there to bump the regs output to something closer to 12.6V...

[Gregg]

Did you try the diode from output to ground only as in the datasheet?

[Gingertube]

Not to hijack this thread, but the topic is related.

Can I put the filaments of two power tubes in series and use 12V transformer to power the filament? Mainly I am referring to 6L6 and 6V6.

Alan,
Yes you can do this.
Just one example: JADIS JA80 HiFi Amp (80W Class A). 4 x KT88 in parallel push pull. The 2 KT88 on each side of the push pull have their heaters in series across a 12.6 Volt AC supply. Frankly it might have been smarter design to pair a KT88 from each side of the push pull so that a failure would also shutdown a tube from the other side. Wiring would have been a little trickier.
Cheers,
Ian

[Wombaticus]

Did you try the diode from output to ground only as in the datasheet?

Not yet, but I will do when I have a chance.

[Dave H]

I’d still like to know how OD’s diode trick works. If it can’t be the diode forward voltage drop because it still works with two diodes in the regulator ground leg and it can’t be the diode slope resistance because that is an order of magnitude lower than the cold heater resistance then how is it working?

[g1]

Is it about the circuit resistance before the diode turns on?

[kg]

what's the schematic of the final (working) circuit?

[jpfamps]

I've retro fitted a simple soft start circuit to regulated filament supplies to counter this problem.

I use a MOSFET in series with the LT supply and use a RC network on the gate of the MOSFET to gradually pull the MOSFET fully on.

I've either used a N-channel MOSFET on the negative rail and pull the gate up to the +ve rail, or a P-channel MOSFET in the +ve rail and pull the gate doen to the negative rail.

The ramping time can be adjusted by varying the time constant in the RC network.

High power low voltage MOSFETS are cheap. The on resistance can be very low with the correct MOSFET.

Alternatively the soft start circuit shown by Mick also works very well.

[Wombaticus]

Did you try the diode from output to ground only as in the datasheet?

So I tried this today, but it didn't work. Although I realize now that it might not have been a very good experiment, since while I did short out the olddawg series diode on the output, I didn't remember to short out the second 1N4148 diode on the ground leg that I added at the same time.

Incidentally, in the datasheet figure that Gregg is referring to, the shunt diodes he's talking about are part of a split supply with a regulator on both the positive and negative rails. They are intended to just keep the negative supply from pulling the output of the positive regulator negative if it happens to start up first (and vice versa). While there's no negative supply in my case, maybe the fact that I put a voltage adjustment diode in the ground leg makes it harder for it start up with a heavy load.. it would be like the regulator seeing a voltage on the output below its "ground" reference. So really I need to do the experiment a few more ways.. once with all diodes shorted, to see if it can start up all six tubes at 12V. If not, add a shunt diode to the output and try again. Finally, add back a single diode in the adjust leg, and see if the shunt diode makes a difference in that circumstance. I'll report back once more.

[Wombaticus]

what's the schematic of the final (working) circuit?

Here's how it is now.

[Gregg]

...I didn't remember to short out the second 1N4148 diode on the ground leg that I added at the same time.

It doesn't matter. I also have a diode there every time.
I think that the issue you're having is specific to the regulator you're using. As I said before I'm using 12V6 regulated (78XX) heaters all the time and never had this problem. Most of the time my arrangement is: bridge rectifier (sometimes Schottky) -> 3300uF cap -> 78XX regulator -> 3300uF cap. A 4007 in the ground leg. Number of preamp tubes 5 or 6.