DHT w/ DC heater - How to rig cathode bias?

[rongon]

I’ve been motivated to make some mayhem recently, and started to work on a line amp based on 45 DHT’s. I have a few that I won’t be using in any power amps, and I only need a gain of 3, so what the heck, one 45 per side for stereo...

Here’s the thing. I’ll be using DC on the heaters. I want to take the 5V tap on the power transformer, using a full-wave diode bridge to make a single DC supply of about 5.6V, then use one LM317 to regulate down to the 2.5V, 1.5A required for each 45.

I’m new to this, so I have a pretty simple question. How should I (or should I) ground the +2.5V filament supply so that it doesn’t interfere with the cathode bias voltage? I always get into trouble with DHT’s, because the cathode IS the heater, so I have trouble visualizing the forces acting on them as being separate.

I found this thread, DHT DC heater question, and I understand the right-most solution is the best one, in this picture:

(.

It looks to me like the heater supply floats while the cathode is grounded.

I whipped up this schematic to illustrate my question:



In the lower right, where I ask where that point goes, is that left floating? No grounding?

I'm sure I could be missing something here, and I don't want to do anything stupid. Thanks!

-=|=-


PS - When you're figuring the cathode bias, does the +2.5V DC of the heater supply add to the voltage from cathode to ground? Or do you ignore that because it's floating on top of the cathode bias?

--

[Frank Berry]

Leave it floating or connect it to chassi ground through a NP capacitor.

[tomchr]

The filament supply floats on top of the cathode bias voltage. As Frank says, just leave your ground connection in question floating. However, beware that you need a separate transformer winding for each DHT tube -- just as you would if you were to use AC heating.

~Tom

[rongon]

Thanks once again Tom.

The filament supply floats on top of the cathode bias voltage.

So they're separate? In other words, if I'm aiming for a -12.5V bias on the DHT (grid to cathode), and if I put two zener diodes in series to elevate the cathode +12.4V from ground, then my job is done? The +2.5V heater supply doesn't influence the bias voltage?

Or should I elevate the cathode +10V, and the +2.5V from the filament supply gets added on top of that to put the filament/cathode +12.5V above ground?

beware that you need a separate transformer winding for each DHT tube -- just as you would if you were to use AC heating.

I was afraid of that. You answered a question that I didn't know enough to ask... and I was going to try using the one 5V winding for both tubes. Good catch!

There is a bright side to that, though. Since I'm thinking of using 45's, If I use an LM317 (Imax = 1.5A) for each 45 (2.5V, 1.5A filament) then that should act as a current limiter on the filaments, yes? I get a little current regulation with my voltage regulation, right?

-=|=-

[quikie22]

Originally Posted by rongon

There is a bright side to that, though. Since I'm thinking of using 45's, If I use an LM317 (Imax = 1.5A) for each 45 (2.5V, 1.5A filament) then that should act as a current limiter on the filaments, yes? I get a little current regulation with my voltage regulation, right?

-=|=-

I would not put it that way. You will be running the 317 at its max specs and risk repeated thermal shutdown or total destruction. Please don't ask how I know.

MAybe you should look at some offerings from your fellow forummers... Rod Coleman comes to mind and his well received DHT heater.

[tomchr]

Originally Posted by rongon

So they're separate? In other words, if I'm aiming for a -12.5V bias on the DHT (grid to cathode), and if I put two zener diodes in series to elevate the cathode +12.4V from ground, then my job is done? The +2.5V heater supply doesn't influence the bias voltage?

You're making the bias adjustable, I hope...

With, say, 12 V on one end of the cathode and 14.5 V on the other end, one could argue that the average Vgk is 13.25 V. However, the electrons don't flow directly from one spot on the cathode to the anode. The electrons are emitted to a space-charge cloud that surrounds the cathode and attracted by the anode. I doubt the +/-1.25 V will make any significant difference in the current flow when compared with the anode potential of hundreds of volts...

In addition, tube-to-tube variation will cause one tube to draw more current than another for the same Vgk. Hence, my hope that you're making the bias adjustable.

How much anode current are you running through those 45's? Can the zener diode survive this current? Recall, that zeners have a max power dissipation rating as well. This rating is usually given for a case (or junction) temperature of 25 deg C. At the dissipated power, they won't be that temperature, hence the max power rating needs to be de-rated...

Originally Posted by rongon

There is a bright side to that, though. Since I'm thinking of using 45's, If I use an LM317 (Imax = 1.5A) for each 45 (2.5V, 1.5A filament) then that should act as a current limiter on the filaments, yes? I get a little current regulation with my voltage regulation, right?

The National Semiconductor version of the LM317 should be able to survive 10 years of operation at 1.5 A --->IF<--- the junction temperature is kept within specified limits. Frankly, I think that's pushing your luck. Also, what if one tube is drawing 1.65 A? That's only a 10 % increase. I'd go with a 2~3A regulator. Regardless of which (linear) regulator you use, you will have to provide it with a rather large heatsink. In fact, I did the math on the regulators I was going to use for my 300B (1.5 A @ 5 V). To guarantee operation under worst case conditions, the regulators would have to burn more power in the heatsink than was delivered to the speaker. I thought this was ridiculous and started looking at switchmode regulators. I'm getting really good results with those. Good enough in fact that I decided to build a universal filament supply. I'll offer boards for sale on my website once I'm done. As others point out, Rod Coleman makes a DHT regultator. As does Guido Tent (TentLabs).

~Tom

[rongon]

Hmmm... Good points.

I checked out the threads on the switched-mode filament regulators and Rod Coleman's current regulated filament supply. All very interesting, but maybe more than I wanted to get into with this project.

I should mention that my little project is meant to be an experiment. I'm hoping to check out what a DHT line stage might sound like, without investing a fortune in parts, chassis, iron, silicon+heatsinks, etc. I have an old chassis, appropriate power transformers and choke, a few 45's.

Re: Regulator power dissipation - 2.5V * 1.5A = 3.75W dissipated by the heater. Assuming a 5.6V input and a 2.5V output, that's a drop of 3.1V in the regulator. Multiplied by the current draw of 1.5A, that's 4.65W. 5W or so through the heatsink isn't too horrible, is it? (Or did I figure that wrong? Probably was wrong...) In any case, I think I'm switching to an LM350, if I do stick with this.

Re: Adjustable bias - I wasn't thinking of that. I was only planning to dissipate about 2.4 watts in each 45. Plate at 120V, cathode at -12.5V. I figured the tube would draw in the vicinity of 20mA.

I wonder if I should scale down my goal to a stab at a pair of 12B4A's with CCS plate loads...


-=|=-

[tomchr]

Originally Posted by rongon

I should mention that my little project is meant to be an experiment. I'm hoping to check out what a DHT line stage might sound like, without investing a fortune in parts, chassis, iron, silicon+heatsinks, etc. I have an old chassis, appropriate power transformers and choke, a few 45's.

Oh... Well that does change things. My calculations of power dissipated in the filament regulators were based on worst case conditions. I want the amplifier to work. Always. For a slap-together test setup, I don't mind cutting a few corners. I've fed 300B's from an LM317 mounted on a monster 0.4 K/W heatsink. I was feeding the LM317 about 15 V.... So a good 15 W of power dissipation. But with the big heat sink it wasn't an issue. For a test setup this works, but I wouldn't want to integrate a 30x15x5 cm heat sink into a tube amp just for the filament regulators.

Originally Posted by rongon

Re: Regulator power dissipation - 2.5V * 1.5A = 3.75W dissipated by the heater. Assuming a 5.6V input and a 2.5V output, that's a drop of 3.1V in the regulator. Multiplied by the current draw of 1.5A, that's 4.65W. 5W or so through the heatsink isn't too horrible, is it?

5 W is pretty manageable to dissipate. Your math is correct. It does, however, assume that you have 5.6 V available...

Originally Posted by rongon

In any case, I think I'm switching to an LM350, if I do stick with this.

Good call.

Originally Posted by rongon

I was only planning to dissipate about 2.4 watts in each 45. Plate at 120V, cathode at -12.5V. I figured the tube would draw in the vicinity of 20mA.

Why not just apply the CCS to the 45 anode? You'd still need the zener (or other voltage reference - LEDs come to mind!) in the cathode, but you'd have 100 % control over the anode current.

~Tom