• These commercial threads are for private transactions. diyAudio.com provides these forums for the convenience of our members, but makes no warranty nor assumes any responsibility. We do not vet any members, use of this facility is at your own risk. Customers can post any issues in those threads as long as it is done in a civil manner. All diyAudio rules about conduct apply and will be enforced.

Universal filament regulator

If you're driving multiple boards by the same input voltage, you could run some solid core wire between the boards. The connectors I use are through-hole.

Note that the filaments supplied by each transformer winding will need to be at the same potential. So if some filaments in the amp are required to float at different potentials (to avoid breaking down the filament-cathode interface for example), they will need to be supplied by separate transformer windings. Hence, they won't be able to share Vin -- or even ground.

In case of directly heated tubes, you'll need one regulator per tube. Just like you need one transformer winding per tube. If one end of the filament is connected to ground for all the tubes you're driving, you could run them on the same Vin but using separate regulators.

Does this make sense?

Do NOT run regulators in parallel to get higher output current. Use the LM22679 for up to 5 A output current. If you need more current because you have many filaments in parallel, separate them out on multiple supplies. If you need more than 5 A out, you'll need a different regulator.

~Tom
 
Last edited:
Because you're working with some high-impedance components (tubes), I'm not going to recommend you to use a switching power supply for filaments unless you're going to build a very good filter pack and place some shields - grounded.
By the way, why do you try to use a regulator? the tubes are already designed to work with a.c.
 
Because you're working with some high-impedance components (tubes), I'm not going to recommend you to use a switching power supply for filaments unless you're going to build a very good filter pack and place some shields - grounded.
By the way, why do you try to use a regulator? the tubes are already designed to work with a.c.

I don't like the intermodulation products that AC heating causes. Hence, my desire for DC heaters.

I've already implemented and tested switchmode supplies for tube heaters. They work well. See my arguments in Post #1 and my 300B Switchmode Supply thread.

~Tom
 
In case of directly heated tubes, you'll need one regulator per tube. Just like you need one transformer winding per tube. If one end of the filament is connected to ground for all the tubes you're driving, you could run them on the same Vin but using separate regulators.

Why couldn't you run two DH filaments in parallel if both 'grounds' are connected and it doesn't exceed the current limit?
 
This supply could be approximated by an ideal battery. If both negative terminals of two identical batteries are tied together and they are driving equal loads, what does it matter if I replace the two batteries with one that is twice as big and connect it to two loads at once? The load is going to be anchored to the same reference and at the same output voltage and the supply is ideally zero output impedance...
 
If you intend to tie both cathodes together then this is fine.There are very few topologies where this is acceptable. Two output tubes - not, PP output Yes, LTP-not, Multiple gain stages - not...

I'm really still not on the same page with you on this. I can think of ways to make all four of these work. In most designs, it doesn't work because cathode bias is frequently used. You can't share supplies because there is no common point to hang these supplies since each tube needs to develop its own voltage to bias itself (excluding shared cathode resistors, of course).

Specifically, Tom stated that "In case of directly heated tubes, you'll need one regulator per tube. Just like you need one transformer winding per tube. If one end of the filament is connected to ground for all the tubes you're driving, you could run them on the same Vin but using separate regulators."

If one end of the filament is connected to ground for all the tubes I'm driving(they are not cathode biased with individual cathode resistors), and they are equal filament voltage, and the regulator can put out the current, what exactly is stopping me from doing this? I know I just tied them together, but why is that a bad thing since I was going to tie them to the same place, ground, individually anyway?
 
If the regulator was indeed an ideal battery, SpreadSpectrum would be right. If one end of both cathodes is at ideal GND, then the other ends of both cathodes/filaments (assuming DHT here) could be tied to the ideal battery. The regulator is very close to an ideal battery -- at least at frequencies below the resonance frequency of the LC output filter (10-ish kHz). However, its output impedance is finite and non-zero. Hence the cathode currents from the two tubes could potentially interfere with each other. It really depends on the ground impedance as well as the output impedance of the regulator.

My advice would be that you need a separate regulator per directly heated tube. Or, say, you convert a design from AC heating to DC heating, you'll need one regulator per transformer filament winding.
For indirectly heated tubes, you can use one regulator to power several tube heaters as long as the heaters are to float at the same voltage.

Latest board incarnation attached. The multiple output caps are options. You'll see that only one is populated.

~Tom
 

Attachments

  • FilamentReg_R3p1_Schematics.png
    FilamentReg_R3p1_Schematics.png
    27.2 KB · Views: 527
  • FilamentReg_R3p1.png
    FilamentReg_R3p1.png
    56.9 KB · Views: 514
Last edited:
Thanks for the response, Tom. It sounds like you are worried about signal(or some undesirable noise/distortion) coupling between the filaments in the shared supply. Is that what you mean by the cathode currents interfering with each other? Please explain if I have this wrong.

I am planning on using two regulators for four 841s. Each regulator would see a total load of 7.5V @ 2.5A. These are fixed bias, sharing a negative filament supply reference. They are also CCS-loaded so there would be very little plate current swing, just plate voltage swing.

I guess I could do four regulators but I only want to do so if there would be a real benefit. What do you think?
 
It sounds like you are worried about signal(or some undesirable noise/distortion) coupling between the filaments in the shared supply.

More a coupling between the cathode currents. Let's say tube A runs at Ia AC anode = cathode current. This creates an AC voltage on the filament regulator output that's Ia * Zout, where, Zout is the output impedance of the regulator. If tube B's cathode is connected to the same regulator, this means tube B's grid-to-cathode voltage will have an AC component on it that's Ia * Zout. In addition to whatever AC component is in tube B's anode = cathode current.

I think the effect will be minuscule -- at least near DC, but I can't deny it's there. I would expect a significant effect at frequencies near the LC resonance.

I am planning on using two regulators for four 841s. Each regulator would see a total load of 7.5V @ 2.5A.

First off, I suggest using WebBench (National Semiconductor | High-performance Analog) to work out the inductance and Cout. Note that the DC resistance of the inductor as well as the ESR of the output cap play a role in the overall stability of the regulator. So you wan't to buy the right parts.

Do the tubes operate in parallel? If so, you could tie the two tubes that are in parallel to one regulator. I don't see a big issue with that as the cathode currents of the two tubes in parallel should be nearly identical.

I guess I could do four regulators but I only want to do so if there would be a real benefit. What do you think?

I think there would be a benefit, but I can't quantify how much. You could start with two boards and get another two if you don't like the results.

~Tom
 
Hope you plan on getting quite a few of these made. I want to put them into my OPUS and My Preamp.

I'm expecting to get 30~40 boards for the first run. I can't have too much money tied up in inventory... Sadly, due to the economics of scale, this also means that the boards will be a touch pricey. I'm expecting the cost to come in around $7.50/each.

I'm toying with the idea of offering assembled boards at a few different voltages as well. I haven't even begun the math on the pricing for those. Just giving you a heads up of my plans.

The lead time on the boards is about a week or 10 days. So should I run out of boards, more can be ordered and be here quickly.

I expect to submit the board for fab this weekend. I'll keep you posted.

~Tom
 
I thought I had this current coupling issue cleverly outmaneuvered since my plate load will be '10M90S cascode || 1.5M || 6pF.' Pretty high impedance, hence very little AC current swing to even worry about. However, this tube is going to require grid current to get the voltage swing I need so there will be cathode current bumps at low plate voltages due to grid current.

I'm using these because I need to develop 200Vrms+ at low distortion and low-gm DHTs tend to be linear across wide voltages (less curve bunching at low plate currents and high voltages). They develop the voltage swing to drive the 50% NFB output stage in a push-pull amp, so they are two push-pull pairs.

I think what I will do is try running a pair on one regulator. I will run a test applying a large signal to the first in the pair with no signal on the second. I will sweep through frequency-wise and watch for a peak in amplitude of whatever is there on the second tube's plate. Should be fun and interesting.
 
I thought I had this current coupling issue cleverly outmaneuvered since my plate load will be '10M90S cascode || 1.5M || 6pF.' Pretty high impedance, hence very little AC current swing to even worry about.

As long as the CCS has sufficient bandwidth. Yeah. That doesn't sound too far off. I'm still having some trouble accepting that you won't get any fluctuations in the cathode current, but I'm willing to be proven wrong on that.

If you're running a simulation of your amplifier, you could try including a circuit to model the output impedance of the voltage regulator in your sim. I would think that an LC parallel resonance with some series resistance would do the trick. You may have to have a large resistor in parallel with the L, C to tame the Q of the tank.

I'm using these because I need to develop 200Vrms+ at low distortion and low-gm DHTs tend to be linear across wide voltages (less curve bunching at low plate currents and high voltages). They develop the voltage swing to drive the 50% NFB output stage in a push-pull amp, so they are two push-pull pairs.

If they are P-P pairs, I would think that you might be able to get away with one regulator for two tubes. If the tubes were perfectly matched, I would think that any change in cathode current of one tube would be outweighed by a similar in magnitude but opposite polarity current in the other tube. Except for the Region of Trouble when the tubes enter grid current.

I think what I will do is try running a pair on one regulator.

Sounds like a plan.

~Tom
 
Sure, I would get fluctuation is cathode current, but they would be as small as I could reasonably make them, certainly nothing like what you would see in a power amp stage.

The 841 is an interesting tube, very linear if the curves in the datasheets are to be believed, but pretty high impedance and you need to drive grid current to get any useful signal output, so unfortunately I will spend a lot of time in the region of trouble. There was probably a better way to do this, but I really want a white hot filament in my amp, but didn't want to pay for 801As, which were a more ideal choice.