• 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

Member
Joined 2009
Paid Member
Folks,

I've been building this 300B amp for the past year or so. I want to use regulated DC for the filaments. I have done the math on linear regulators and found that I will end up dissipating more power in the filament regulators under worst case conditions than the amp will deliver to the speakers. I find that ridiculous... Needless to say, I started dabbling in switchmode regulators. They actually work really well...

So I was thinking. Wouldn't it be nice with a universal switchmode filament regulator? Universal, as in programmable by a resistor change to different output voltages. I'm thinking to support 2.5 V for 2A3 and the like, 4.1 V for the Russian tubes, 5.0 V for 300B, 6.3 V for just about everything, and 12.6 V for the tubes with series filaments.

Are there any other common filament voltages that should be supported?

I'm thinking to use an LM3102 as it allows me to hit the above voltages without too much fuzz. Its max output current is 2.5 A (just enough for a 2A3 I think). It's also in a package that's hand-solderable by most humans.

I'm throwing this out there to get the process started and to gauge the level of interest in such a project.

~Tom
 
Member
Joined 2009
Paid Member
Tom, were you going to start with a smps plugpak with regulated 12V output, and use the buck for lower voltage options?

Nah... I'm not a fan of the plugpak/wall-warts. I figured I'd use a small-ish 12 V transformer -- or two 6.3 V heater windings in series to power the filament regulator.

In the ideal world, the regulator would run off of one 6.3 V winding. However, if you do the math, you'll figure out that by the time you account for voltage ripple, regulator drop-opt, etc. under worst case conditions it just is not possible to make 6.3 V regulated DC from a 6.3 V AC winding. Unless.... Unless you go into a buck/boost converter topology. But even then switchers tend to not operate at their optimum point when Vin is close to Vout. I considered using a regular voltage doubler to get 12 V from a 6.3 V winding, but with the currents involved, I'm not sure it's such a hot idea.

~Tom
 
Member
Joined 2009
Paid Member
Start out with a 12.6V source, FWB gives you around 18V which is plenty of headroom to get the efficency up above 90%.

Yep. That's the plan.

I just looked at buck/boost converters. Between National, TI, ADI, and LTI, only LTI has a buck/boost converter that can deliver more than 2.5 A. It's $9/each at Digikey and requires a bunch of external components. With the LM3102 I think the total parts budget will come in under $9 (plus the PCB of course).

So sticking with the plan. 12 V AC in, step-down to the filament voltage.

~Tom
 
I had a similar plan as well. Except mine had modules using a flyback so the output is isolated from the source. Have them all be auto ranging 85-265 vac input and dip switch selectable for the output. Build an all DHT amp with these and only need a high voltage winding...
 
I had a similar plan as well. Except mine had modules using a flyback so the output is isolated from the source. Have them all be auto ranging 85-265 vac input and dip switch selectable for the output. Build an all DHT amp with these and only need a high voltage winding...

I built a project like that for my final project in college. Except, I wanted +/-45 and +/-60 V out for a class AB sand power amp. It worked pretty well.

But the more I look at mains supplied switchmode, the more I realize that sourcing the magnetic components is a royal pain. Single SMD inductors used in the low voltage regulators are plentiful - and fairly inexpensive, but transformer cores and bobbins needed for the high-voltage supplies are hard to come by. They were hard to come by in the 1990'ies, but now even more so.

This filament regulator will not use custom parts. Just off-the-shelf stuff.

~Tom
 
I have found the same problem with high power class d amps.
I now wind my own.
In most cases it works out cheaper than buying in specials.

I was referring to the cores and bobbins needed for winding my own magnetics when I commented on the lack of availability. For anything but the simple IC regulated supplies, I would plan on winding my own magnetic components. But I have yet to find a good source for the parts. The vendors I have found haven't had any data on what they actually had for sale.

The low-voltage regulator ICs are so much easier to work with as the magnetic components needed are available commercially and the components are generally well spec'ed.

~Tom
 
National has a part out that works for buck/boost (Zeta) with a Constant On Time regulator which is discussed in their POWER designer application note #130.

http://www.national.com/assets/en/appnotes/national_power_designer130.pdf

The design is for a fixed output 600ma regulator, but since it has an external pass transistor and uses a resistor divider for feedback it should be straight forward to change it to selectable output and greater current capacity(bigger FET and Inductor?). Minimum V-out is 1.25 (limited by the internal ref). But could be made lower by using acouple of tricks like post regulation (drop voltage with a series schottky diode).

How far can one go on the reange of voltage boefore the inductor has to change?
 
National has a part out that works for buck/boost (Zeta) with a Constant On Time regulator which is discussed in their POWER designer application note #130.

Nifty...

How far can one go on the reange of voltage boefore the inductor has to change?

There are quite a few variables to play with. Inductance, (Vin-Vout), Vout, switching frequency, inductor current ripple, etc. I'll need to set up a spreadsheet and find a minimum number of inductance values that will work with the output voltages listed. Hopefully, I can reduce it to just one inductance. Given that National makes the LMZ-series with the internal inductor, I'm guessing I'll be able to find a good compromise.

~Tom
 
Hi Tomchr

Interesting thread. Have you seen the SM regulator modules from TI
particularly the PTH/PTN series. They are ready built modules and
look like they could cover the requirements of most DHTs. Only problem
might be that while they have a soft start feature Im not sure if it is
"soft" enough and I dont know if it can be altered. Might be other
problems I havent thought of.
 
Have you seen the SM regulator modules from TI
particularly the PTH/PTN series. They are ready built modules and
look like they could cover the requirements of most DHTs. Only problem
might be that while they have a soft start feature Im not sure if it is
"soft" enough and I dont know if it can be altered.

I hadn't seen those. They're pretty neat. However, not all that flexible and actually rather expensive. Not only are the modules expensive, they also require two expensive external ceramic caps (100 uF, X5R is not cheap). The ripple voltage is also higher than I'd like (1 % of Vout). But it is pretty impressive that they can crank 45 W out of a small module like that. 95 % efficiency... Nuttin' wrong with that.

Usually "soft start" means it takes about 1 ms for the regulator to start up... :) I'm planning to extend that to about 5 seconds. - Though, if you want it to be different, just change the soft-start cap... In my experience, a 5 second start-up is enough to avoid hitting the current limiter when powering up into a cold 300B filament.

The PTH series goes up to 5.5 V - but only 2.25 A so not enough for a 2A3, but enough for the other low-voltage tubes. The PTN-series can deliver enough current (3 A), but only go down to 3.0 V, so again the 2A3 lose out...

This is exactly what I ran into in the previous incarnation of my filament regulator. The National LMZ-series are awesome and have rock solid performance, but they only go to 5.5 V, so the 6.3 V tubes will need a different regulator. The regulator I used for the 6.3 V couldn't deliver the output current needed by a 300B, so I ended up with two different designs. Nothing wrong with that, but it did make me wonder if it wasn't about time that someone threw together a design that could handle all the voltages. I'd much rather have one design that support all the needed voltages with a simple tweak of the passives.

~Tom
 
It turns out I can support the range of common tube voltages (1.5, 2.5, 4.2, 5.0, 6.3, 7.5, and 12.6 V) with a single inductor value. For some of the output voltages, I may have to increase the switching frequency from the nominal 500 kHz to keep the inductor ripple current within reasonable limits (250-ish mA). But that's easily done by swapping out a resistor.

Latest schematic and board layout are attached. The component values on the schematic are for 5.0 V out. The board is etching as I type... Just a quick prototype before I commit more serious money to a fab run.

~Tom
 

Attachments

  • FilamentReg_1p0.png
    FilamentReg_1p0.png
    48.9 KB · Views: 1,454
  • FilamentReg_1p0_Schematic.png
    FilamentReg_1p0_Schematic.png
    40 KB · Views: 1,541
First hack...

Folks,

The first results are in. I assembled the circuit as shown on the schematic, except I omitted D1, C1~C5 and just fed the circuit from a lab supply. I also didn't have any 68 kOhm resistors in 0603 package for R2, so I used 47 kOhm. As a result, I had to accept a higher switching frequency of approx 800 kHz versus the 500 kHz I designed for. Oh, well...

At Vout = 5 V; Iout = 2 A, the regulator shows a few mVpp of ripple voltage. The start-up time is about 5~6 seconds. Not bad for a day's work, eh?!

~Tom
 

Attachments

  • FilamentReg_1p0_Proto.jpg
    FilamentReg_1p0_Proto.jpg
    290.5 KB · Views: 1,388
  • FilamentReg_1p0_Ripple_5V_2A.jpg
    FilamentReg_1p0_Ripple_5V_2A.jpg
    182.5 KB · Views: 1,322