Audio Asylum Thread Printer
Get a view of an entire thread on one page
|For Sale Ads|
I bought a few cheap switchers through eBay recently to see how they might work for filament supplies. The 12V models are a no-brainer; just turn the adjustment to 12.6 VDC output, and you're done. The 5V / 4A unit I bought didn't have enough adjustment, so I opened it up.
Looking at the parts, the adjustment pot is 200 ohms. The pot is in series with a 4.7K SMT chip resistor, 0603 size. I assume this is typical of most units. After some experimentation, I replaced the resistor with a leaded 3.3K 1/8W. I was able to solder it between the pot and another pad away from the original resistor, so it was a perfect fit down against the board. The supply now adjusts from roughly 5.6-6.4 VDC. As a final test, I connected the output to a 1.5 ohm resistor bank and let it run for an hour or so. The case wall used as a heatsink was only lukewarm after that time, so I think the units will be OK for this use.
I really dislike the cheap Chinese case these are in, so I'll make new ones if I put any of these into actual use. Will also replace the electrolytics with name-brand 105° caps from Mouser or DigiKey. That should improve reliability.
Now I just have to decide whether to use these in new builds, or to stick with the linear approach. A transformer/rectifier and LM338 costs more and dissipates more heat (and takes up more space), but it's hard to beat for long term use. I have LM317s working great in test gear more than 25 years old. It's not clear whether these little switchers can be expected to survive so long.
Buy Chinese. Bury freedom.
Yesterday I tested a few switchers and an eBay DC-DC buck converter. The converter is rated at 8A, and it uses a different regulator than the more common 3A units that sell for a dollar or two. I still need to try those, but they haven't arrived yet. The converter I used is adjustable for both output voltage and current limiting and is pictured below. I used a 811A tube for all the tests. The filament draws 4A at 6.3V, and cold resistance is very low.
The first test was a 5V/4Amp switcher that I've converted to 6.3V. When AC power is applied, it produces a series of three short output pulses, then it starts and heats the tube. Total time to stabilize is about four seconds.
The second test was a 12V/2Amp switcher driving the buck converter. I adjusted the converter in advance to 6.3V and set the current limit just above 4 Amps. The switcher wouldn't run properly in this situation, even with the converter's current limiter reduced to near-zero. Output of the switcher would slowly creep up to a maximum of 2 or 3 volts over a period of 10 or 15 seconds. That was enough to get the buck converter working a little, and the 811A would light dimly. Despite the heated filament, the switcher's output would not increase further.
Next, I tried a 12V/5A switcher driving the buck converter. This worked fine, but the switcher is much larger than the other two. It won't fit on a 2" sidewall like the others, so it would take up a lot of space under the chassis.
Finally, I powered the buck converter from a linear 10A bench supply operating at 15V. This also worked well, and it provided the opportunity to test efficiency. Comparing input power to output, I calculated the efficiency at 85%. This agrees with the datasheet for the regulator IC used in this particular converter. It also means the 12V/2Amp switcher would need to output 2.5A to heat the tube, even after the filament is warm. So, in addition to the failure to start, that's simply the wrong way to get this done.
A single large switcher is more difficult to mount than two small ones, and this seems to be leading the charge in determining the best way to resolve this. I now think the two best approaches to dual voltage heating are A) two small switchers, a 6.3V and a 12.6V, or B) a linear 12.6V supply driving a 6.3V buck converter. A third possibility would be a 6.3v/4A switcher (same small size as the 12V/2A version) along with a boost converter to power a few 12V preamp and driver tubes. The best configuration will obviously depend on how much current needs to be supplied at each of the two voltages. In any event, I'm very encouraged by these results. They do seem to indicate that no linear, heat-producing regulators will be needed in order to power all the 6V and 12V tubes on a single chassis.
Incidentally, 120 Hz ripple from these switchers was less than 20mV pk-to-pk, and HF/RF noise was even lower. No external output filtering will be required to use any of these devices for heating.
Buy Chinese. Bury freedom.
Hi Triode Kingdom,
I like your approach. I have repurposed switchers for various purposes but I just get SMT resistors of the value which I need.
You may have to over size your switcher to start properly with a tube heater load. Tube heaters draw ~ 10X the normal current when cold.
I'm making a crystal pickup phono stage for a ~ 1946 78 changer I'm restoring. I was going to use a 12AU7 as a cathode follower. I have a sealed board mount 12V, 0.5 A switcher with which I was planning to light the tube connected for 12 volts. Since the switcher is potted I planed to just live with 12 volts. The 150 ma heater load caused the switcher to make several repeated attempts to start. Each try heated the heater somewhat so eventually it got there. This would work but seemed flakey to me. I was thinking of some active soft start circuit.
I found a cute little dual triode 6111 which I just had to use. This tube has a single 6.3 V, 300ma heater. I still want the 12 volts for a the cathode bias for the cathode follower. I tried the 6.3 Volt, 300 ma heater with a dropping resistor from the 12 V 500 ma switcher and it started on the first try.
I haven't decided yet on the best way to handle the startup surge. An oversize switcher defeats the purpose of keeping everything small. The switcher has to fit under the chassis, so its size is more critical than a filament transformer mounted topside.
I'm trying to focus more now on alternatives driven by the electrical/mechanical requirements, rather than starting with the solutions themselves. For example, let's say a raw supply (transformer-rectifier-filter) is used to provide DC filament power. That saves under-chassis space, but the output needs to be regulated down in order to create the right voltage. That's where the problem comes in. The regulator usually controls a fairly large voltage drop, often at several amps, so it gets hot and needs a heatsink. There goes the additional space freed up by locating the transformer topside. In my case, the issue is even more difficult, because I want the output to be switchable from 6.3V to 12.6V. That means the raw supply must be at least 15V, and the drop to 6.3V will create a lot of heat.
From a problem-solving perspective, the best solution might be to combine a raw supply using a topside transformer with a small buck converter under the chassis. The buck converter could be programmed to provide 6.3 and 12.6 VDC from the same raw supply, and because it's so efficient, no heatsink would be needed. The only thing is, I don't know whether the typical buck converter will start the tubes when they're cold. I'll test that on the bench soon.
Buy Chinese. Bury freedom.
Hi Triode Kingdom,
It is true the grossly oversizing will cut into the advantage of the switcher.
I sure did not want to throw a damper on this idea. There may not even be a problem. First off if the switcher just current limits at its rated current as opposed to shutting down and restarting again the voltage across the tube heaters will rise steadily with no drama. Even my 500 ma PS would eventually (tens of seconds) light the 150 ma heater. For each successive attempt the tube heater would be a bit hotter and drop a bit more voltage. If I wasn't watching the voltage rise with a scope I would have been unaware that the PS was having to make several repeated attempts to start. Actually it may have been fine. I just didn't like it.
Even if your switcher shuts down and retries as mine as opposed to current limiting there are solutions. The most obvious would be to have a series resistor limiting the current at the supply's rating and switching it out in a few seconds as Chip647 suggested.
An other approach would be to follow the switcher with a low drop linear regulator with current limiting. The switcher output would be set to the desired voltage plus the drop out voltage of the LDO linear. Linear Technology makes a variety of LDO linears.
Yes, if only one voltage is required, the combination of a switcher and LDO regulator (with current limiting) is probably the smallest package and least heat. It's also very inexpensive. The need for 6.3V and 12.6V on the same chassis changes all that. In my case, I want to be able to use 6V and 12V output and driver tubes. I also want to be able to switch between 6V and 12V versions of the same output tube. Current draw with a quad of 6V outputs will be double that of 12V tubes, so a linear regulator doesn't seem optimal for this use.
I only see a couple options here. One would be to use a separate, small switching supply for each voltage. That would require a delay circuit to be sure the supplies start reliably when the tubes are cold. I'd like to avoid that additional complication if possible.
The second option would be a power supply with at least 18 VDC output driving a DC-DC buck converter. The buck converter could reduce the voltage to either 12.6V or 6.3V, selectable by simply switching a resistor. If a linear supply is used, nothing else would be required. Use of a switcher might require a LM338 wired as a current limiter between the supply and the buck converter. The downside here is that 12.6V and 6.3V wouldn't be available simultaneously, and some of my upcoming designs will need that.
As it now stands, I'm leaning toward the use of a single supply and two buck converters to get this done. The converters are very inexpensive,
only a few dollars each, and also very small if 3A or less is needed. Either a linear supply or a small 24V/2A switcher could be used. The switcher might still require a current limiter, but it's also possible the self-limiting of the buck converters will be sufficient.
Looks like I have more experimenting to do...
Buy Chinese. Bury freedom.
or just a manual load switch. Soft-Start and stand-by are always nice.
Think of the progress that has been made with these in the past 5 years or so with the switchers. In the next 5 years there will be much greater improvement which will allow for an upgrade path. LED lighting is driving a lot of the R&D
Post a Message!
This post is made possible by the generous support of people like you and our sponsors: