If you need to get extra headroom.... you can use 2 FETs to make an "Active" rectifier.... real simple without extra parts..driving the gates from the transformer.... SiC FETs are best, problem is getting you hands on them is not easy.. Reverse leakage current on Schottky diodes is real bad with temperature coefficient....Unfortunately I cant use them in many applications I work in..
If you can work with surface mount, these look very nice for heater rectification... They only drop 95mV @ 10A!
LX2410A | Microsemi
LX2410A | Microsemi
Sorry for taking a while to respond to this. I appreciate all the input. I took a serious look at DC regulation for 7 12AX7s. That's 2.1A of heater current at 6.3VDC so quite a load. I actually did find a lot of options for regulators that handle this current in a single unit with low overhead, my favorite probably being the LT1764 so far.
However, the surge current on the diodes is insane. By the time the circuit has enough filtering before the regulator to give it .7v-.9v to operate with, the surge current is already >30amps for 7ms.
It would be nice if there was a non linear device like a varistor but for current instead that I could put in series between the transformer secondary and rectifier to suppress that surge current. Then I could put whatever caps in I want. I might try a mosfet or high power NPN with feedback from the filtering cap after the rectifier.
I realize I could get high current diodes but I want this on a PCB with, at most, 2mm 2oz traces. Honestly, I don't know if I can simply time average the current for the first second and get the right trace size, but I know 30-60 amp bursts isn't the best for it.
Wow I see you are thinking about the surge current and overhead as well hahaha. That is a nice diode? for this. I am limited by affordable PCB traces though. I am not sure how to determine the "burst current" that a PCB trace is capable of. Seems like they'd act like a fuse and just blow.
However, the surge current on the diodes is insane. By the time the circuit has enough filtering before the regulator to give it .7v-.9v to operate with, the surge current is already >30amps for 7ms.
It would be nice if there was a non linear device like a varistor but for current instead that I could put in series between the transformer secondary and rectifier to suppress that surge current. Then I could put whatever caps in I want. I might try a mosfet or high power NPN with feedback from the filtering cap after the rectifier.
I realize I could get high current diodes but I want this on a PCB with, at most, 2mm 2oz traces. Honestly, I don't know if I can simply time average the current for the first second and get the right trace size, but I know 30-60 amp bursts isn't the best for it.
Wow I see you are thinking about the surge current and overhead as well hahaha. That is a nice diode? for this. I am limited by affordable PCB traces though. I am not sure how to determine the "burst current" that a PCB trace is capable of. Seems like they'd act like a fuse and just blow.
Last edited:
You dont need diodes.... Use FETS instead to make an "ACTIVE" rectifier... There is a simple way of connecting the Gates to the transformer so you do not need a separate gate driver circuit...
If you do it the old-fashioned way with a 60 Hz power supply and/or linear regulation, yes, it is a fairly hefty load.
But times have changed. If you look around, we hardly ever see beefy linear power supplies any more. Instead, a little switch-mode wall-wart supplies clean, ripple-free DC for your devices. It weighs maybe one-tenth of an equivalent linear power supply, is tiny by comparison, and the DC output is already regulated. All this magic works because the frequency is tens of kilohertz instead of 60 Hz, and modern electronics makes processing this to DC easy and reliable.
So the answer to your problem is simple. Just use a small DC switching power supply for heater power. It's quite easy to find a 3A, 12V quite inexpensively. The voltage is already regulated, ripple is already small (and is above audio frequency, so the thermal mass of the heater wire ensures actual heater temperature has virtually no ripple.)
Wire the 12AX7s for 12.6V operation, and the little SMPS will power them happily.
If you are forced to wire the heaters in 6.3V mode for other reasons, then I suggest getting an SMPS that puts out a little more than 6.3V DC, and using a series dropping resistor. I use an old thrift-store Sony 8.4V DC power supply and a series resistor to power the 6.3V heaters in a guitar preamp.
-Gnobuddy
That's true. A SMPS circuit would be simpler than what I've put together to suppress the surge current. I've had pretty good luck with PCB layouts for 12VDC to 350VDC circuits. I assumed the switched mosfet would be getting loaded too much to put out 2-3 amps but at 12VDC output that's just not a big deal. At least at 12VDC it would 1.5 amps instead. I could just take taps straight of the mains then put in a varistor and fuse for safety.
Last edited:
I really dont think a SMPS is best option here... Even though I design them for a living for MIL and space applications.. You basically have a fixed load .... A well designed and robust SMPS will still need a controller chip, diodes and or FETS, Caps and a small transformer inductor..
Even some SMPS I have made switching at 2.2Mhz ....the magnetics may be small but the wire gauge needed will still require the magnetics to be big... It will be smaller, easier and quicker to use 2 FETs and a filter cap... to make an active rectifier...
Even some SMPS I have made switching at 2.2Mhz ....the magnetics may be small but the wire gauge needed will still require the magnetics to be big... It will be smaller, easier and quicker to use 2 FETs and a filter cap... to make an active rectifier...
Sadly, this is not always true; actually, I've mostly negative experiences with the cheap SMPS. Just now, I needed a 5V source to test/select some MOSFet for a capacitance multiplier, so I just grabbed a D-Link branded 5V 1.2A; guess what: unloaded: 5.19V; at 0.6A, 1/2 of the nominal load, it only delivers 4.62V. And this is true for most of the wallwarts I have around the house.
You're over-thinking. I have 5x12A?7 and 2xE80cc for a total of 1.35A @ 12V, and I use a ready made kit for $10 and it works fine, with the regulator screwd to the aluminum chassis. But it has over-spec'd rectifiers though.
Looks like that circuit hits about 27 amps for ~8ms in our application (7 12ax7 style heaters). It looks like they're using beefier rectifiers because of this and reducing the filtering cap accordingly but aren't worried about the PCB traces, although the only complaint about the board on ebay are the traces lifting. That regulator seems a bit noisier than the LT1764 I've sim'd up under the same conditions.
I'm probably overstating the impact of the noise haha.
I'm probably overstating the impact of the noise haha.
I wasn't suggesting a DIY SMPS, which I agree is beyond the capability of most home DIY electronics enthusiasts (including myself). Rather, I was suggesting an off-the-shelf one.
Most people have several wall-warts lying unused around the house, the thrift-stores are littered with them these days, and if need be, Digikey and Mouser stock lots of new ones.
Here's an example. 12V at up to 3A for just under $20 Canadian dollars (it'll be considerably less in the USA): PSM36W-120L6 Phihong USA | Power Supplies - External/Internal (Off-Board) | DigiKey
There are SMPS modules designed to mount inside your own housing, as well. This one is non-stocked (there will be a delay supplying one), but it's also 12V @ 3A for less than twenty bucks Canadian: LRS-35-12 MEAN WELL USA Inc. | Power Supplies - External/Internal (Off-Board) | DigiKey
-Gnobuddy
Honestly, I've never seen an SMPS that dropped as severely as that; perhaps it has ridiculously thin, long output wires? I've seen that defect on some very compact models; the output wires have enough DC resistance to drop a fair bit of voltage at full output current.
Even allowing for imperfect SMPS with poor load regulation, the reference for comparison is a 90-year-old engineering design: completely unregulated AC from a 60 Hz transformer!
The last valve power transformer I used (an old Hammond) put out 7.3 volts AC RMS at the heater windings, unloaded. I had to use a series resistor to drop it to 6.3V AC RMS for the valves.
So how do the cheap SMPS compare with this? They're better, right?
In North America, AC mains voltage has increased over the years, but many valve power transformers are still wound to vintage specs, and now put out too much voltage. Add in the manufacturer's attempt to compensate for voltage drop under load, and you can have way too much heater voltage, as in this case.
For use as valve heater power, poor load regulation from an SMPS isn't much of an issue - the load is fixed, so all you have to do is ensure it's correct when loaded. And any line regulation at all is better than the "classic" solution.
-Gnobuddy
You won't see the voltage drop with a multimeter because it just gets noisier as it degrades in regulation. You'd need an o-scope to see the 120hz noise.
Not at all: the MOSFet I mentioned will serve as a cap multiplier after the regulator board, mostly to improve the ripple rejection, about -60~70dB for most regulator.
Both Canada and Switzerland are free countries, so we're allowed to choose any faith we please. Me, it's a linear regulator where it matters NOT, as in heaters, and a LCLC filter where it matters, as in the HV supply.
But then I'm old... 🙂
Fair enough! 🙂
Back in the early 1990s, I remember building a power supply for my op-amp projects using a 7815/7915 pair of 3-pin linear voltage regulator ICs. I think that was actually the last time I used linear regulators. Life took some unexpected turns and I stopped doing any hobby electronics for a long time.
Nearly thirty years later, we have quite remarkable off-the-shelf products like this, which, for under sixteen bucks (CAD), is powered straight from the AC mains, spits out +/-15V DC at up to 340 mA, and is about the size of a single "C" cell: RAC10-15DK/277 Recom Power | Power Supplies - Board Mount | DigiKey
As another diyAudio member pointed out, it will even work on DC input voltage; so it could be used to generate +/- 15V rails for op-amp circuitry straight from the several-hundred-volt B+ rail in a valve guitar amp.
-Gnobuddy
- Status
- Not open for further replies.