Toroidal transformer secondary voltage.

[davelectronic]

Not sure about this as I've never tried it yet. Is it possible to get a regulated 12 volts power supply from a 12 volt toroidal transformer. Using a fairly large capacitor bank, a low dropout voltage regulator, and a shottky bridge rectifier. Oh and with the voltage drop of a series pass power transistor for extra current and power. Thanks for reading thoughts appreciated.

[George Edmonds]

Hi

Max DC from 12VAC is 16.98VDC, then start subtracting voltage drops.  You did not say current required, but why two voltage regs?

G Edmonds

[wraper]

Depends on particular transformer (how much it sags under the load), how much you load it and how much mains voltage sag it must accept. Note that transformers with large voltage sag often have quite a bit more voltage than nominal under no load as voltage normally is specified under load.

Oh and with the voltage drop of a series pass power transistor for extra current and power. Thanks for reading thoughts appreciated.

Avoid that if you need to minimize voltage drop. It would be better to just make an opamp based regulator rather than beefing up an LDO.

[T3sl4co1l]

Not really, if you need to include low-line condition and modest output ripple.  (You can use excessively large capacitors to get slightly higher unregulated output with less ripple, but the power factor goes down as well; and it's already about 0.5 at typical values.)  In the average case, sure.

Preferably you wouldn't have a boost transistor in there (presumably you mean a PNP in the usual connection?).  Fortunately there do exist LDO controllers, somewhat of a boutique item, but they are out there.  See this for example: https://ww1.microchip.com/downloads/en/DeviceDoc/mic5156.pdf

Notice you will need a higher supply voltage for the reg, since it's driving an N-ch (follower) pass device.  This is easily implemented on a transformer supply by making a half-wave doubler from one winding to "on top of" the +16V supply.  Use a small-ish series capacitor so the supply has a modestly high impedance, and use a zener for one of the diodes to clamp peak voltage -- a C-dropper supply without being direct mains fed.  (Or just let it full double the voltage, if it's within ratings.  At 16V (32V when doubled) out of a 36V rating, I think I'd rather toss a zener in, simple enough.)

Tim

[themadhippy]

Max DC from 12VAC is 16.98VDC

rubbish,from 12v AC i can give you almost any DC voltage you would care for

Is it possible to get a regulated 12 volts power supply from a 12 volt toroidal transformer. Using a fairly large capacitor bank, a low dropout voltage regulator, and a shottky bridge rectifier.

have a search for voltage doubler rectifier

[davelectronic]

I'm understanding most of your descriptions, but the voltage doubler would take to high for the current I'm looking for. That's approximately 15 Amps at 12 volts. I know 12 volts RMS ends up at 16.968 Volts after the capacitor bank. The 12 volts regulator is a LM2940 CT and I was going to find some high power shottky diodes for the bridge rectifier. Yes there PNP series pass transistors to boost the current to 15 Amps, 5 Amp max per transistor. The transformer is a vigatronix toroidal transformer, 2 X 115 primary and 2 X 12 volts secondary. They should be a bit over 12 volts RMS when powered on. I've got quite a few 10000uf 25 volts 105 °C electrolytic capacitors for the capacitor bank. So if I've got this right, there's 0.5 Volts drop for the voltage regulator, about 1 volt for the bridge rectifier, and 0.7 volts drop for the 3 X parallel series pass transistors. So a total voltage drop of around 2.2 volts. This should just about be enough over head voltage for the circuit to deliver the 12 volts regulated output....

[themadhippy]

2 X 115 primary

even though its supposedly 230v if your in the uk its gonna be more like 240+v so that should gain you an extra 1/2 volt or so

[T3sl4co1l]

LM2940CT is one of those archaic LDOs, the early ones with "low" as in "not two diode drops" (though it does happen to have a proper Vce(sat) as the limit, not a diode drop like LM1117), and all the stability and current limit issues (but, at least it doesn't suffer from excessive GND current in dropout).  Pay close attention to the required output capacitor ESR.  And, I have no idea what boosting will do to stability; probably safe to assume it'll make it worse.  Notice they don't suggest such a connection in the datasheet; while that's not an active prohibition against it, it's not promising.

Hm, I wonder if that's any different, as far as stability in boosted configuration, with modern (low-ESR compatible) LDOs.

Tim

[tunk]

I think a rule of thumb is to use ~2000uF/A and that this will
give you 4-4.5V ripple: 2000*15*4.5/(16.9-14.2)= 50000uF
(Don't know how power factor will influence this.)

[Doctorandus_P]

There are lots of factors, it's not so straight forward.

Nominal transformer output voltages are (usually?) specified under full rated load current, and at a low current output, the voltage will be higher. This can be a few percent (for a big transformer) to 20% or more (for a small transformer). And of course it also fluctuates with mains voltage changes. Just now I measured 232Vac on a mains socket (so quite close to nominal 230Vac here in the Netherlands) but it can be up to something like 253Vac, or dip to something like 210Vac. Such limits vary by country.

Most basic rule for a capacitor is:
I = C *dV/dt

Mains voltage here is 50Hz, so a half cycle is 10ms
So if you pull a current of 1A out of a 2200uF capacitor, it's discharge over a half cycle is:
dV = I*dt/C = 1*1e-2/(2200e-6) = 4.54V

So that is quite close to tunk's "rule of thumb".

[Doctorandus_P]

So if I've got this right, there's 0.5 Volts drop for the voltage regulator, about 1 volt for the bridge rectifier, and 0.7 volts drop for the 3 X parallel series pass transistors. So a total voltage drop of around 2.2 volts. This should just about be enough over head voltage for the circuit to deliver the 12 volts regulated output....

That seems quite optimistic.
A bridge rectifier has two diodes in series with the load, and even for Skottky's 500mV drop is quite low at 15Arms current (and maybe 45A peak current). You can add a shitload of bulk capacitors. this will reduce ripple, but at the cost of ever growing peak currents, and you may saturate the transformer during those peak currents. Also, resistive (copper) loss is quadratic with current. so 45A peak will have a 9x higher copper loss, on top of saturation problems. But anyway, those peak currents will reduce the output voltage of your transformer further.

---------------------------------
It's not saturation, but overload problems I think, I'm not too well versed in transformer details.

---------------------------------
Also, another thing to consider:
1. Use the voltage doubler (Only one diode in series so already less loss).
2. Use a moderate capacitance.
3. Add an SMPS circuit to get 12Vdc.

It's higher efficiency both for the SMPS and the more even load for the transformer, and you avoid problems with headroom.

[T3sl4co1l]

A bridge rectifier has two diodes in series with the load, and even for Skottky's 500mV drop is quite low at 15Arms current (and maybe 45A peak current). You can add a shitload of bulk capacitors. this will reduce ripple, but at the cost of ever growing peak currents, and you may saturate the transformer during those peak currents.

Ap bap bap*--- saturation is a magnetization thing, not a load current thing.  Load current doesn't affect primary voltage (at least, not in an unbalanced way, given the full wave rectifier).  Indeed, flux density decreases as load current increases, because of primary DCR.  Which is a small effect, for the most part.

*Um, hmm, I have no idea if there's a preferred transcription for that... it's a, not really a turn of phrase, more a sound.  I imagine it's a simplification of something like "ah, but but--".  ...Things I'd never have thought about, but for foreigners/ESLs being present..!

Also regarding transformer impedance / regulation / efficiency, toroids are generally quite good at that, so even if it's a small transformer (and it sounds like it's not), I'd expect pretty good regulation, even while using large filter capacitors.  Again, assuming you don't mind the low power factor.

It's not saturation, but overload problems I think, I'm not too well versed in transformer details.

Yup, just RMS current to worry about here

1. Use the voltage doubler (Only one diode in series so already less loss).
2. Use a moderate capacitance.
3. Add an SMPS circuit to get 12Vdc.

It's higher efficiency both for the SMPS and the more even load for the transformer, and you avoid problems with headroom.

Note that doublers draw worse PF as well; at least for the same output ripple.  Though if capacitance should be "moderate", that should work out favorably overall?  For certain values of "moderate".  I see your point.  It's not like the full doubling is needed; a lot more ripple means higher PF and (somewhat) lower DC out.  Which, being that DC would be "too high" here, that's advantageous.

I think at that point, I'd be more worried about the capacitor(s) themselves -- electrolytics even at line frequency tend to be rather unhappy with more than say 10% ripple.  Polymers might handle it fine?  They're rather pricey though, for the kind of values needed here, nevermind film or ceramic...

SMPS can also be a SEPIC type, allowing boost or buck output (freely and continuously above or below input) -- but any SMPS is probably a bit more advanced than OP is looking for here, unfortunately.

There's also the boost/buck (transformer) strategy, where you take a low voltage transformer in series with the main winding to make a smaller change to it.  Well, in this situation, good luck finding say a 2Vrms transformer, right? -- but, being it's a toroidal, there might be some room to add windings around the thing by hand.  Hacky, but it's probably in the couple-turns-per-volt range, so, I'd guess less than a dozen turns would do, and, just use whatever hookup wire is rated for the (RMS) current required.  Does make mounting it a bit annoying -- avoid crushing the added wires.

Tim

[davelectronic]

Ok, I see it's not an easy straight forward answer that I was hopping for. The transformer is 300 VA toroidal type transformer. I've never made a voltage doubler, so would have to look this up on how it's done. I'm not aware of other LDO voltage regulators, so would have to research some suitable types to use. Here in the UK I've never measured less than 240 Volts AC, but it might dip lower at peak times of day. My maths isn't the best in the world, I can work out most basic formula with a bit of help from the internet. I'm unsure about different scenario power factors of different completed circuits. I know I can build this type of circuit with an 18 volt RMS secondary transformer, because I've done it several times, with good results. But I've never tried to get the same regulated voltage as the Volts RMS of the intended transformer I'd plan to use. Thanks for all your replies.

[wasedadoc]

Also, resistive (copper) loss is quadratic with current. so 45A peak will have a 9x higher copper loss, on top of saturation problems. But anyway, those peak currents will reduce the output voltage of your transformer further.

That statement about "quadratic" and "9x higher" is misleading in the context of voltage loss. Power loss is quadratic with current but voltage drop from copper loss is close to linear with current until the current is high enough that the heat generated increases the copper resistance.

[Doctorandus_P]

That statement about "quadratic" and "9x higher" is misleading in the context of voltage loss. Power loss is quadratic with current but voltage drop from copper loss is close to linear with current..

Yes, you have to keep the distinction between voltage and power loss clearly separated to avoid confusion.

The whole SMPS idea was probably not so bright either. I was thinking along the lines of buying some pre-made module, but those shiny plated rectangular SMPS bricks with holes are at such a price point that it is probably a better option.

About the voltage doubler... There are quite a lot of ways to combine a transformer, diodes and capacitors to make a DC voltage. The attached circuit (found in 10 search) is one of the simplest.
The capacitor alternately charges C1 or C2 to the peak output voltage of the transformer.

[Gyro]

Ok, I see it's not an easy straight forward answer that I was hopping for. The transformer is 300 VA toroidal type transformer. I've never made a voltage doubler, so would have to look this up on how it's done. I'm not aware of other LDO voltage regulators, so would have to research some suitable types to use. Here in the UK I've never measured less than 240 Volts AC, but it might dip lower at peak times of day. My maths isn't the best in the world, I can work out most basic formula with a bit of help from the internet. I'm unsure about different scenario power factors of different completed circuits. I know I can build this type of circuit with an 18 volt RMS secondary transformer, because I've done it several times, with good results. But I've never tried to get the same regulated voltage as the Volts RMS of the intended transformer I'd plan to use. Thanks for all your replies.

Don't bother with the voltage doubler, it's a very good way of squandering more than half of your transformer's secondary current rating and landing yourself with a lot of voltage overhead to shed. It will need some very chunky (more than double the planned value) and rather stressed electrolytics at that sort of current (basically you have two caps in series, which are 'pumped up' alternately at 50Hz, rather than 100Hz as you would with a bridge rectifier).

Don't rely on the 240V mains these days either. Our supply step changed from 250V down to 230V (228V - 233V observed) without any warning few weeks back, when the DNO switched our local area to a lower output tap on the substation transformer. You really want a solution that works across the specified tolerance rainge (230V - 6% +10%) to avoid sudden disappointment.

I think this is one of those too marginal cases where you need to carefully re-evaluate all of your requirements and choices, maybe go SMPS for instance? I don't think you specified what sort of load this is for.


EDIT: T3sl4co1l's suggestion of an overwind on the secondary is probably the most practical fit for what you're intending.

[wasedadoc]

The attached circuit (found in 10 search) is one of the simplest.
The capacitor alternately charges C1 or C2 to the peak output voltage of the transformer.

The diodes and capacitors arrangement is the equivalent of the input circuit of a PC PSU with a 110/220 Volt switch set in the 110 position. See schematics in the third reply at https://electronics.stackexchange.com/questions/479800/why-do-some-smps-power-supplies-require-an-input-voltage-select-switch

[Doctorandus_P]

A very simple and effective technique is to just add some extra windings to your toroidal transformer. The only downsides are that it "looks ugly" and it costs some time. Just put the extra wire in series with the existing windings. (note: you can either add or substract, depending on polarity). (Oops, I see now that Gyro mentioned that T3sl4co1l mentioned this too.  )

It is also a good exercise to get some hands-on experience with transformers. But do make sure your wire is thick enough to handle your current.

============
And there are other demons lurking too.
One of them is that when you turn the thing on, the empty capacitors form a near short for your transformer. This can lead to very high primary current peaks. They can be enough to blow the mains fuses in your house. For toroidal transformers over 500VA, a "switch on current limiting" circuit is almost mandatory, for smaller transformers it's often omitted. Your 300VA transformer is in the region where this may or may not be necessary. Such a circuit varies from a very simple NTC to a power resistor (10 to 30 Ohm or so) in series with the primary winding and that gets shorted by a relay after a few seconds. audio power amplifiers often use this circuit and schematics are easy to find.

[T3sl4co1l]

One of them is that when you turn the thing on, the empty capacitors form a near short for your transformer. This can lead to very high primary current peaks. They can be enough to blow the mains fuses in your house. For toroidal transformers over 500VA, a "switch on current limiting" circuit is almost mandatory, for smaller transformers it's often omitted. Your 300VA transformer is in the region where this may or may not be necessary. Such a circuit varies from a very simple NTC to a power resistor (10 to 30 Ohm or so) in series with the primary winding and that gets shorted by a relay after a few seconds. audio power amplifiers often use this circuit and schematics are easy to find.

This, however, is also exacerbated by saturation!  Toroids are notorious for inrush due to saturation, when the switch is closed near a zero crossing (which happens randomly).  An inrush limiter (salvage an NTC resistor from some random power supply, if nothing else) is good cover for both effects.

Tim

[Doctorandus_P]

Some 15 years ago I had a gadget that had a low power standby mode (which kept big elco's topped up) and a high power mode in which a 600VA 12V transformer had to be switched on quickly and reliably with a solid state relay.

The solution I finally settled on was to measure zero crossings with a micrcocontroller, and then determine a (fixed) phase shift to turn on the solid state relay. When I got it completely wrong, the 16A fuses in the house blew. I had a 1A wirewound (slightly inductive?) shunt resistor to measure inrush current, and it just topped out at the mains voltage. So it was either 300A peak or something wrong with my measurement setup. When I got it a bit wrong the transformer hummed heavily for a few seconds before it settled.

When I got it right (I think ideal phase delay was about 3ms.) I still had heavy humming sometimes. It took me a while to realize the humming was about half of the times the thing was turned on. Solution was to turn the transformer on during a rising mains flank, and turn it off on a falling flank, and after that it worked excellently. so it turned out that the remanent magnetic field in the transformer core was enough of an extra phase shift to change saturation effects on turn-on significantly.

Transformers look simple on the outside, but it needs a surprising amount of work to get something like that just right. If I had known that in advance, I would have used a regular inrush current limiter.

[davelectronic]

I won't bother with a voltage doubler, I don't know if to chance 300 VA 12 volts transformers, or parallel two 160 VA 15 volts transformers. I get that extra few volts at the expense of a little less space in the case. But two 160 VA 15 volt transformers would fit. A few posts back mentioned what it's for, it's for HF radio equipment, so needs to be reasonably electrically quite when in use. I can find some NTC inrush thermistors for current limited during turn on. My knowledge doesn't stretch to microcontrollers, but that seems overkill for a linear power supply.

[Vovk_Z]

Is it possible to get a regulated 12 volts power supply from a 12 volt toroidal transformer.

My own statistics says it is 50% possible. If you are ok to have a bit smaller regulated output voltage, e.g. 11.0-11.5 VDC then chances are higher. Or, possibly you have oversized low-impedance transformer, or you may use syncronous rectifier instead of Shottky diodes.

[T3sl4co1l]

When I got it right (I think ideal phase delay was about 3ms.) I still had heavy humming sometimes. It took me a while to realize the humming was about half of the times the thing was turned on. Solution was to turn the transformer on during a rising mains flank, and turn it off on a falling flank, and after that it worked excellently. so it turned out that the remanent magnetic field in the transformer core was enough of an extra phase shift to change saturation effects on turn-on significantly.

Transformers look simple on the outside, but it needs a surprising amount of work to get something like that just right. If I had known that in advance, I would have used a regular inrush current limiter.

I even once had a situation where, ... well let me set it up.  Low frequency inverter -- a couple kHz, I forget what range exactly (maybe 1-3kHz?); induction heating application.  Transformer was output matching (and isolation) between inverter and resonant tank.  At the low frequency, a stripwound iron core was chosen (thinnest gauge GOSS available). First test, assumed pulse width was balanced (H bridge inverter) and wired it directly to the inverter.  Well, DCR was low enough, and mu_eff high enough, that didn't work of course.  It mostly worked, but the peak current at turn-off was pretty significant, and only on one edge, because of saturation.  It took some seconds to drift over into saturation, but would inevitably saturate.

The weird thing is, after it was first magnetized in this way, it never recovered: it would always... hmm wait now what was it, it would still start randomly (depending on turn-off + ringdown phase; initial phase on the controller was always 1/4 cycle however) I think, but always drift back towards saturation to one side (less severe than before, at least), despite having installed a coupling capacitor after that first test.  It wouldn't recover magnetization on sitting (IIRC?), but only while driven at AC.

Whereas a "virgin" transformer not subject to saturation, was always well behaved, quiet running and balanced.

While I hate the term "flux walking", which normally isn't any kind of strange dynamic process, just the literal DC balance in the circuit; this truly seemed to be some asymmetric, dynamic effect, for which "walking" might be an apt name.

So, there's some weird magnetization history stuff, and remnant and permanent magnetization all going on here. Ferromagnetic materials are -- to be precise -- as a physicist -- weird.

NiZn ferrites also typically carry a warning not to saturate them, or cause mechanical stress (force, shock, etc.).  Ferrites in general have some stress response, where mu can go up or down depending on relative stress and field orientation, there's weird stuff like that too. (And not just because of magnetostriction; I mean change in mu_i.)

Tim

[IanB]

approximately 15 Amps at 12 volts

Seriously? Why do we keep seeing absurdly high currents and linear voltage regulators in the same paragraph?

You don't just bung a few capacitors and and LDO after a transformer to get a regulated 15 A supply. Linear power supplies that can provide 15 A at 12 V have the size and weight of a dozen bricks and have cooling fans.

If you were asking for 1 or 2 amps, this thread would make more sense.

[PCB.Wiz]

I'm understanding most of your descriptions, but the voltage doubler would take to high for the current I'm looking for. That's approximately 15 Amps at 12 volts. I know 12 volts RMS ends up at 16.968 Volts after the capacitor bank. The 12 volts regulator is a LM2940 CT and I was going to find some high power shottky diodes for the bridge rectifier. Yes there PNP series pass transistors to boost the current to 15 Amps, 5 Amp max per transistor. The transformer is a vigatronix toroidal transformer, 2 X 115 primary and 2 X 12 volts secondary. They should be a bit over 12 volts RMS when powered on. I've got quite a few 10000uf 25 volts 105 °C electrolytic capacitors for the capacitor bank. So if I've got this right, there's 0.5 Volts drop for the voltage regulator, about 1 volt for the bridge rectifier, and 0.7 volts drop for the 3 X parallel series pass transistors. So a total voltage drop of around 2.2 volts. This should just about be enough over head voltage for the circuit to deliver the 12 volts regulated output....

You need to also include the ripple and the winding resistance, as well as some mains margin.
Just wire something up on the bench, and load it up.

As you increase the total C to lower ripple, you increase the peak current demanded from the transformer and diodes, and that means more droop.

So if I've got this right, there's 0.5 Volts drop for the voltage regulator, about 1 volt for the bridge rectifier, ...

You can also consider MOSFET ideal bridges, as your overhead here is tight...
 
https://www.analog.com/en/parametricsearch/11518#/

Another helper, for tight designs, is to generate a 20V rail from a simple charge pump, that is used as a regulator bias rail.