Proper Use of a Choke - Power Supply

[SkrillBill]

Hello,
this is my first post on the forum so i feel i must put the obligatory letter head on my post. I've been watching Dave for years, he's actually the reason I'm into electronics at all. Looking to make a career out of it eventually but for now it's a fun hobby.


My question is converting AC to DC. I will have 120v(North American) from my wall going to a 24v transformer. That will be passed into full-wave rectifier, and from there passed on to the rest of the circuit. My question is, do i need a choke after the rectifier, or is the rectifier enough? There will be filter caps on the rectifier as well, and my understanding is that chokes are typically used on the inputs more than outputs.

I'm building a very simple and basic bench power supply. I could just as easily use a pc PSU but i'd rather just make one for the experience. it will only be used to power small things like Arduino projects or other random creations.

Much appreciated!

[TimNJ]

In my experience, the bulk of linear AC-DC power supplies only use a capacitor after the diode bridge. There are probably a couple of reasons for that.

First, pros for LC:

An LC filter will give you a more tightly regulated output voltage, which might be important if you are not planning on using a regulator. It also lets you get away with using a smaller capacitor (than would be required for similar performance in a capacitor-only filter). That might be important if you are designing a very high current supply.

Cons for LC:

There is a voltage drop across the series inductor so your average voltage will be less. This may or may not be an issue. Probably the biggest con is that they are expensive and bulky. In general, engineers try to deal with inductors as little as possible.


I'd say for relatively low currents and when you are using a voltage regulator, like you likely are, a capacitor filter is generally good enough. As long as your minimum voltage across the filter capacitor is greater than your maximum desired output voltage + drop-out voltage of the regulator, you are covered.

[TimNJ]

Also, historically, capacitors used to be physically quite a bit larger compared to a modern cap of the same value. So, older gear might be more likely to use LC filters because cap technology was not as good (tube amps come to mind).

[SkrillBill]

There will be at least 1 regulator, maybe two if im feeling ambitious. Max output im wanting to shoot for is 30v @500ma as i can't imagine needing a full amp for anything, unless it makes more sense from a design perspective to just build it up to 1A.

So, no choke as long as i have filter caps and regulators. LC's aren't used as much anymore anyways?

[IanMacdonald]

Although, a disadvantage of just using one very large capacitor is that this puts much heavier stress on the transformer and rectifier for a given output current. This is because the capacitor only charges during a small part of the AC cycle when the rectified voltage is higher than the residual charge on the cap.

Thus a circuit which is powering a 1A load might be drawing current in short pulses of maybe 10A for a tenth of the time. Since P=I²R that will cause more heating than 1A continouus current. You therefore need to rate the rectifier and transformer somewhat higher than you might expect.

This is sometimes termed 'bad power factor' although it's not quite the same thing as power factor in AC circuits. The effects is much the same though, in that components which in theory ought to be adequate for the job get a mite sweaty under the armpits, so to speak. 

The choke extends the conduction time of the rectifier (inductors try to keep current flowing) and so reduces the strain.

In practice though, nobody uses chokes these days.

[IanB]

Using a choke (or inductor) as part of an LC filter after the rectifier is a good technical design. It provides better and more efficient filtering than reservoir capacitors alone. (The choke goes in series after the rectifier, feeding the capacitor.)

Old equipment needing an HT supply for vacuum tubes would often use this arrangement.

Downsides are that the required choke is large, heavy and expensive (uses lots of copper). So modern designs leave out the choke and pay the penalty in efficiency, given that a voltage regulator after the reservoir capacitor can fill the gap and provide a smooth and regulated output voltage.

If you are building a home brew supply and have a suitable choke to experiment with, you could try it and see how it goes.

[TimNJ]

Although, a disadvantage of just using one very large capacitor is that this puts much heavier stress on the transformer and rectifier for a given output current. This is because the capacitor only charges during a small part of the AC cycle when the rectified voltage is higher than the residual charge on the cap.

Thus a circuit which is powering a 1A load might be drawing current in short pulses of maybe 10A for a tenth of the time. Since P=I²R that will cause more heating than 1A continouus current. You therefore need to rate the rectifier and transformer somewhat higher than you might expect.

This is sometimes termed 'bad power factor' although it's not quite the same thing as power factor in AC circuits. The effects is much the same though, in that components which in theory ought to be adequate for the job get a mite sweaty under the armpits, so to speak. 

The choke extends the conduction time of the rectifier (inductors try to keep current flowing) and so reduces the strain.

In practice though, nobody uses chokes these days.

Not to get off topic, but I believe the power factor associated with the non-sinusoidal cap charging current is really the same thing as power factor in AC circuits. In purely AC circuits, all the waveforms are sinusoidal so the only thing that determines power factor is the phase difference between I and V. But more generally, the power factor is determined by the total harmonic distortion (THD) of the current wave and the phase difference. Because there is no distortion in purely AC circuits, then you can just neglect that part.

But that's proably besides the point. You do make a good point about reducing stress on the power components. That said, I think they parts are usually over-spec'd anyway.

[SkrillBill]

I planed to use some generic 1N4001's for the rectifier. Is the max current the rectifier can handle calculated as a sum across all the diodes, or is the max current still that of a single diode?

If i plan on using the 4001's I'll have to do 500ma. I for some reason had thought they were higher current than 1A.

@IanB:
You replied as i was posting. I have a couple chokes i pulled out of an older power supply. I don't know their specs or anything. My main question was if I needed one or not, which it sounds like i'll be fine w/o one. However i may take your suggestion of just adding one if i have one that works. This whole circuit is going to be a learning experience for me anyways and if it actually works that's just a bonus!

[TimNJ]

Each diode in a bridge rectifier has to be rated for whatever your max expected current is. Yes, 1N400x series is rated for 1.0A continuous, but they can handle much larger surges with no problem (30A). You don't have to use anything fancy but I would recommend using 1N4007 since they are literally 0.01$ more expensive and rated for 1kV peak reverse voltage.

[SkrillBill]

Each diode in a bridge rectifier has to be rated for whatever your max expected current is. Yes, 1N400x series is rated for 1.0A continuous, but they can handle much larger surges with no problem (30A). You don't have to use anything fancy but I would recommend using 1N4007 since they are literally 0.01$ more expensive and rated for 1kV peak reverse voltage.

I already have 1000 of the 4001s...

But i'll order them from DigiKey. I have to order some regulators too since I don't have any suitable ones. Kinda waiting to have a full schematic drawn up so i know what else i need too.

How frequent can the surges spike above 1A before it causes the component to fail?

[TimNJ]

Take a look at the datasheet: http://www.vishay.com/docs/88503/1n4001.pdf

You're going to get a current "peak" on every conduction cycle of 60Hz (actually 120Hz for full wave rectified). If you see the parameter I-FSM, you'll note that they specify an 8.3ms half sine wave input. Guess what the period of 120Hz is. (It's 8.3ms.) So basically it's telling you its okay to draw momentary high peak currents at 120Hz (up to 30A absolute max). The maximum current spec is really a function of junction temperature.

However, you will also note there is a square wave pulse current rating which the diode can only handle a number of times before it begins to fail. See Fig. 2.

You can also just buy a single package bridge rectifier.

EDIT: Hold that thought. I might have been misinterpreting the I-FSM spec myself. I believe that 8.3ms spec is actually just one 60Hz half-period pulse...Hmm...Maybe someone else has an idea. Sorry!

[SkrillBill]

Take a look at the datasheet: http://www.vishay.com/docs/88503/1n4001.pdf

You're going to get a current "peak" on every conduction cycle of 60Hz (actually 120Hz for full wave rectified). If you see the parameter I-FSM, you'll note that they specify an 8.3ms half sine wave input. Guess what the period of 120Hz is. (It's 8.3ms.)

So basically it's telling you its okay to draw momentary high peak currents at 120Hz (up to 30A absolute max). The maximum current spec is really a function of junction temperature.

However, you will also note there is a square wave pulse current rating which the diode can only handle a number of times before it begins to fail. See Fig. 2.

You can also just buy a single package bridge rectifier.

I was just thinking, "dont they make a package for this?". I'll be using Perfboard, and having all of those interconnects for the diodes would be messy.
Found this: https://www.digikey.com/product-detail/en/diodes-incorporated/KBP210G/KBP210GDI-ND/278625
I don't really know parts that well, but this one seems like a good enough fit. Looking at max specs it fits with what it's going to be expected to do in my circuit.

[tautech]

There will be at least 1 regulator, maybe two if im feeling ambitious. Max output im wanting to shoot for is 30v @500ma as i can't imagine needing a full amp for anything, unless it makes more sense from a design perspective to just build it up to 1A.

So, no choke as long as i have filter caps and regulators. LC's aren't used as much anymore anyways?

No, cost and bulk weighs against chokes in modern design.

These days bigger C's are used to get ripple to a level where the regulator looks after any residue but as others have indicated you must have sufficient headroom in current capabilities of the diode bridge.

CLC is often seen in switch-mode today.

[Circlotron]

With a 100VA transformer feeding a bridge and a big capacitor you can only expect to get about 60 Watts DC before the transformer is fully loaded. This is because of the greater heating effect of the short sharp current pulses from the very tips of the rectified voltage. For a small power supply this is no big deal but if you were doing one with a continuous rating of more than say 1kW then a choke would let you use a smaller transformer and filter capacitor and save money. Nowadays most big supplies are switch mode type though so most people have forgotten just how good filter inductors really are. Another useful feature is that if you have two supplies of the same rating and ripple voltage output, the one with the LC filter will produce a 100 or 120Hz *sinewave* ripple whereas the one with only a filter cap will produce a sawtooth ripple. If it is feeding sound equipment the sawtooth ripple is a much more audible buzz compared to the hum of a sinewave. Big deal back in the days of telephone exchanges being powered by a 54 volt 1000 amp supply.