AC to DC power [solved]

[abishur]

I was testing out an AC->DC circuit I built. It's a no muss, no fuss full wave rectifier. AC comes in, DC goes out. I then hooked up my voltage step down circuit, seems to be working. I take out my cheap-o multimeter, set it to DC, put one lead on the plus, the other lead on the negative (the correct way I promise ) and BAM! the voltage step down pops out of existence. Now I don't care too much about the why it did it. It was a cheap multimeter, I shouldn't have been lazy and gone and fetched my fancy one that I actually know how to 100% work.

Obviously I had a sudden in-rush of current and the little step-down circuit can only handle 2A max. The bridge rectifier has a max of 6A (overkill I know, but it's what I have on hand) I tried putting a resistor on there but that got hot to the touch almost immediately (I put 26V AC into the diode rectifier, 24V DC comes out) My question is what it the best way to limit the current of the DC output to 2 amps? Is there a better way then putting a resistor in the mix?

[aTao]

Something a bit missing here.. Like are you sure the meter was reading volts, not amps?

Also if you put 26Vac into a bridge rectifier you (should) get 35Vdc out.

[abishur]

Oh, I just had a thought. Rather than putting the resistor between my 24V DC source (the diode rectifier) and my voltage step down, it would probably be better to put it after I drop the voltage down to 5V as that's a whole lot less voltage to constrict. Still is there an even better answer? I suppose as long as I know what I'm attaching to it, do I even need to worry about putting it there in the first place?

[abishur]

Something a bit missing here.. Like are you sure the meter was reading volts, not amps?

Also if you put 26Vac into a bridge rectifier you (should) get 35Vdc out.

Yep, I promise the multimeter was set to read volts.

You know I also thought I should get a lot more Voltage out, but I assure you (and if you really want I can verify by posting pics ) that the numbers I provided are what is actually going on in there.

[aTao]

In simple PSU design it is better to "know" what load will be connected and slightly over spec the current supply to suit. A current limit resistor will cause poor regulation.
If you want a cheap, dirty current limit then pick a transformer with appropriate VA rating. The saturation in the magnetic circuit will act as a power limit. Note though that it will get hot and possibly burn out the windings, but it will provide much better limiting than a resistor.

[abishur]

In simple PSU design it is better to "know" what load will be connected and slightly over spec the current supply to suit. A current limit resistor will cause poor regulation.
If you want a cheap, dirty current limit then pick a transformer with appropriate VA rating. The saturation in the magnetic circuit will act as a power limit. Note though that it will get hot and possibly burn out the windings, but it will provide much better limiting than a resistor.

Thanks aTao, I had a feeling as much. I have no idea why that stupid multimeter blew the circuit. To be fair, I had the voltage regulator sitting there with no load and only wanted to test it again because I was smelling something weird, it's possible some metal shard got on the step down circuit and the leads were just the proverbial straw. Que sara!

[Jim Manley]

Hi Matt!

Are you using any large electrolytic capacitors to smooth the output of the rectifier? Otherwise, you're actually delivering 120 Hz sinusoidal pulses where the voltage varies from zero to full voltage. Depending on the capacitive and/or inductive impedance of the load, all sorts of mayhem can ensue. Also, how are you "stepping down" the output of the rectifier - those terms imply using a transformer, but you're already doing that from 120 volts. If you're using a regulator, there are small-value conditioning capacitors that are needed for stability, or just the kind of behavior you described can occur. If you're just using a resistor off the rectifier, note that its value will drop significantly as it gets really hot. Use a wattage and composition appropriate to the current load, e.g., a high-wattage ceramic wire-wound resistor, or your rectifier might cause a recti-fire! Also, if the current will vary due to the load, the voltage drop across the resistor will change, which may affect the current through the load, which will change the voltage drop across the resistor ... can you see where this is headed? The time dimension is often not accounted for by innocent bystanders, but there are major areas of science, engineering, and industries based on it and it turns out we engineers really do get paid to do something useful occasionally

[gritz]

Remember also that what is commonly referred to as (in your example) 26V a.c. refers to the RMS voltage (this is what aTao is referring to). When you apply that to a full wave rectifier you get 26 * 1.414 = 36.8V peak (minus a diode drop or two). Apply this to a reservoir capacitor and it'll fill in the troughs between the peaks (more or less, depending on the size of the cap and the amount of current you're drawing from the supply).

So, *very basically* your regulator may be seeing nearly 1.414 (square root of two) times the voltage that you think it is. Power supply design is deceptively complicated. Remember that volts dropped across a resistor or linear regulator * current through that component = heat in Watts, so you also need to know how many Watts your components will happily dissipate. Maybe the regulator will need heatsinking, etc. Rather than starting from first principles it may be worth hunting out a few tutorials on the web. Let's keep that magic smoke where it belongs!

[abishur]

Hi Matt!

Are you using any large electrolytic capacitors to smooth the output of the rectifier? Otherwise, you're actually delivering 120 Hz sinusoidal pulses where the voltage varies from zero to full voltage. Depending on the capacitive and/or inductive impedance of the load, all sorts of mayhem can ensue. Also, how are you "stepping down" the output of the rectifier - those terms imply using a transformer, but you're already doing that from 120 volts. If you're using a regulator, there are small-value conditioning capacitors that are needed for stability, or just the kind of behavior you described can occur. If you're just using a resistor off the rectifier, note that its value will drop significantly as it gets really hot. Use a wattage and composition appropriate to the current load, e.g., a high-wattage ceramic wire-wound resistor, or your rectifier might cause a recti-fire! Also, if the current will vary due to the load, the voltage drop across the resistor will change, which may affect the current through the load, which will change the voltage drop across the resistor ... can you see where this is headed? The time dimension is often not accounted for by innocent bystanders, but there are major areas of science, engineering, and industries based on it and it turns out we engineers really do get paid to do something useful occasionally

Yeah, the resistor is gone, threw that concept away by the third post

Technically, no the rectifier portion of my circuit has no capacitors, but I do have a nice capacitor on the input and a second one on the output (and some other stuff since it's adjustable) on my step-down regulator. I haven't thought to check my voltage after the input capacitor, but before the step-down voltage regulator, but I've tuned it such that the output is giving me my nice 5V and after testing it on some disposable USB devices (mouse, blutooth headset, etc) it's been happily running my model A pi.

Side note, I went with a switching regulator and haven't see a lot of heat build up, I'll probably still stick a heat sink on it just to give me peace of mind for these hot Texan summers.

[Burngate]

Is it just me?
All these words and narry a piccie anywhere?
Just one diagram would help me work out what you're doing!

[abishur]

Is it just me?
All these words and narry a piccie anywhere?
Just one diagram would help me work out what you're doing!

I'll take a pic in a bit, but the truth is y'all already answered my question from my OP. I just wanted to know about forcing a current limit with a resistor and if that was a good idea or a bad idea. Answer: It's a bad idea for a variety of reasons including:

1) Bad design practice (it's better to know how much draw you'll be using and not exceed the capabilities of the unit
2) Can result in excessive heat if done improperly
3) Depending on my load, I could create a condition where I was getting wacky voltage / available current conditions

and the list goes on

As for the end application:

I have a 24V AC power transformer (Though it technically reads ~26V AC), it goes into my 3A max full wave rectifier/diode bridge/whatever-your-favorite-thing-to-call-it-is, according to my multimeter (not the cheap one, I think I'm going to return that one, I went and grabbed my fancy provided by my work place) I now have about 24V DC.

From this I plug in my voltage regulator circuit. I had made one, it was huge and did the job, but I swapped it out for a pre-made one of the same design that's about 1/6th of the size so it will work better for my designs . It's an adjustable voltage regulator, so I got the voltage set right and did a couple of test to make sure everything was working right and have been running my pi successfully for a bit now

[timmoore46]

It is enormously easy to plug the multimeter red lead into the 10Amp socket and cause then difficulties !

A good way to learn basic electronics, building PSU's ! But the chance of cooking you RPi is very real !

A tip is draw out a circuit diagram and then work out maximum currents and power dissipation in each component especially the regular (if you have one).

stating with 34V DC for 5V or 3.3V output is a good way to get something very very hot !

34 v -3.3 v = 30.7 v to get rid of. So at 1 Amp = 30.7 Watts ! Ouch ! but at 100mA that is only 3 W so care is still needed.

: )

Tim

[wartstew]

I agree that stepping 24V down to 5V using a "linear" (analog) regulator, will mostly make a lot of heat!

If that transformer has a "center tap" (third "middle" wire on the secondary winding), then you might try using it to do a "full wave" rectifier circuit using the center tap wire which then would use only 1/2 of the bridge rectifier (or just two diodes if you wish) in a circuit that would only produce 12 volts DC to the regulator (at twice the capable current), and thus would be a lot more efficient. It would produce less heat and *could* provide more current. I say "could" because you might be using an LM317 or similar linear adjustable regulator that limits current at 1.5 Amps (unless it has to limit it further to save itself from overheating) anyway.

Really, when dealing with low voltages at high currents, you really are off using a "switching" power supply design, buy such things are harder to "home brew".

[abishur]

I agree that stepping 24V down to 5V using a "linear" (analog) regulator, will mostly make a lot of heat!

I fully agree, which is why I'm using a switching regulator

[Jim Manley]

When the Apple ][ came along in 1977, one of the many differences between it and legendary "boat anchor" S-100 systems (as Byte "Chaos Manor" columnist and science fiction author Jerry Pournelle liked to call them), such as the MITS Altair series, was the then completely-novel idea of using a switching power supply. To the uninitiated, they were strange, off-gold, anodized aluminum boxes with perforated holes to allow air to circulate and that provided a glimpse of a mysterious array of exotic components that made up an oscillator and also included a triac and a bizarre-looking small toroidal transformer. They really did look like alien technology to a population of knuckle-draggers used to buying ever-larger iron-core, wire-wound, step-down transformers and related rectifiers, capacitors, etc., to generate the +5, +12, and -12 volts at numerous amps consumed by pre-CMOS (complementary metal oxide semiconductor), pre-low-power Schottky, TTL (transistor-transistor logic) inverters, gates, counters, etc., used in microcomputers of the Pleistocene era.

The list price on those sleek, lightweight, switching puppies was around $120 in 1977 cash, which would be closer to $600 in today's inflation-scarred Samolians. That assumed you had connections to be able to buy them in single-unit quantities from a shady character in a back alley behind a bar that catered to the employees of the only major industry in Cupertino at that time, the quarry on the North edge of town (even HP hadn't been so hard-up for real estate yet that it had needed to build a campus in Cupertino, which is now the site of Apple's new "flying saucer" HQ that is under construction and should be opening next year). That the same functionality in a package half the size of a draftsman's eraser is now available for a few dollars (dimes, in 1977 money) is just astounding beyond comprehension, sorta like what the Pi is.

Arthur C. Clarke was right, "Any technology, sufficiently advanced, is indistinguishable from magic."

[gritz]

I'm old enough to remember when switching power supplies would routinely sacrifice an output transistor in order to save a few pence worth of fuse - and I'm not even that old (well, I suppose it's all relative...)

Progress isn't always a bad thing.

[abishur]

As promised here's a pic of the setup



My AC input comes in from the left, DC 5V output goes out through the right. I've temporarily attached a usb extender to the end for testing purposes.

[cyrano]

Using a simple diode instead of the full-wave rectifier will produce lower voltage and put less strain on the regulator.

I'd still add a buffering capacitor (500-1000 microfarad), since the one on the regulator board is only 220 microfarad.