Smoothing capacitor value ?

[Radio Scotland]

Putting together a 12v / 4-amp adjustable power supply using a full wave rectifier, What's the best value to use for the smoothing cap ?



Cheers
Jay

[murphyv310]

OK, so what is the circuit? Simple PSU with bridge and smoothing cap with tappings on mains TX or resistive dropping and unregulated or with adjustable regulation or regulation and active smoothing?

A basic non regulated will need between 4700uf and 10,000uf, suitably rated cap for no load, minimum of 25v, 35v advised.

[Radio Scotland]

Originally Posted by murphyv310

OK, so what is the circuit? Simple PSU with bridge and smoothing cap with tappings on mains TX or resistive dropping and unregulated or with adjustable regulation or regulation and active smoothing?

A basic non regulated will need between 4700uf and 10,000uf, suitably rated cap for no load, minimum of 25v, 35v advised.

Centre tapped transformer with two rectifying diodes, Going to add an adjustable LM-338 Regulator, It's actually an old car battery charger rated at 4 amps


Cheers

[Radio Wrangler]

Simple answer:

It depends.

If you have a rather small smoothing capacitor, your circuit will discharge it further between peaks of the mains voltage. So you get more ripple. On an unregulated supply, this translates to more hum.

A regulated supply will take off the ripple UNLESS the smoother voltage falls so low that the regulator can't handle it. This can be very bad.

If you opt for a massive smoothing capacitor, it won't discharge much between peaks of the mains. This means that the mains waveform will be almost at the top before the rectifier on the positive-going side turns on and starts to recharge the smoother. This means that the smoother is recharged in a very short amount of time. The same amount of charge has to go in, in order to drive the load for the period between peaks, and this in turn means that the charging current is dramatically increased.

So with too big a C, the transformer, the rectifier and the capacitor all have unnecessarily hard lives from the bigger current surges. Efficiency falls and heating increases as resistive losses and diode drops are all scaled up by the increased current.

So like goldilocks, you want a capacitor which is not too small and also not too large. Both cause problems, though different problems.

Finding where the upper and lower limits are depends on things you'd think of as mere details or stray effects. They are actually important. There are some web-based calculators which have been mentioned on here, as being good. I can't steer you to them as I just use LTspice. It's a full-blooded circuit simulator and some people are scared of it, seeing it as excessive for the job. The advantage of using a sledge hammer to crack a walnut is that you not only have a cracked walnut, you also get to keep the sledge hammer and you've started to learn how to use it on things where it really makes a difference.

David

[Radio Scotland]

Originally Posted by Radio Wrangler

Simple answer:

It depends.

If you have a rather small smoothing capacitor, your circuit will discharge it further between peaks of the mains voltage. So you get more ripple. On an unregulated supply, this translates to more hum.

A regulated supply will take off the ripple UNLESS the smoother voltage falls so low that the regulator can't handle it. This can be very bad.

If you opt for a massive smoothing capacitor, it won't discharge much between peaks of the mains. This means that the mains waveform will be almost at the top before the rectifier on the positive-going side turns on and starts to recharge the smoother. This means that the smoother is recharged in a very short amount of time. The same amount of charge has to go in, in order to drive the load for the period between peaks, and this in turn means that the charging current is dramatically increased.

So with too big a C, the transformer, the rectifier and the capacitor all have unnecessarily hard lives from the bigger current surges. Efficiency falls and heating increases as resistive losses and diode drops are all scaled up by the increased current.

So like goldilocks, you want a capacitor which is not too small and also not too large. Both cause problems, though different problems.

Finding where the upper and lower limits are depends on things you'd think of as mere details or stray effects. They are actually important. There are some web-based calculators which have been mentioned on here, as being good. I can't steer you to them as I just use LTspice. It's a full-blooded circuit simulator and some people are scared of it, seeing it as excessive for the job. The advantage of using a sledge hammer to crack a walnut is that you not only have a cracked walnut, you also get to keep the sledge hammer and you've started to learn how to use it on things where it really makes a difference.

David

Very indepth reply, Thank's for that, Seems like its not going to be as simple a project as i originally thought, Think i'll start with 4700uf and see how that goes, As long as i remember to get the polarity right there shouldn't be any massive bangs or explosions,


Cheers

[joebog1]

I think you have it first guess Scotland. My old rule of thumb is 1000 uF per amp. So 4700 is perfect.
Fits most of Davids arguments too. Dont forget a plastic .1 uF across it too!!.

Joe

[Radio Scotland]

Originally Posted by joebog1

. Dont forget a plastic .1 uF across it too!!.

Joe

In parallel with the smoothing capacitor ?

[joebog1]

Yes!! Exactly. It helps a bit with the diode switching noise.

Joe

[Radio Scotland]

Originally Posted by joebog1

Yes!! Exactly. It helps a bit with the diode switching noise.

Joe

Never knew that,


Cheers
Jay

[Robert Gribnau]

I always thought that placing capacitors of about 0.01 uF parallel to the diodes was an effective way to minimise switching noise. But my knowledge doesn't go deep enough to explain the pros and cons.

[Radio Wrangler]

4700uF per amp. With a full-wave rectifier and 50Hz mains, it's 10ms between charging pulses so roughly 10ms discharging.

Q = C*V = I * T

So V = (I * T)/C 4700uF, 10ms and 1 amp gives 2.13v peak-to peak ripple. maybe 10% less if charging is a bit slow.

If your circuit can handle this, then Joe's rule of thumb will work for you.

Electrolytic capacitors are made from long foils and papers in a tight swiss roll. So they are inductive, which spoils their high frequency behavoiur.

Adding a high frequency capacitor across them looks like it will create a resonant circuit. It does, but the ESR of the electrolytic damps the resonance. So, it usually works.

Start with quiet (relatively) diodes... fast ones counter-intuitively

David

[Craig Sawyers]

If you want to eliminate diode reverse recovery artefacts, you need a snubber across the transformer secondary.

The so called noise is a result of the diode recovery transients kicking off the resonance between the mains transformer leakage inductance and winding capacitance. So every time the bridge diodes switch off, you get a damped burst of ringing usually in few hundred kHz range.

I've posted about this recently on another thread, with links about how to kill that dead, the detailed values depending on the transformer.

But putting a small C across an electrolytic does nothing at all, and may actually make things worse. The lead inductances, and the relative capacitance values can give rise to a large resonant peak, quite the opposite of what you want.

Craig

[KeithsTV]

Originally Posted by Craig Sawyers

If you want to eliminate diode reverse recovery artefacts, you need a snubber across the transformer secondary.

I can confirm that as many years ago I was working on an amplifier with an AM radio on nearby. When the amp was powered up there was 50Hz buzzing on the radio. with just the amp PSU running with no load, the interference disappeared. I put a capacitor across the transformer secondary and I could run the amp with no interference to the radio.

Keith

[Craig Sawyers]

a 10-47nF capacitor across the Tx secondary can often sort it out. The full solution puts an RC across that capacitor (a so-called CRC snubber).

But if just a capacitor sorts it out, that is nice and simple.

BTW as a lad of fifteen or so, I built my first preamp. I decided to do what I'd seen done - a capacitor across the transformer primary. 240V, right? So I put a paper capacitor of what I thought was a generous voltage rating of maybe 300V in there, and was greeted some minutes later with an explosion and capacitor guts all over the place. That was when I discovered the difference between RMS and peak!

Craig

[lesmw0sec]

Originally Posted by Craig Sawyers

If you want to eliminate diode reverse recovery artefacts, you need a snubber across the transformer secondary.

The so called noise is a result of the diode recovery transients kicking off the resonance between the mains transformer leakage inductance and winding capacitance. So every time the bridge diodes switch off, you get a damped burst of ringing usually in few hundred kHz range.

I've posted about this recently on another thread, with links about how to kill that dead, the detailed values depending on the transformer.

But putting a small C across an electrolytic does nothing at all, and may actually make things worse. The lead inductances, and the relative capacitance values can give rise to a large resonant peak, quite the opposite of what you want.

Craig

Yes - but don't confuse that with the requirement for local RF bypasses on the supply lines close as possible to any I/C's on the PCB - for other reasons.

[Craig Sawyers]

Goes without saying - yes local bypassing. One way of isolating capacitor interaction is to use a small RF choke, then the capacitor on RF (or any other) IC.

That idea goes way back. The Racal RA17 used an LC as a filter before each chassis module, sometimes with an R as well to damp LC Q.

Craig

[Radio Scotland]

Thanks for the replies / info, I'll post back once i have the circuit up & running, As i said above i'll start out with a 4700 capacitor and see how that fairs, A bit of experimenting maybe needed,


Cheers

[Radio Scotland]

4700uf seems to be fine, After having some weird problems with my homebrew PSU, See thread below, All seems good..
https://www.vintage-radio.net/forum/...64#post1503864

Powers my car radio with no signs of hum