Power supply "home made"

[skips]

Hi.

First of all sorry for my english 

I am constructing a low voltage power supply(-12V to 12V),  this power supply is only for "testing" and to learn something.

What i've done until now was the schematics and the simulations in spice.

i put in attachements the schmatics and the simulation results.

Now the questions 

What do you think about it?
For the transformer i was thinking in  pcb transformer, something like this:

The next step was export the schematics to eagle to create the pcb.

[Zero999]

The virtual earth has a high impedance - 5k so it won't work under load.

Try rerunning the simulation with a load connected from +V to 0V.

The simulation of the transformer is also way off. The transformer's RMS voltage is 24V but its peak voltage is 24*sqrt(2) = 34V. in real life it will have an internal resistance and the output voltage will be much higher when there's no load.

You need to use a centre tapped transformer.

How much current do you intend to pull from this power supply?

[skips]

About the current i was hoping get the maximum of 1.5A(according to the load after the power supply), this is the maximum of the LMxxx output, but a few mA would be fine.
The objective is to supply a breadboard with some ic's and resistors.

I don't understand what you say about the ground impendace, what are the practical consequences? The ground shouldn't offer non conductor surface and  a fixed potential at 0v?

I was considering that the transformer can supply 24v of amplitude, 48v peak to peak.
Now I added the transformer to simulation, but without considering the imperfections of the power supply, that results on a very high current in primary side. By the way what are the normal values for series resistors with the power supply?

Thanks for the help 

[Zero999]

About the current i was hoping get the maximum of 1.5A(according to the load after the power supply), this is the maximum of the LMxxx output, but a few mA would be fine.
The objective is to supply a breadboard with some ic's and resistors.

You'll need larger capacitors for C1 and C2, 6800uF minimum, and a higher voltage transformer: 30V minimum which needs to be centre tapped i.e 15V-0-15V.

I don't understand what you say about the ground impendace, what are the practical consequences? The ground shouldn't offer non conductor surface and  a fixed potential at 0v?

R1 and R2 form a potential divider which you're using as a virtual earth. The total impedance is 5k so your circuit effectively has a 5k resistor connected in series with the 0V rail.

I was considering that the transformer can supply 24v of amplitude, 48v peak to peak.

No.

The a transformer outputs 24VRMS which is 33.9V peak or 67.8V peak to peak.

Now I added the transformer to simulation, but without considering the imperfections of the power supply, that results on a very high current in primary side. By the way what are the normal values for series resistors with the power supply?

You need to simulate the power supply under load conditions, otherwise there's no point.

So you think your circuit should be able to supply 1.5A? Connect an 8 Ohm resistor from the 12V output to 0V and watch the voltage on the 12V rail fall to nearly nothing.

You need to use a centre tapped transformer with the tap connected to 0V. The circuit won't work without a transformer with a centre tap.

[Bored@Work]

Use a delon circuit with freaky big caps if you don't want to use a centre tapped transformer. Approx. 100 mF x Ohm / Rload(min) per cap.

[Simon]

I'd use either a centre tap transformer or use two half wave rectifiers, there is no need for the virtual earth and like Hero said it will never work and to make it slightly work you will need resistors so small they will waste power, this is a power supply not a voltage reference circuit ! you can reduce your part count and have a much better design or just use a centre tap transformer.

[Zero999]

Use a delon circuit with freaky big caps if you don't want to use a centre tapped transformer. Approx. 100 mF x Ohm / Rload(min) per cap.

That will work but it will cause the transformer's core to saturate causing the windings to burn out, if the current flowing through the the 0V node is too high. Drawing a large current from a half wave rectifier connected to a transformer creates a net DC flux in the core which can cause it to saturate.

Using a lower voltage transformer (9V is probably optimal) and two Greinacher circuits (one for +V and another for -V) is better but it will need massive capacitors for a decent output current.

[skips]

I followed your advices and a changed the 4 diodes rectifier for a centered tapped transformer with 2 diodes.

I'm doing some tests and if the load impedance stays near 3.3Ohms in 5V line (the lowest value that maintains the current bellow the LMXX current output limit) the circuit
output is a oscillation between 4.64 and 4.74V, but when the load is High the circuit outputs a constant 5V tension.

I noticed that in primary side of the transformer  the input wave have a strange deformation in the ascending flank. I think this is because of the bad chooses for the inductance's values.

[jahonen]

I noticed that in primary side of the transformer  the input wave have a strange deformation in the ascending flank. I think this is because of the bad chooses for the inductance's values.

I'd make it still as a full-wave rectifier, it does not make sense to save two diodes here. That will also reduce the ripple somewhat.

The deformation is due to 10 ohm source resistance in your voltage source. Reduce it to something like 10 milliohms and it probably disappears.

Regards,
Janne

[skips]

Thanks, now it work fine, when the load is low it continues with a very small oscillation   but the LMXX can correct this and delivery approximately 1A.

This ripple when the load is very low can be due to unadapted  impedance that cause line reflections?! The spice can simulate this!?

[alm]

This ripple when the load is very low can be due to unadapted  impedance that cause line reflections?! The spice can simulate this!?

Impedance mismatches? Do you have an 1GHz mains frequency? A rule of thumb is that transmission line theory should be used if the length exceeds 1/4 times the wave length. The wave length for 120Hz is something like 3Mm (6 megameter), it's quite unlikely that you'll reach this in any power supply that fits on this earth.

[Simon]

I followed your advices and a changed the 4 diodes rectifier for a centered tapped transformer with 2 diodes.

No only use two diodes if your using a single output transformer to get positive and negative voltages. If your using a centre tap then use a full bridge or your wasting 1/2 the power

[skips]

I forgot to update my post, but after jahonen comment i put a full bridge rectifier.

Next step is to add Adjustable LM317 rectifier and get a regulated output.

[Zero999]

Good, you're on the right track.

The load is non-linear so the AC waveform will be distorted slightly.

You've set the mains voltage incorrectly.You need to set Vin to the peak voltage of the mains in your area, not the RMS. Refer to my first reply to the thread and Google, if you don't know what RMS and peak voltage are or how to calculate them.

What's the total current you're planning to draw, including all outputs?

If it's higher than 1.5A or so, you will need to use some diodes with a higher current rating when you come to build it. The 1N4007 is only rated to 1A but it should be fine up to 1.5A as the diodes only conduct for half a cycle but you shouldn't push beyond 1.5 times the continuous rating.

If you want to limit the current from both the 12V and 5V outputs, you could connect the 5V regulators to the outputs of the 15V regulators. This will reduce the power dissipation in the 5V regulators and improve the ripple. On the down side, the maximum current from the 12V and 5V combined will be 1A.

[skips]

I've corrected the voltage to 325V (220V*sqrt(2)).

About the current, i was thinking to use a 24V-0-24V transformer with a max output current of 1A, this will protect the diodes and leave the current "management" for the voltage regulators. In this case my current limitations would be imposed only by the regulators, am i right?

EDIT: Forget the last part, it won't work i will need in worst case 1.5A entering each one of the LMXX so that he can delivery the max current when output load is low.
So i will need a transformer with a higher current output, and  diodes who can handle all that current.


The max current i will need is about 1A, its only for some circuits with a few ic's and a Arduino.

[skips]

Hi again.

I already have a sketch of the pcb, i would like to hear some opinions.

I didn't put the transformer in the pcb, the pcb only have 2 connectors for the 12V of the transformer secondary.


Now i have question, i need to connect the center of the transformer to ground, what ground i connect, the mains ground or the ground plane of the pcb?
Should i connect the ground plane to center of the transformer and leave the mains ground connected to the box chassis for safety?

[Zero999]

The centre tap of the transformer needs to be connected to the 0V ground of the power supply which is the ground plane of your PCB.

The PSU ground doesn't need to be connected to the mains ground and it's better to leave it floating so there's no risk of a short circuit when connecting an oscilloscope to a circuit powered from it. The mains ground should be connected to the box chassis for safety.

The transformer should have a fuse and switch in series with the primary.

[skips]

Thanks

For the transformer i will use a Veroboard, because they are a little expensive and i will need the transformer for other "projects", so, this way i can have a "portable" transformer.

 About the fuse, i was thinking in a ceramic 2A slow-blow.

I am gathering all the components but probably next week i already have this mini power supply working.

[Zero999]

Using veroboard for anything mains is a bad idea because there's not much clearance between the tracks so you can't be sure there's enough insulation between the primary and secondary.

A 2A fuse is better than nothing but I'd recommend 500mA for a small project like this.