Filtering when designing a 4A linear power supply

[srb1954]

You're ignoring the issue I outlined in reply #16 completely. You really need to take it into account.

Fair enough, I ignored it because I don't know the characteristics of my transformer yet but maybe I'm too optimistic of actually being able to get 2V of ripple with 4-5A. What values are relevant for the problem you outlined? Is the series resistance of the secondary enough to model it in SPICE or should I get the inductance as well as the coupling factor? I'm not sure. The inductance is not very hard to measure but I guess it will strongly depend on the saturation of the core which is a mess to model.

I'll measure the secondary resistance of the one I've got and hope the new one will have similar characteristics.

To model a transformer correctly you need to include the reflected primary resistance as well as the secondary resistance: R_tot = R_sec + R_pri/N² where N is the primary/secondary turns ratio.

On larger transformers you also need to include the effects of leakage inductance but on small transformers the winding resistance dominates the transformer regulation and the leakage inductance can generally be ignored.

[jtruc34]

Well, I've just measure between 140 and 150 mOhms on both secondaries. With a bit of luck it will be similar on the new transformer. Now that I think about it, it's really the leakage inductance that's important here, which I could assume is negligeable for a toroidal transformer, the secondary impedance is probably dominated by the DC resistance. Do you agree?

EDIT: Well, I didn't see srb1954's answer, which answers my problem. Thanks!

[Benta]

Which at an optimistic peak current of 15 A results in around 2.1 V peak voltage drop. This is consistent with my own experience. Plus, the primary resistance is not included yet. It'll be around the same.

[jtruc34]

Which at an optimistic peak current of 15 A results in around 2.1 V peak voltage drop. This is consistent with my own experience. Plus, the primary resistance is not included yet. It'll be around the same.

Taking into account the primary resistance of my 12V transformer, I simulated with Schottky diodes (I think there is no real reason not to use them, and it's also easier for me for thermal design in this particular case), I got the following result (image linked below). As you can see, the current spikes doesn't really seems to depend on the value of the capacitor, and increasing it really does seem to reduce ripple.

I don't fully understand this result, maybe I've done something wrong.

[Benta]

I'm not familiar with your simulation program, but isn't I1 a current source?

[jtruc34]

It's LTspice. Yes, it's a current source, but here I would call it a load.

[Benta]

Right.
Just for fun, I threw your circuit together quickly, but had to scal the current to 10 % (I didn't have any 20 A schottkys in my Spice library), so it's operating at 0.4 A.
I use just a resistive load, this makes things simpler.
the schematic is super-simple.
Plot 1 is cap voltage and R1 current. You see the crest factor in the current plot, and that the peak current is in this case three times higher than average. Plus, the DC voltage is disappointing.
Plot2 is the voltage at the AC input of the rectifier bridge. The peak voltage "capping" is typical of linear power supplies.

[jtruc34]

OK, but I still don't really understand the result of my simulation. You said increasing the capacitance would increase the current spike which does not seem to be the case in my simulation. The DC voltages does become better by increasing the capacitance.

[Benta]

Try replacing I1 with a suitable resistor and let's see if we get similar results. That might be interesting.

[jtruc34]

Try replacing I1 with a suitable resistor and let's see if we get similar results. That might be interesting.

I've just tried and the difference is insignificant, but by the way, I do have similar results to you. I have a pretty significant drop in voltage (about 2 to 3V), but the only thing is that you said there is a diminishing return in picking increasing values of capacitance, and that the current spikes would get higher and higher, which it doesn't, or at least not in the values of C I've tested.

[Benta]

I've just tried and the difference is insignificant, but by the way, I do have similar results to you. I have a pretty significant drop in voltage (about 2 to 3V), but the only thing is that you said there is a diminishing return in picking increasing values of capacitance, and that the current spikes would get higher and higher, which it doesn't, or at least not in the values of C I've tested.

Yes, that statement is only true up to a certain point. The transformer internal resistance saves the situation here.

[Kleinstein]

Schottky diodes tend to have relatively high series resistance. This can cause the relatively identical and not that short current peaks.

In addition to the transformer resistance one could consider adding a fuse (or 2). The added resistance is not that bad.

[jtruc34]

Well, thank you for all your support, I think you have given me pretty much everything I need to know. One small thing, still: do you think it's a problem that I exceed the ripple current for the first cycles when charging the capacitor? Except from blowing the secondary fuse (which I can just choose smartly to avoid that problem).

[Kleinstein]

There is usually not problem with a higher paek current for the first few cycles.  The worst case loss for the series resistance on charging a capacitor is equal to the energy stored in the capacitor.  Normal electrolytic capacitors have no problem absorbing that much as heat without getting really hot.

A large size (e.g. more than some 100 VA) toroidal transformer may need some kind of soft start anyway. There can be a quite large current spike from driving the core to saturation due to remanence and random start phase.

[srb1954]

Well, thank you for all your support, I think you have given me pretty much everything I need to know. One small thing, still: do you think it's a problem that I exceed the ripple current for the first cycles when charging the capacitor? Except from blowing the secondary fuse (which I can just choose smartly to avoid that problem).

Electrolytic capacitors are generally OK for absorbing the high initial peak current, which will be limited mainly by the transformer resistance anyway.

However, you need to pay attention to the surge currents through the diodes. The diode will be specified with a maximum surge current I_FSM which should never be exceeded even on a single cycle. The diode datasheet may also have a derating curve for that I_FSM figure that shows the maximum surge current that can be tolerated over durations longer than a single cycle. You should always ensure that the peak diode currents come under the allowable limits of that curve at all times during the startup period.

[jtruc34]

I'm resuscitating this post because someone questionned the usage of POL regulation. From what I understand, the goal is to minimize the effect of stray wire resistance and inductance when the load is far away from the source, but I don't see how POL is achieving more there than simply putting bypass capacitors near the loads.

[Kleinstein]

Bypass capasitors are for the higher frequencies / shorter time scale. The local regulator is maily for the longer time scale / lower frequency part, e.g. regulate after the resistive drop on traces. With relatively low voltage parts (e.g. 1.8 V or 3 V) the tolerance in absolute voltage got smaller.