开云体育

PSUD gives me this result on a particular power supply (yes, the choke is underspecified on purpose).? Playing with rectifier leak mitigates the problem somewhat, but it still looks unnatural.? Is this a math problem or is my rectifier in actual danger of blowing up?

?

?

There could be a danger of damage to the rectifier; the problem is the choke value being too low.

It's essential on a choke supply for the current to be continuous through the choke without dropping to zero at any point during steady state operation. If it's doing this, you need a bigger value, and it's a good idea to monitor I(L1) to see what is happening and yield some great clues as to what will work and what won't work.

Had a look at your circuit and the current collapses to zero during the cycle due to the low choke value, this will cause the voltage to head towards infinity until something breaks down, either insulation in the transformer or more likely the rectifier breaking over with the excess voltage. This could cause excess heating in the rectifier and subsequent failure. You can increase rectifier leakage for a visibly better result, however this is just a band aid for the underlying problem.

There are various formulae around for figuring out the correct minimum choke value, but I just did some hit and miss values and got 30mH as about as low as I'd want to go. If you try this value, or higher, it will give much better results.

Have a look at I(L1) without changing anything, then look again after upping the value to 30mH and you'll see the difference it makes. Hope this helps.

Regards,
Duncan

On 4/20/2022 7:18 AM, Duncan Munro wrote:

There are various formulae around for figuring out the correct minimum choke value

Try "Figure out the load resistance, E/I (Volts, Amps), then divide that by 1200. The result is in Henries."

In 50Hz lands use "1000".

So for 24VDC and 0.6A we have 40r, divide by 1200 is 0.034H.

Or for 50Hz, 40r, divide by 1000 is 0.040H. (40mH)

You have 0.007H. Less than 1/5th what the formula suggests for minimum.

Yes, it may kick violently.

Other tip: whenever possible! don't use "current sources". No practical component acts as a perfect current source. I suspect if you replace your I1 with a 40 Ohm you will still get kicks but not divide-by-zero spikes of a math model.

I would also simplify first. Your R1 and C2 probably have little effect on the kicking and rectifier stress.

However none of this changes the fact that 6.8mH is much too small for a 24V 0.6A load. The simplified supply shows (as Duncan says) choke current going to zero every half-cycle, which spoils the advantage of a choke-input filter, while adding the fun of "infinite" voltage kicks.

As the current doesn't instantaneously drop to zero in real life, the voltage generated will be limited.? HOWEVER, your power line is not a perfect sine wave forever... there are inductive devices being switched elsewhere, lightning induces impulses, even without a direct hit, and of course your supply gets turned on and off. As a result, ANY solid-state rectifier used with a choke-input filter should have overvoltage protection - an MOV (between DC max and diode PIV rating) or a small value input cap will do.? In an industrial application, I saw a 1400V, 240A three-phase bridge (used on 208VAC) that would fail every few hundred on/off cycles. A .047 uF across output fixed it for good.

Tom Bavis

Scratch the reverse-biased diode idea: adding a tiny capacitor (10 nF) before the choke and looking at the voltage shows the choke input goes below zero by about a diode drop at the zero-crossing of line power.? Not for long, 100 or 200 microseconds each cycle, but enough to make a mess.

Thanks to both Duncan and Paul, that's exactly what the sim indicates.? Well, that sure showed me for trying to take a shortcut.? I now have a CRC network which works like a charm.

Tom's MOV suggestion is a good one as well - powering the system off would present an open circuit (modulo transformer impedance) to the choke so it would return a flyback voltage.? Would a reverse-biased diode, such as used across relay coils, also work when placed between the choke input and ground?? From stored energy = LI^2/2 I get about 60 millijoules for a 60 mH choke filtering 1.4 amps, so as long as one doesn't get too energetic with the power switch a 1N400x should be capable of dealing with this, yes?