PSUD is not only good for hollow-state, it's also nice for sand-state. I've run a bunch of sims and soldering up hardware shows voltages & ripples are pretty well on the money. Sweet!
If anyone runs across this sort of output, here from a lightly loaded choke input filter with a 400 ohm load resistor emulating about 50-60 mA from the supply, Duncan's already on the case.
From the Help menu:
And the fix, from Help->Technical Information->Program Limitations:
For this circuit, setting 10k in Options->Rectifier Leak sorted the funky output nicely. Firebottle choke input filters could be happy with higher values.
If anyone runs across this sort of output, here from a lightly loaded choke input filter with a 400 ohm load resistor emulating about 50-60 mA from the supply, Duncan's already on the case.
From the Help menu:
And the fix, from Help->Technical Information->Program Limitations:
For this circuit, setting 10k in Options->Rectifier Leak sorted the funky output nicely. Firebottle choke input filters could be happy with higher values.
Last edited:
It will. Since the current will be steady at 2A, have you considered an inductor before the first capacitor to make the filter LCRC? The right inductor will knock down ripple voltage to about a tenth of a CRC section's, and thus power by about 20 dB (power is proportional to voltage squared), with what's left having almost no content above twice line frequency. You only have to make sure the initial power-on doesn't have excessive voltage overshoot; less than 10% is usually acceptable. Using an inductor will also reduce the supply output voltage since LC sections tend to average the input rather than ride the peaks, so selecting a higher voltage for the transformer secondary will solve that problem.
Happy simulating!
Last edited:
PSUD has a table which shows the maximum, minimum, and difference for each parameter, quite handy if all you want is a number. trobbins' tip will get you to the steady state of your filter after it's settled down from power-up.
Compare the curves for I(T1) and I(I1): I(I1) will be flat, indicating smooth DC current flow, whereas I(T1) will be a train of narrow spikes. RMS stands for "Root-Mean-Square", which means you take the square of the value over a time interval, divide by the time, and then take the square root. For DC values this is equivalent to the mean, but the mean for AC should be zero, which is what you see for I(T1). However, taking the RMS value of a sinusoid yields a number giving the equivalent energy of a DC value.
This assumption falls down for signals with a high peak-to-average ratio, where the square term blows up and predominates over the time interval. For example, say the signal is 1000 volts for 1 millisecond, recurring at one second intervals while being zero elsewhere. The average would be one volt, but the RMS would be sqrt(1000*1000 * 1 millisecond / 1 second), or 33 volts!
So why use RMS at all if it's subject to this madness? Power determines heating in circuit elements and power goes as resistance times the square of current, so RMS is useful for computing how much stress a component will undergo, such as the transformer equivalent resistance at about 78 ohms in your circuit.
Since you have a capacitor-input filter, the current going through T1 would be highly spiky (that's a technical term, be careful where you use it
) so it would have the problem mentioned above. Choke-input filters are better behaved vis-a-vis peak-to-average power ratio so the RMS value would be much closer to the DC average at I1 in your circuit.
I'm a big fan of choke-input filters since the above shows they're much easier on transformer and rectifiers and the subsequent capacitor sees a happy little sinusoid atop DC instead of a train of narty looking spikes. However, choke-input filters really want a steady load: they're not the best choice for something like a class-AB or class-D amplifier, with low idle currents and high maximum currents, since the output voltage would rise to that of a capacitor-input filter at idle and collapse to a choke-input filter at high power demand. Class-A amps, on the other hand, are at their best with choke inputs.
This assumption falls down for signals with a high peak-to-average ratio, where the square term blows up and predominates over the time interval. For example, say the signal is 1000 volts for 1 millisecond, recurring at one second intervals while being zero elsewhere. The average would be one volt, but the RMS would be sqrt(1000*1000 * 1 millisecond / 1 second), or 33 volts!
So why use RMS at all if it's subject to this madness? Power determines heating in circuit elements and power goes as resistance times the square of current, so RMS is useful for computing how much stress a component will undergo, such as the transformer equivalent resistance at about 78 ohms in your circuit.
Since you have a capacitor-input filter, the current going through T1 would be highly spiky (that's a technical term, be careful where you use it
) so it would have the problem mentioned above. Choke-input filters are better behaved vis-a-vis peak-to-average power ratio so the RMS value would be much closer to the DC average at I1 in your circuit.I'm a big fan of choke-input filters since the above shows they're much easier on transformer and rectifiers and the subsequent capacitor sees a happy little sinusoid atop DC instead of a train of narty looking spikes. However, choke-input filters really want a steady load: they're not the best choice for something like a class-AB or class-D amplifier, with low idle currents and high maximum currents, since the output voltage would rise to that of a capacitor-input filter at idle and collapse to a choke-input filter at high power demand. Class-A amps, on the other hand, are at their best with choke inputs.
You're welcome I never would have known about PSUD if it weren't for someone here telling me about it years and years ago so, y'know, carry the information forward.
As a point of interest, PSUD also works well simulating low-voltage solid state supplies. One preamp design (LCRC) showed 20 volt rails with 200 uV ripple before the regulators, and by golly throwing a scope (AC coupled of course) on +-Vcc showed that within the limits of error.
And, yes, full credit to Duncan Munro for a really useful program. It does one thing, one thing only, incredibly well.
I never would have known about PSUD if it weren't for someone here telling me about it years and years ago so, y'know, carry the information forward.As a point of interest, PSUD also works well simulating low-voltage solid state supplies. One preamp design (LCRC) showed 20 volt rails with 200 uV ripple before the regulators, and by golly throwing a scope (AC coupled of course) on +-Vcc showed that within the limits of error.
And, yes, full credit to Duncan Munro for a really useful program. It does one thing, one thing only, incredibly well.
The software is simple in concept. 2 capacitors in parallel is the same as 1 capacitor. If you think it is still worthwhile trying to simulate 2 separate capacitors, then use an R-C section for the 2nd capacitor and make R a very low value (ie. the resistance of the wire connecting the 2 caps).
After I took the time to make sure all values entered matched the parts used my results are spot on. Just as with any simulation. This includes transformer secondary DCR and also the actual secondary voltage it gives under load. Capacitor ESR also makes a (small) difference and so on...