I am making a rectifier for a single N-Scale train building. It takes DC, but the train controller gives out 19VAC. I have 4 1N4001s and some capacitors, but probably not the right size/uf. I only need to make the 19VAC into 19VDC for this one model. I do not know its current draw however.

  • \$\begingroup\$ This is the building: patstrains.com/View_Details/?ID=11220 \$\endgroup\$ Feb 10 '19 at 1:46
  • 1
    \$\begingroup\$ An appropriate capacitor can't be chosen without some idea of the required current. Do you not have a multimeter to measure the current? In any case you aren't getting 19VDC with this setup, you'll need more components. \$\endgroup\$
    – user133493
    Feb 10 '19 at 2:10
  • \$\begingroup\$ What is being powered? If it's just incandescent bulbs, it doesn't need to be DC at all. If it's LEDs with current-limit resistors, then you can power it with un-filtered DC. If it's more complicated, then the situation is more complicated. \$\endgroup\$
    – TimWescott
    Feb 10 '19 at 2:11
  • \$\begingroup\$ I do not have a power source to test with, and it is a LED but I do not know if any resistors are with it. I could make a source with 2x 9V batteries if needed. \$\endgroup\$ Feb 10 '19 at 2:12

using Q = C *V, take the derivative with respect to Time, and get

dQ/dT = C * dV/dT + dC/dT * V

Now define dV/dT to be zero (the capacitance is constant), and we have the useful ripple prediction formula

dQ/dT = C *dV/dT

Now define dQ/dT to be the current (I) and we have

I= C * dV/dT

So what? Let I = 1 ampere, dV = 1 volt, dT = 1/(2*50Hertz) = 0.01 second

What must the smoothing capacitor be?

1 amp = C_farads * 1 / 0.01

and we have

1= C* 100,


C= 1/100 or 10,000uF


Example with ONE UF and 1 AMPERE load:


simulate this circuit – Schematic created using CircuitLab

  • \$\begingroup\$ That is a very conservative answer. High performance audio amps use about 2,000uF per amp of current consumed, on both + and - supply rails. A 1,500 watt RMS mono-block has +/- 120 volt rails with 15 amp fuses. Current peaks are about 10 amps, which calls for a 100,000uF cap. IRL they use 20,000uF caps. +1 \$\endgroup\$
    – user105652
    Feb 10 '19 at 3:24
  • \$\begingroup\$ @ Sparky256 They may also use cascaded output-emitter-follower_plus_cascade, so the Early Voltage of the emitter-follower does not produce120Hz hum in the speaker. And they make be depending on the low-frequency gain of the servo-loop to push the hum way down. To avoid the "singing" of rectifier surges --- each looking like a wide-band impulse ---- they can install 1uH inductors in series with each rectifier diode. Assuming 100 amps peak thru the diodes, turning on in 10US, we have V = L * dI/dT, or V = 1uH * 100 amps/10uS = TEN volts across the 1Uh. And that 10 volts will SLOW down surges. \$\endgroup\$ Feb 10 '19 at 3:36
  • \$\begingroup\$ @analogsystemsrf - Can you explain dT = 1/(2*50Hertz) = 0.01 second ? \$\endgroup\$ Feb 13 '19 at 21:45
  • \$\begingroup\$ @analogsystemsrf - Please correct me: .500mA = C * 19VAC/1/(2*60Hz) .500 = C * (19/0.008333) .500 = C * 2280 and this is where I am bad at math... \$\endgroup\$ Feb 13 '19 at 21:52
  • \$\begingroup\$ At 60Hz recharge rate, a 1Amp current will cause 1 volt sag if the capacitor is 8300 uF. Does that match your math? If your transformer produces 19VAC RMS, or about 30 volts peak, (60 volts Peak_Peak, ), and you take that peak of 30 volts, and allow only ONE volt of ripple, what do you get? \$\endgroup\$ Feb 14 '19 at 17:27

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.