EEVblog Electronics Community Forum
Electronics => Beginners => Topic started by: Mp3 on August 06, 2021, 05:01:44 pm
-
I just bought 50 1000uf caps off Mouser the other night and realized this morning I selected caps with just under 1A tolerance for ripple current. Argh!
For an analog / non switching power supply (three individually fused AC lines from the transformer, two go into voltage regulators, one has its voltage quadrupled)
Unfortunately i already removed the old caps. Should i new ones in, measure the ripple on a scope, and change the caps again if they exceed the ratings and hope they don't give the filter caps more than 1A of ripple? Or should i go back and order the 1000uf's with a much higher ripple current?
-
for mains frequency use, some 1 A of ripple current rating is normally sufficient for a 1000 µF cap. Usually one has a ripple current about has hiogh as the DC current and 1 A more want some 2-5 mF of capacitance to kepp the ripple voltage at a reasonable level. The number depends on the voltage - lower valtage wants less ripple and this more capacitance.
The ripple current may be higher with the voltage multiplier - so here one may want better caps. Still it depends on the circuit how large the ripple current really is - the specs are usually for RMS, not peak current.
-
If you need a detailed calculation, I recommend the freeware "PSUD 2" from Duncan Labs.
https://www.duncanamps.com/psud2/ (https://www.duncanamps.com/psud2/)
It calculates actual ripple current through the filter capacitors, as well as rectifier current, DC voltages, and ripple voltages.
(Note that the transformer must be characterized by open-circuit secondary voltage and equivalent series resistance. The instructions tell you how to deal with that.)
-
for mains frequency use, some 1 A of ripple current rating is normally sufficient for a 1000 µF cap. Usually one has a ripple current about has hiogh as the DC current and 1 A more want some 2-5 mF of capacitance to kepp the ripple voltage at a reasonable level. The number depends on the voltage - lower valtage wants less ripple and this more capacitance.
The ripple current may be higher with the voltage multiplier - so here one may want better caps. Still it depends on the circuit how large the ripple current really is - the specs are usually for RMS, not peak current.
Aha, that is a handy rule of thumb, I will keep it in mind for future cap selection if I don't want to go calculate or simulate a circuit. ;)
The higher voltage caps used by the voltage multiplier, I bought some Vishay's that are rated something crazy like 125C and 10,000 hours, so I'm sure those will be fine. ;)
I also didn't even think to consider RMS ripple of my circuit so thank you.
If you need a detailed calculation, I recommend the freeware "PSUD 2" from Duncan Labs.
https://www.duncanamps.com/psud2/ (https://www.duncanamps.com/psud2/)
It calculates actual ripple current through the filter capacitors, as well as rectifier current, DC voltages, and ripple voltages.
(Note that the transformer must be characterized by open-circuit secondary voltage and equivalent series resistance. The instructions tell you how to deal with that.)
Never heard of this but this is about to be my new favorite app!! :)
Thank you very much for that link!
-
When using the software, be sure to choose the relevant parameter from the available results, e.g. rms current through the filter caps but peak-to-peak voltage at the output. Also, allow some elapsed time in the computation. The load can be a resistor or a current or a stepped current. In some circuits, the constant current load gives bad results during initial turn-on, but that can be ignored.
-
Alternative to the special calculations /simulations, one can also do a simulation with the normal spice SW (e.g. LTspice or Tina). Just make shure to include the transformer parasitic effects like resistance at leat approximate.
For a PSU / regulator design a simulation is a good idea anyway, not just for the ripple, but also the stability.
-
I believe PSUD2 uses Spice as the calculation engine. I find its user interface useful, and it includes lots of device models, including some vacuum tube rectifiers.
It is limited to transformer-driven single-phase sinusoidal rectifiers, of the common configurations at (selectable) 50 or 60 Hz.