1
\$\begingroup\$

I am trying to make a 24VAC to 5VDC converter. My constraint is to keep the height of the entire power module to as small as possible (ideally < 5mm) I have the buck converter design with the height of all components <= 5mm. However for the input capacitor I can't find a capacitor with a small enough height.

My requirements for the power module are:

  • Vin = 24VAC
  • Vout = 5VDC
  • Load current = 1.5 A

Would it be worth considering to put a diode/diodes where the "forward voltage drop" is high enough so it reduces the voltage requirements on the choice of the capacitor.

\$\endgroup\$
  • 5
    \$\begingroup\$ Why not use multiple smaller capacitors? \$\endgroup\$ – Ignacio Vazquez-Abrams Jun 28 '14 at 22:46
  • \$\begingroup\$ I am planning on using LM2596S-5.0 for the SMPS.It can easily accomodate 4.5V to 40V. So I think it should be able to handle decent amounts of p-p. I should have been clearer in my question. The load current that I expect is the load current on the output of the switcher. I can even live with roughly 1A. Based on the datasheet of my SMPS it looks like the following cap should be sufficient: Voltage = 50V. Ripple Cur. rating = 600mA. Capacitance = 100uF. I was able to find a cap(UCL1V101MCL6GS) with a height of 6.6MM,capacitance=100uF,Volt=35V.Perhaps I can put 2 in series. Would this cap work? \$\endgroup\$ – user3496475 Jul 3 '14 at 4:00
  • \$\begingroup\$ You'll need 4 of them in a 2S2P configuration, since putting them in series will halve their capacitance. \$\endgroup\$ – Ignacio Vazquez-Abrams Jul 3 '14 at 4:02
3
\$\begingroup\$

Given that the input filter capacitor(s) will probably determine the useful life of this gadget, consider how much you want to compromise for small size.

You really have two things to play with (I'm going to assume you've got 50Hz input, the answers are not much different for 60Hz)- you can have more ripple voltage (which makes the SMPS work harder and tends to heat the capacitor more or you can have more capacitance (which makes the capacitor bigger).

So let's just say you can live with 5V of ripple peak-to-peak. Your current at 31V or so will be around 300mA assuming 85% efficiency. So with a full-wave rectifier, C ~= 600uF. Let's try around 470uF to 1000uF. Ceramic assemblies meet your height requirement, almost, at 5.5mm, but you'd need 5 pieces at $36 each to get to 500uF. Kinda expensive. 22uF 0805 you'd need maybe 25 of them at 0.65 each (about $16) but a lot of area (maybe you could fill the bottom of the board with them). Very thin.

At 500uF, ripple will be about 11V p-p, still possibly tolerable with ceramic caps.

Polymer SMT caps are more like 11mm high, so forget those.

As to regular electrolytics, you could cut away the board to allow the capacitor to protrude through- gaining maybe 2-3mm that way, so a 7-8 mm diameter cap might be as acceptable as a 5mm tall SMT part. Perhaps one or two of these parts might do you EEU-FR1V471LB. 0.20 each in 1K quantity, so cheap, and 8mm diameter, and they're 105°C 9000h so pretty reliable.

\$\endgroup\$
  • \$\begingroup\$ 5V ripple would be hilarious on a 24V to 5V converter. :) \$\endgroup\$ – dext0rb Jun 29 '14 at 0:53
  • 3
    \$\begingroup\$ The ripple is in the input filter, so even 10Vp-p ripple is not insane, provided the SMPS can handle the range (30V to 20V input range, plus whatever RMS input range has to be accommodated). \$\endgroup\$ – Spehro Pefhany Jun 29 '14 at 1:11
  • \$\begingroup\$ I am planning on using LM2596S-5.0 for the SMPS.It can easily accomodate 4.5V to 40V. So I think it should be able to handle decent amounts of p-p. I should have been clearer in my question. The load current that I expect is the load current on the output of the switcher. I can even live with roughly 1A. Based on the datasheet of my SMPS it looks like the following cap should be sufficient: Voltage = 50V. Ripple Cur. rating = 600mA. Capacitance = 100uF. I was able to find a cap(UCL1V101MCL6GS) with a height of 6.6MM,capacitance=100uF,Volt=35V.Perhaps I can put 2 in series. Would this cap work? \$\endgroup\$ – user3496475 Jul 3 '14 at 3:59
  • \$\begingroup\$ I think the RMS ripple current will be way too high..and the capacitance way too low. It appears that it's rating is lower at mains frequency, whatever that is for you. Remember the current goes up as the input voltage drops- a constant current approximation isn't good enough when the voltage changes a lot. The radial parts I pointed to in my initial answer sitting in a horizontal cutout are probably the best you can do without spending a fortune (probably equivalent to a 5-6mm SMT part). \$\endgroup\$ – Spehro Pefhany Jul 3 '14 at 10:42
2
\$\begingroup\$

One option is to mount a leaded capacitor sideways on the edge of the PCB (or even in a cutout big enough to accept it. If you are allowed 5mm above the PCB, plus the PCB thickness (often 1.6mm), and maybe 1.5mm below it, that would allow an 8mm diameter electrolytic, as long as you have room for, giving you more options for value, voltage etc.

\$\endgroup\$
  • \$\begingroup\$ I am planning on using LM2596S-5.0 for the SMPS.It can easily accomodate 4.5V to 40V. So I think it should be able to handle decent amounts of p-p. I should have been clearer in my question. The load current that I expect is the load current on the output of the switcher. I can even live with roughly 1A. Based on the datasheet of my SMPS it looks like the following cap should be sufficient: Voltage = 50V. Ripple Cur. rating = 600mA. Capacitance = 100uF. I was able to find a cap(UCL1V101MCL6GS) with a height of 6.6MM,capacitance=100uF,Volt=35V.Perhaps I can put 2 in series. Would this cap work? \$\endgroup\$ – user3496475 Jul 3 '14 at 4:00

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.